"Codec Wars" : The attempt of an objective AVC/h.264 versus VC-1 benchmark [Text View] - Page 3

RBFilms

06-18-08, 07:19 AM

Are you talking about the SDK and implementation? I am still not clear on what you mean.

To answer your question ... I think ... we used Scenarist to test VC-1 and tried a variety of AVC Encoders as well. VC-1 performed well but AVC performed better in our opinion.

We do realize that not all encoders are created equally. There are differences between them from what we have seen .... albeit there does not seem to be much you can really do with any of them that is all that far "out of the box" so to speak.

However, there are some things that Tiger AVC seems to handle well ... which may be things that have nothing to do with the basic SDK or AVC encoding process. I am not sure about that ... nor do I care .. as I have never claimed to be an engineer.

I am just a very dedicated and passionate Producer of Audiophile, Videophile, and Home Entertainment products. I go with the differences I can see and hear and could care less what the specs or engineering minds have to say. In my opinion, many typical engineers are focused on a lot of things that do not always translate in to real world results. If it looks or sounds better ... that is all I need to make a decision.

However, I was informed by several engineers that Tiger handles pattern based dithering extremely well ... better than any other encoder in fact. Also, there may be a difference in how encoders handle downsampling from a 10 bit color source.

I also know that many folks agree Tiger produces superior results. In the "old days" ... with DVD Encoders ... we clearly saw a big difference between encoders and had one that we favored over all others and it clearly did the best job on most content.

We are seeing a similar difference with AVC encoders as well. I am not sure about VC-1 as the only once we have any experience with is what is included with Scenarist.

Right now, we like Tiger. Even the $300,000 "dumbed down" version of Tiger ... the Tiger Cub .... incorporates some of the most important technology only available with Tiger ... so I have been told.

Rumor also has it that Sony's new AVC Encoder uses some proprietary Tiger technology. Tiger is not for sale but would probably cost a fortune if it was.

We are using the the proprietary Tiger Encoder at Technicolor for all of our future projects at this point. It appears to do a great job.

I found this online for what it is worth. Probably too basic for you engineering types but perhaps helpful for consumers reading the thread.

I am not interested in participating in this thread on any serious level. I just wanted a second "unbiased" opinion about Tiger ... albeit, we have already decided that it is the best encoder available for our application.

-------------

A Quick Introduction to Blu-ray Codecs

Blu-ray Disc supports three video codecs: Mpeg-2, SMPTE VC-1 and h.264/AVC.

Mpeg-2 is the venerable codec that we have worked with in standard DVD for over a decade. There are many reliable Mpeg-2 encoders that can output a compliant HD stream. The downside of Mpeg-2 is that it is not nearly as efficient as VC-1 or AVC, so it requires higher data rates (and larger file sizes) to achieve similar quality results. Most of the initial Blu-ray releases were Mpeg2.

SMPTE VC-1 is a new standard that is based on the Windows Media 9 codec from Microsoft. VC-1 is more efficient than Mpeg-2 and can achieve high quality results in the 15mbs range. There are numerous licensing entities involved in the VC-1 standard which has lead to the greatest number of encoders falsely claiming BD output. Most HD DVD titles were VC-1, however very few Blu-ray titles have used this codec.

AVC is the newest of the three codecs and is the most advanced. It is also the favored codec for Blu-ray Discs. AVC encoded streams can often achieve the same quality of Mpeg-2 at one half the data rate. The relative complexity of the codec compared to VC-1 and Mpeg-2 and the fact that it is new means there are not as many encoding options for AVC.

benwaggoner

06-18-08, 11:21 AM

Are you talking about the SDK and implementation? I am still not clear on what you mean.
The SDK is an implementation.

To answer your question ... I think ... we used Scenarist to test VC-1 and tried a variety of AVC Encoders as well. VC-1 performed well but AVC performed better in our opinion.
Scenarist is an authoring tool. Do you mean CineVision? Original flavor or PSE? What version? You'll get massive differences in VC-1 output between Cinevision and Cinevision PSE. And there are a very wide array of improvements between the original PSE release and the current one being used for many Blu-ray titles today.

As far as I know, your original disc is the only widely released title done with the non-PSE Cinevision VC-1 encoder, so is not representative of any other VC-1 HD DVD or Blu-ray titles.

...I was informed by several engineers that Tiger handles pattern based dithering extremely well ... better than any other encoder in fact. Also, there may be a difference in how encoders handle downsampling from a 10 bit color source.
This is something that the Hollywood compression houses have always felt was something that our tools did extremely well. So what are they comparing Tiger to?

We are seeing a similar difference with AVC encoders as well. I am not sure about VC-1 as the only once we have any experience with is what is included with Scenarist.
Ah. If that's the only one you tried, than you haven't been exposed to the VC-1 that is being used for actual discs.

Right now, we like Tiger. Even the $300,000 "dumbed down" version of Tiger ... the Tiger Cub .... incorporates some of the most important technology only available with Tiger ... so I have been told.

Rumor also has it that Sony's new AVC Encoder uses some proprietary Tiger technology. Tiger is not for sale but would probably cost a fortune if it was.

We are using the the proprietary Tiger Encoder at Technicolor for all of our future projects at this point. It appears to do a great job.

I found this online for what it is worth. Probably too basic for you engineering types but perhaps helpful for consumers reading the thread.

I am not interested in participating in this thread on any serious level. I just wanted a second "unbiased" opinion about Tiger ... albeit, we have already decided that it is the best encoder available for our application.

-------------

A Quick Introduction to Blu-ray Codecs

Blu-ray Disc supports three video codecs: Mpeg-2, SMPTE VC-1 and h.264/AVC.

Mpeg-2 is the venerable codec that we have worked with in standard DVD for over a decade. There are many reliable Mpeg-2 encoders that can output a compliant HD stream. The downside of Mpeg-2 is that it is not nearly as efficient as VC-1 or AVC, so it requires higher data rates (and larger file sizes) to achieve similar quality results. Most of the initial Blu-ray releases were Mpeg2.

SMPTE VC-1 is a new standard that is based on the Windows Media 9 codec from Microsoft. VC-1 is more efficient than Mpeg-2 and can achieve high quality results in the 15mbs range. There are numerous licensing entities involved in the VC-1 standard which has lead to the greatest number of encoders falsely claiming BD output. Most HD DVD titles were VC-1, however very few Blu-ray titles have used this codec.

AVC is the newest of the three codecs and is the most advanced. It is also the favored codec for Blu-ray Discs. AVC encoded streams can often achieve the same quality of Mpeg-2 at one half the data rate. The relative complexity of the codec compared to VC-1 and Mpeg-2 and the fact that it is new means there are not as many encoding options for AVC.
Where did you find THAT? It's not helpful to consumers, it's just factually wrong.

"There are numerous licensing entities involved in the VC-1 standard which has lead to the greatest number of encoders falsely claiming BD output"?
The sole licensing entity for VC-1 is MPEG-LA, the same entity that does MPEG-2 and H.264.

And encoders falsely claiming BD output? I've never even heard a rumor about that.

"...very few Blu-ray titles have used this codec"?
Manifestly false. All three codecs have seen quite a bit of use, and there are certainly many more new titles using VC-1 than MPEG-2.

RBFilms

06-19-08, 08:28 AM

Thanks for the feedback. I found that info online.

http://dostudio.netblender.com/wikipapers/Encoders1.asp

Nature's Journey was an early HD-DVD / BD release and we used the latest and greatest and most current version of VC-1 that was available at the time. Technicolor did the encodes.

Honestly, I understand your need to promote CODEC and Technology as part of your obligation to your employer ... and your commitment to the development team.

I have no such affiliations, obligations, or loyalty to be concerned with. We honestly see diffrences between them ... as I am sure you yourself would admit ... otherwise ... one would not be any different from the other. For anyone to claim VC-1 is better immediatley confirms there is a difference and leaves the discussion open to debate regarding AVC verses VC-1.

I am not here to debate as I am already sold on AVC and so are all of the authoring houses I have spoken with.

VC-1 is good ... and I would use it in some applications depending on content.

I find AVC ... compared to the flavors of VC-1 I have tried .... looks sharper and more detailed, handles blacks and shadow detail better, and appears to take better advantage of the additional bandwidth I tend to give my encodes.

These are only my observations. I am not here to promote or bash one or the other. However, I do have some special edition High Bit Rate releases coming out in 4th quarter and I will be using Tiger.

I woudl be happy to do test encodes using the latest version of VC-1 and do a new comparison to Tiger. If you can help me cooirdinate something like this I will be happy to revisit AVC verses VC-1.

I would even be willing to put the test encodes on any of my upcoming releases as an Easter Egg for AVS Forum fans.

Who can you recommend for a test encode at 36mbps to 38mbps using VC-1 with our content? WHo is the best encoding technician you know? I will be happy to give this a try.

The SDK is an implementation.

Scenarist is an authoring tool. Do you mean CineVision? Original flavor or PSE? What version? You'll get massive differences in VC-1 output between Cinevision and Cinevision PSE. And there are a very wide array of improvements between the original PSE release and the current one being used for many Blu-ray titles today.

As far as I know, your original disc is the only widely released title done with the non-PSE Cinevision VC-1 encoder, so is not representative of any other VC-1 HD DVD or Blu-ray titles.

This is something that the Hollywood compression houses have always felt was something that our tools did extremely well. So what are they comparing Tiger to?

Ah. If that's the only one you tried, than you haven't been exposed to the VC-1 that is being used for actual discs.

Right now, we like Tiger. Even the $300,000 "dumbed down" version of Tiger ... the Tiger Cub .... incorporates some of the most important technology only available with Tiger ... so I have been told.

Rumor also has it that Sony's new AVC Encoder uses some proprietary Tiger technology. Tiger is not for sale but would probably cost a fortune if it was.

We are using the the proprietary Tiger Encoder at Technicolor for all of our future projects at this point. It appears to do a great job.

Where did you find THAT? It's not helpful to consumers, it's just factually wrong.

"There are numerous licensing entities involved in the VC-1 standard which has lead to the greatest number of encoders falsely claiming BD output"?
The sole licensing entity for VC-1 is MPEG-LA, the same entity that does MPEG-2 and H.264.

And encoders falsely claiming BD output? I've never even heard a rumor about that.

"...very few Blu-ray titles have used this codec"?
Manifestly false. All three codecs have seen quite a bit of use, and there are certainly many more new titles using VC-1 than MPEG-2.

madshi

06-19-08, 09:20 AM

@Richard, would you mind telling us exactly which VC-1 encoder you used for Nature's Journey? Was is Cinevision or was it Cinevision PSE or was it something else? Saying "we used the latest and greatest and most current version of VC-1 that was available at the time" is not very specific at all.

Being the producer of the disc it should surely be possible for you to find out which VC-1 encoder (product name + version) was used for the disc exactly?

Thanks! :)

MovieSwede

06-19-08, 09:37 AM

Who can you recommend for a test encode at 36mbps to 38mbps using VC-1 with our content?

Would be nice to see how the codecs behave in this two scenarios

A BD50 with a 120 minute long movie one lossless audiocodec (24/48) and one DD640.

And the same restriction on a BD25.

That would allow us videopeak and avarage videobitrate of 40mbs on the BD50 and on the BD25 we would have 40 in peak but an avarage bitrate of 20-22 mbs for the video.

So PBR/ABR 40/40* and 40/20 are 2 very interesting limits to test how a codec perform for various BD projects.

* = that its not limit on the avarage bitrate in this scenario.

It also would be fun to see how a BD25 can hold up against BD50 using modern codecs.

scaesare

06-20-08, 09:59 AM

{SNIP}

Nature's Journey was an early HD-DVD / BD release and we used the latest and greatest and most current version of VC-1 that was available at the time. Technicolor did the encodes.

{SNIP}

Unfortunately, an analysis of the disc indicates the encoding paramters were incorrect.

Corellianrogue

06-20-08, 10:59 AM

Unfortunately, an analysis of the disc indicates the encoding paramters were incorrect.

For some reason your post makes me think of Star Wars.

Imperial Officer: We have recovered the Probot Lord Vader.

Darth Vader: Excellent. Display its recovered disc on my viewscreen.

Imperial Officer: I'm sorry Lord Vader but unfortunately an analysis of the disc indicates the encoding parameters were incorrect... *gasp* *gargle* *thud*

Darth Vader: Apology accepted.

:D

RobertR1

06-20-08, 11:46 AM

Wow, someone at Microsoft actually got back to Richard :)

RBFilms

06-21-08, 10:58 AM

Checking to be sure...but Nature's Journey was an early title...as was Chronos ... which used MPEG-2 since neither VC-1 or AVC was mature enough at the time to yield good results.

@Richard, would you mind telling us exactly which VC-1 encoder you used for Nature's Journey? Was is Cinevision or was it Cinevision PSE or was it something else? Saying "we used the latest and greatest and most current version of VC-1 that was available at the time" is not very specific at all.

Being the producer of the disc it should surely be possible for you to find out which VC-1 encoder (product name + version) was used for the disc exactly?

Thanks! :)

RBFilms

06-21-08, 11:01 AM

RBFilms

06-21-08, 11:02 AM

That is pretty funny...you have obiusly been reading the forums for a while...:)

Wow, someone at Microsoft actually got back to Richard :)

sperron

06-21-08, 11:13 AM

Hey Ben, I was wondering if you could comment on what went wrong with the VC-1 encode of Sweeney Todd. Xylon posted a bunch of screenshots here (http://www.avsforum.com/avs-vb/showthread.php?t=1040485) along with the specs.

Here's a screenshot off the Blu-Ray:
http://i166.photobucket.com/albums/u98/adzez/Sweeney%20Todd/6f213c76.png

It's just absolutely littered with compression artifacts.

amirm

06-21-08, 12:48 PM

Checking to be sure...but Nature's Journey was an early title...as was Chronos ... which used MPEG-2 since neither VC-1 or AVC was mature enough at the time to yield good results.
Here we go again :(. This the review of Chronos: http://www.dvdtalk.com/reviews/25835/chronos/

"Regarding Chronos specifically, the picture is sharp and detailed, with robust colors and a good sense of depth. Film grain is present but well compressed. It's quite a lovely disc, with a couple of caveats. The first is that the source elements have minor dirt and specks. They're not severe enough to be bothersome. More distracting are some edge enhancement artifacts, most noticeable when dark objects are sharply contrasted against bright backgrounds (check out all the shots of Stonehenge for one example). The ringing is fortunately low-amplitude, but is persistent in many scenes. Also problematic is some minor instability and jitter in a few places, especially the end credits.

In any case, Chronos is a movie loaded with great imagery, and the HD DVD delivers it nicely by large measure.

[Update: Since first publishing this review I've been in contact with a rep from R&B Films. In our discussion, it appears that the 1080i labeling on this disc's packaging is largely a matter of semantics and interpretation of the HD DVD format specs. It turns out that the Chronos HD DVD is actually encoded on disc as 1080p24 video frames with 2:3 reverse field cadence headers for ease in decoding to 1080i by the player. This is exactly the same thing that all of the major studios do. A player can use the flags to output the video in 1080i format, or can ignore the flags and output the raw 1080p24 data (at the time of this writing no HD DVD players offer that function yet). I feel that labeling the disc "1080i" misleadingly implies that the data is stored in interlaced form and is somehow inferior to a true 1080p master. The R&B rep disagrees. For what it's worth, the Chronos HD DVD is mastered in the same manner as all of the film-based content released by any of the major studios.] "

Did you apply edge enhancement to the image before encoding it? If not, then then the ringing is caused by MPEG-2 compression artifacts.

Here is a review of a VC-1 movie which was encoded probably a year before Chronos was: http://www.dvdtalk.com/reviews/21506/phantom-of-the-opera-the/

"The Phantom of the Opera HD DVD is encoded on disc in High Definition 1080p format using VC-1 compression. .... The film opens with and has occasional interludes in grainy black and white. The grain looks a little noisy, but I suspect that it was added digitally for effect and is not genuine film grain. From there the movie segues directly into glorious color, which the HD DVD delivers with astonishing image quality. The deep, rich colors pop off the screen in every vibrant shade. The picture is razor sharp with fine object detail visibility far surpassing anything possible on standard DVD, yet without a trace of edge enhancement artifacts. The contrast range is perfectly delineated with rich blacks, very good shadow detail, and a fine sense of depth. Since this is a launch title, I hesitate to rate the disc too highly, but at this point in time I can't find anything serious to complain about. "

Hard to imagine the above being said about a technology that is not capable of "yielding good results."

amirm

06-21-08, 01:16 PM

Really...and you know this how? [that wrong encoding parameter was used]
He is talking about a thread you had created on your production practices where Stacey said: http://www.avsforum.com/avs-vb/showpost.php?p=11979429&postcount=245

"I finally got around to looking at the bitstream on Nature's Journey and I am puzzled. It seems to be using the interlace frame encoding mode of PEP instead of the interlaced field mode. There are two interlaced modes, field and frame. Frame is used when there is no motion or very slow motion while field is used in normal motion. You can force an encode to be one, the other or both (auto). Normally we recommend the auto mode. A higher bitrate is required if you use the less effecient mode for the motion in the frame.

The encode is virtually CBR, which can be wasteful in the bitrate department. On one hand, the easy scenes make a look a tad better, but the more complex scenes may suffer.

Truth be told, frame is the red headed step child. Our focus is always progressive mode first, followed by field and when time permits, frame."

You are saying that Technicolor ... who spoke with Microsoft about this tritle and had their input ... did something wrong?
Since interlace encoding was (and is) a rare occuring in these circles, it is easy to see how Tehnicolor's judgement could have been wrong. Best way to have avoided that was to stay in the loop between Microsoft and whoever is encoding your content.

Give me details and I will check it out.

Give me specific details instead of just a genaral comment please.
Did you miss it when Stacey post it the first time in your thread?

scaesare

06-22-08, 11:31 AM

Really...and you know this how? You are saying that Technicolor ... who spoke with Microsoft about this tritle and had their input ... did something wrong?

I am always amused by the folks who claim their was a mistake if they do not care for the results.

Give me details and I will check it out.

Give me specific details instead of just a genaral comment please.

Stacey Spears already did back when this was first brought up. You dismissed it then, suggesting that Technicolor must have done it correctly after having spoken to MS.

However, when pointed out that the code on disc nonetheless appears to have been encoded using some odd settings, you opted to not respond.

scaesare

06-22-08, 11:33 AM

For some reason your post makes me think of Star Wars.

Imperial Officer: We have recovered the Probot Lord Vader.

Darth Vader: Excellent. Display its recovered disc on my viewscreen.

Imperial Officer: I'm sorry Lord Vader but unfortunately an analysis of the disc indicates the encoding parameters were incorrect... *gasp* *gargle* *thud*

Darth Vader: Apology accepted.

:D

"I find your lack of progressive encoding... disturbing."

;)

Joe Bloggs

06-22-08, 12:10 PM

"I find your lack of progressive encoding... disturbing."

;)
But it was recorded at the high rate of 60i. To encode progressively would mean it would have to have been recorded at the rate of 24 slow frames per second. There's no option of encoding Blu-ray content at 60p despite cameras being available capable of 1080p60. I suppose more could be shot at 30p and encoded at 60i and you shouldn't notice much difference (than encoded 1080p30 content), if any, if it was encoded correctly (at 1080i60).

Andrew_HD

06-22-08, 01:52 PM

However, there are some things that Tiger AVC seems to handle well ... which may be things that have nothing to do with the basic SDK or AVC encoding process. I am not sure about that ... nor do I care .. as I have never claimed to be an engineer.

....

However, I was informed by several engineers that Tiger handles pattern based dithering extremely well ... better than any other encoder in fact. Also, there may be a difference in how encoders handle downsampling from a 10 bit color source.

I also know that many folks agree Tiger produces superior results. In the "old days" ... with DVD Encoders ... we clearly saw a big difference between encoders and had one that we favored over all others and it clearly did the best job on most content.

.....

Right now, we like Tiger. Even the $300,000 "dumbed down" version of Tiger ... the Tiger Cub .... incorporates some of the most important technology only available with Tiger ... so I have been told.

Rumor also has it that Sony's new AVC Encoder uses some proprietary Tiger technology. Tiger is not for sale but would probably cost a fortune if it was.

Almost everything above is true. Tiger is a very good encoder, but there is a lots of hardware, so price is very high. It has nice workflow, designed with Technicolor compressionists and it produces great results.

There is only one thing which I wouldn't agree: you can buy it. If you have money than there is no problem. It's on the Thomson site as one of the many products which you can buy. I've seen demo and it's a very professional product.

Tiger is an old name (when it was only for Technicolor use), new name is NexCode. More info:

http://www.thomson.net/GlobalEnglish/Nexcode/Nexcode/Pages/default.aspx

Andrew

benwaggoner

06-22-08, 02:52 PM

Hey Ben, I was wondering if you could comment on what went wrong with the VC-1 encode of Sweeney Todd. Xylon posted a bunch of screenshots here (http://www.avsforum.com/avs-vb/showthread.php?t=1040485) along with the specs....

It's just absolutely littered with compression artifacts.
It's hard to say what's going on there without seeing the source. The bitrate should have been more than sufficient for the content, so I think there's a good chance we've got source issues. The posters in the original thread seem to think it's due to "digital airbrushing" which is certainly plausible on the faces at least.

I've been pretty much out of the loop on the blow-by-blow for particular titles for a while now; this is the first I've heard anything about this disc.

scaesare

06-22-08, 05:40 PM

But it was recorded at the high rate of 60i. To encode progressively would mean it would have to have been recorded at the rate of 24 slow frames per second. There's no option of encoding Blu-ray content at 60p despite cameras being available capable of 1080p60. I suppose more could be shot at 30p and encoded at 60i and you shouldn't notice much difference (than encoded 1080p30 content), if any, if it was encoded correctly (at 1080i60).

So you are saying Stacey's analysis and subsequent comments are incorrect?

Joe Bloggs

06-22-08, 05:52 PM

So you are saying Stacey's analysis and subsequent comments are incorrect?
I was just responding to the "I find your lack of progressive encoding disturbing" comment, I didn't read the analysis bit properly, maybe it was encoded incorrectly or maybe it wasn't I don't know. But from your comment it seemed you were implying that it should have been encoded progressively, which, given that Nature's Journey was recorded using interlaced video cameras, seemed to be an incorrect way to encode it.

Actually - just reading Staceys comment, Stacey seems to be saying that progressive frame mode encoding was used instead of interlaced, whereas interlaced should have been used instead.
It seems to be using the interlace frame encoding mode of PEP instead of the interlaced field mode

So the "your lack of progressive encoding" comment may not be quite right?

benwaggoner

06-22-08, 06:34 PM

Actually - just reading Staceys comment, Stacey seems to be saying that progressive frame mode encoding was used instead of interlaced, whereas interlaced should have been used instead.
There are three different modes:

Progressive
Interlaced Frame
Interlaced Field

Frame<>Progressive in this context.

Joe Bloggs

06-22-08, 06:38 PM

They have to complicate it in these encoders :) (they could have just had two easy options - frame or field based) - but I'm still correct that it would have been wrong to encode Nature's Journey, something recorded on interlaced cameras, as progressive though ;)

scaesare

06-22-08, 09:40 PM

I was just responding to the "I find your lack of progressive encoding disturbing" comment, I didn't read the analysis bit properly, maybe it was encoded incorrectly or maybe it wasn't I don't know. But from your comment it seemed you were implying that it should have been encoded progressively, which, given that Nature's Journey was recorded using interlaced video cameras, seemed to be an incorrect way to encode it.

Actually - just reading Staceys comment, Stacey seems to be saying that progressive frame mode encoding was used instead of interlaced, whereas interlaced should have been used instead.

So the "your lack of progressive encoding" comment may not be quite right?

Ah.. that was simply a throw-away line to Corellianrogue's "Darth Vader" post, which I found quite funny. :D

RBFilms

06-23-08, 12:53 AM

Answer from the lab. I forgot that we used the MS Encoder....it was along time ago....they reminded me.

MS was INVOLVED with the encoding. They consulted with Technicolor on the phone regarding the encoding of Nature's Journey

Here s their comment:

We used the Microsoft encoder.

We did not use the Cinevision.

Best,

Rich

@Richard, would you mind telling us exactly which VC-1 encoder you used for Nature's Journey? Was is Cinevision or was it Cinevision PSE or was it something else? Saying "we used the latest and greatest and most current version of VC-1 that was available at the time" is not very specific at all.

Being the producer of the disc it should surely be possible for you to find out which VC-1 encoder (product name + version) was used for the disc exactly?

Thanks! :)

RBFilms

06-23-08, 01:13 AM

Yes, there was omse edge enhancemnt and MPEG-2 yielded better results than VC-1 at the time. The VC-1 CODEC was not mature enough that he time to compete. That is a fact.

CHRONOS is a 30 plus year old film shot for $500,000 by a bunch of young pioneering filmmakers.

Phantom of the Opera is a visually interesting big budget film shot within the last few years and an early HD release ... meaning there was not a lot to compare to at the time. In the early days, I am sure this looked stunning ... and I am sure it probably still hols up .. but there is more to compare to on the makret at this point.

These films are miles apart and do not make sense for comparison. Once again, you are using selective choices that are not relative or comparable to make your point.

Two reviews do not make an argument ... plus you can't argue VC-1 is great back in the day and then say there have been massive improvements in another thread.

Once again, your arguments do not serve to convince me of anything other that the fact that you have a very focused agenda that is seriously skewed by your investment in VC-1 and HD-DVD.

That si fine...but you saying I am wrong..does not make me wrong. VC-1 has improved over time and was simply not ready for Prime Time when we created CHRONOS. We looked at it ... with our content ... and we know.

Review from IMDB

Beautiful film, 23 November 2004
9/10 -Author: lisa_nyc from New York

" I was able to view this at a special screening and was very impressed. It is a visually stunning movie - the costuming and sets are as extravagant and lavish as the music...."

Here we go again :(. This the review of Chronos: http://www.dvdtalk.com/reviews/25835/chronos/

"Regarding Chronos specifically, the picture is sharp and detailed, with robust colors and a good sense of depth. Film grain is present but well compressed. It's quite a lovely disc, with a couple of caveats. The first is that the source elements have minor dirt and specks. They're not severe enough to be bothersome. More distracting are some edge enhancement artifacts, most noticeable when dark objects are sharply contrasted against bright backgrounds (check out all the shots of Stonehenge for one example). The ringing is fortunately low-amplitude, but is persistent in many scenes. Also problematic is some minor instability and jitter in a few places, especially the end credits.

In any case, Chronos is a movie loaded with great imagery, and the HD DVD delivers it nicely by large measure.

[Update: Since first publishing this review I've been in contact with a rep from R&B Films. In our discussion, it appears that the 1080i labeling on this disc's packaging is largely a matter of semantics and interpretation of the HD DVD format specs. It turns out that the Chronos HD DVD is actually encoded on disc as 1080p24 video frames with 2:3 reverse field cadence headers for ease in decoding to 1080i by the player. This is exactly the same thing that all of the major studios do. A player can use the flags to output the video in 1080i format, or can ignore the flags and output the raw 1080p24 data (at the time of this writing no HD DVD players offer that function yet). I feel that labeling the disc "1080i" misleadingly implies that the data is stored in interlaced form and is somehow inferior to a true 1080p master. The R&B rep disagrees. For what it's worth, the Chronos HD DVD is mastered in the same manner as all of the film-based content released by any of the major studios.] "

Did you apply edge enhancement to the image before encoding it? If not, then then the ringing is caused by MPEG-2 compression artifacts.

Here is a review of a VC-1 movie which was encoded probably a year before Chronos was: http://www.dvdtalk.com/reviews/21506/phantom-of-the-opera-the/

"The Phantom of the Opera HD DVD is encoded on disc in High Definition 1080p format using VC-1 compression. .... The film opens with and has occasional interludes in grainy black and white. The grain looks a little noisy, but I suspect that it was added digitally for effect and is not genuine film grain. From there the movie segues directly into glorious color, which the HD DVD delivers with astonishing image quality. The deep, rich colors pop off the screen in every vibrant shade. The picture is razor sharp with fine object detail visibility far surpassing anything possible on standard DVD, yet without a trace of edge enhancement artifacts. The contrast range is perfectly delineated with rich blacks, very good shadow detail, and a fine sense of depth. Since this is a launch title, I hesitate to rate the disc too highly, but at this point in time I can't find anything serious to complain about. "

Hard to imagine the above being said about a technology that is not capable of "yielding good results."

RBFilms

06-23-08, 01:21 AM

You are correct....for maybe 3 million they would sell it to you..:)

Almost everything above is true. Tiger is a very good encoder, but there is a lots of hardware, so price is very high. It has nice workflow, designed with Technicolor compressionists and it produces great results.

There is only one thing which I wouldn't agree: you can buy it. If you have money than there is no problem. It's on the Thomson site as one of the many products which you can buy. I've seen demo and it's a very professional product.

Tiger is an old name (when it was only for Technicolor use), new name is NexCode. More info:

http://www.thomson.net/GlobalEnglish/Nexcode/Nexcode/Pages/default.aspx

Andrew

amirm

06-23-08, 03:07 AM

Yes, there was omse edge enhancemnt and MPEG-2 yielded better results than VC-1 at the time. The VC-1 CODEC was not mature enough that he time to compete. That is a fact.
A fact is something you can prove. The only proof point you are offering is your opinion. Therefore, you are stating an opinion, not fact :).

These films are miles apart and do not make sense for comparison.
I was not comparing movies. You said that VC-1 was less mature than MPEG-2. So surely it would have generated subpar results if it were less performant than MPEG-2. Yet the reviewer, despite trying, can't find codec artifacts in Phantom or a dozen other movies encoded with it before your title came out. At the same time, he had no trouble though finding artifacts in Chronos. Hard to integrate these things and arrive at the conclusion that you reached regarding VC-2 and MPEG-2.

Once again, you are using selective choices that are not relative or comparable to make your point.
I could list you a number of technical reasons why VC-1 is superior to MPEG-2:

1. Adaptive block size. MPEG-2 only has fixed block size.
2. More advanced entropy coder.
3. Finer motion estimation logic.
4. In-loop filter.
5. More advanced encoding control.

And so on. You could turn off all the advanced features of VC-1 and have it be very similar to MPEG-2, yet still a bit more efficient. Did your compressionist try that before you declared MPEG-2 better? I am sure not.
Two reviews do not make an argument ...
And how many HD titles you have produced so far? I seem to recall only two ;) :). Yet you have gone from MPEG-2 to VC-1 and now AVC.

plus you can't argue VC-1 is great back in the day and then say there have been massive improvements in another thread.
VC-1 has improved substantially. But even day one, it produced better quality or Warner would not have bothered to use it . After all, they had access to plenty of MPEG-2 encoders, including the high quality one Sony used for BD titles. Recall that they could not wait to get back on VC-1 after having to use MPEG-2 for a short while. Is there someone here who thinks their quality went up with MPEG-2 and then back down with VC-1? No. And that is a company with in house encoding expertise which does not farm out work like you do.
Once again, your arguments do not serve to convince me of anything other that the fact that you have a very focused agenda that is seriously skewed by your investment in VC-1 and HD-DVD.
I have passion around VC-1. I am up front about that and will make no apology for it. The guy who led the engineering of Porsche suspension has similar passion I am sure. I am very proud of the team who has pushed the state of the art in codec research and software engineering in that manner. Are you up front about your agenda regarding Microsoft and its technologies?

Oh I remember. You had said that your agenda is to produce the best picture quality possible. If that is your #1 concern, then go and take Stacey out to lunch and learn more about video compression and quality than you can imagine. Really, if you have to come here and ask which encoder to use, then I can’t emphasize enough how much you need to improve your relationship with the people who build such things for a living (see any other content distributor asking the same question here?). You owe that to your customers who buy this stuff as eye and ear candy and demand that you be best informed about the choices you make. Don’t for heaven’s sake expect Microsoft to come to you. The only person to gain something out of that is you, so you need to “go to the mountain”, not the other way around ;) :).

scaesare

06-23-08, 12:19 PM

Andrew_HD

06-23-08, 03:09 PM

You are correct....for maybe 3 million they would sell it to you..:)

No, the price is more realistic, less than 500K $. Are you based on the story about first version of Tiger set-up for Technicolor for 1M $:)
NexCode is a next step after Tiger and priced more realistic, but still to expensive for most of the small studios.

Andrew

RBFilms

06-23-08, 10:56 PM

I have...and will yet again for those who may have missed it.

We used the Microsoft encoder.

We did not use the Cinevision.

Yes, the encodes for the HDDVD and the BD were virtually CBR due to the
very little variance between the averages and maxes.

Yes, these were maxed out to the ceiling. VBR would not have made a
difference at all.

Yes, we did contact MS and they were consulted for the encodes and very involved ... they absolutely had solid input on Nature's Journey with respect to the VC-1 Video Encodes.

For the HD-DVD disc we cannot peak higher than a 26 maximum due to the audio streams (a bandwidth issue):

DTS-HD High Res audio at 3mbps

DD+ 5.1 at 640kbps

DD 2.0 at 192kbps

So it was either a 24 average with a 26 max, or we max it all out at 26
CBR since there's not much variance between 24 and 26 and we have the
disc space.

As for the BD, right now the max is looking like it could be around 37
due to the DTS HD Master Audio which we are estimating will peak at least
around 8 Mbps. I will plan to process the BD stream separately at the
higher rate.

At the time we did this, these were probably the highest bit rate video encodes on the market.

Does that answer all of your questions? Clearly those speaking out of turn have misinformed many of the folks here on AVS ... and will continue to do so no matter how many times I post the facts.

These are the absolute facts....and MS was involved....and the encodes were done correctly, properly, professionally, and are perfectly well done.

We maximized the video encodes on the HD-DVD disc to the point where we had no headroom at all to spare. The BD Version still had space and I think we still had about 4mbps of hearom left.

Whoever said we would gain from VBR is absolutely incorrect. There would be no gain in performance or PQ as there was no headroom left to work with.

Richard, are you going to respond to the specific you asked for?

RBFilms

06-24-08, 12:02 AM

FACT - It was the “opinion” of everyone I spoke at every lab I dealt with before choosing Technicolor to do the work that VC-1 was not yet a mature enough CODEC and that we should go with MPEG-2 for CHRONOS. I did my homework on this before making a choice.

FACT - MPEG-2 was a mature CODEC and VC-1 was a much more recent technology ... with some growing pains. It was not a mature CODEC compared to MPEG-2 when we produced CHRONOS. This is an undeniable fact no matter how good or not good it may or may not have looked at the time.

FACT – If you argue about something you have not seen with your own eyes … or heard with your own ears … I think that makes it your opinion. Correct me if I am wrong.

FACT - Encode the exact same movie you are talking about back in the day with MPEG-2 and VC-1 and then compare the two ... and we would then have an apples to apples comparison and not the opinion of the reviewer or yourself.

FACT - You have no direct comparison between MPEG-2 and VC-1 from the early days with this film. Today, I know VC-1 would fare much better ... I know this for a fact … because it is more mature and better developed now then it was then …:).

FACT - All Video CODECS I have worked with have improved with age. There is no denying that VC-1 has improved with time ... as it has matured. This seems to be fairly normal for the evolution of a Video CODEC.

FACT - CHRONOS is a 30-year-old film shot for 500K with prints .. not negative …. that we pulled from a vault that had not been touched in years. We had very little money to do much cleanup ... albeit we did what we could. The Phantom feature was a recent big budget film shot on modern day stock using modern day lenses and equipment along with a far superior film transfer process. Also, the source material is better quality than CHRONOS. It is clean and has far less dirt, noise, and other stuff going on that hog up all of the “Video Encoder Horsepower.”

FACT - There was some minor edge enhancement done for CHRONOS by Crest National. You yourself admit this may be the cause of any artifacts you see.

FACT - The video encodes for CHRONOS look good compared to the source. We could only use Crest at the time with their old Tube Telecine ... since nobody else had a Telecine with a 15 Perf pulldown … except maybe for Imagica in Japan.

FACT - If I had my choice, I would have much preferred to use FotoKem or some other lab and done a Spirit transfer ... which is probably what they used for Phantom. However, I could not have afforded this even if it was an option.

FACT - You statement that VC-1 is superior is an OPINION ...not a fact. For it to be a fact, we need a direct comparison under proper laboratory conditions without any funny stuff going on. Has anyone done this yet so we can end this argument once and for all?

FACT - Trying to compare Phantom to CHRONOS does not give us a valid debate. Even film stock back in the day CHRONOS was shot could not come close to what we have today … not to mention everything lese involved.

FACT - MS worked with Technicolor on the Nature’s Journey Video Encodes … so they did what you … or whoever they spoke with at MS … told them to do. Are you saying you failed to tell them how to do the best job with VC-1 even though I gave you the opportunity to work directly with my lab?

QUESTION - Might there possibly be another reason why Warner uses VC-1 that you failed to mention? Don't I hear more people on AVS complaining about Warner Video Encodes than any other?

FACT - I am not arguing that MPEG-2 at lower bit rates is better than VC-1 now...just that it was then.

FACT - I agree that current day VC-1 is better than MPEG-2 at lower bit rates … but I still prefer AVC at lower bit rates in most cases. However, given enough bandwidth, I would still look at MPEG-2 before making a final choice for a Video Encode.

FACT - I am 6'8" tall and 240lbs. I do not fit in a Porsche so I have no idea of what their suspension is like. I just know I could crawl over one with any of my Land Cruisers ... especially those with MODS .... and crush it like a bug...:aren’t they just fancy, overpriced
Volkswagens anyway? :)

FACT - I have built great relationships with the people who "get it" on a level you either don't .. or do not care to admit ... and who are not so biased and invested in just one technology to the point that they rule out all others ... and who are open minded to looking at real world results ... and who are not constantly telling me I am wrong or crazy about what I see or hear.

FACT - I have a great relationships with the folks that I believe have the absolute best Video Encoder in the world...Technicolor. I did not come here to be fed the usual healthy dose of propaganda by MS.

FACT - Honestly, in the beginning, I did try to work with MS and was left in a lurch.

FACT - MS is so used to pushing their “stuff” on folks …. they know no other way. I assume you still think Vista is a good thing? Just because you tell me so .. it isn’t so if I can prove differently. I love my MAC … the funny part is that the only thing that really ever crashes on it is MS Office…:)

All silliness aside … that which I consider all of this to be … I am happy to do test encodes using the exact same content and let folks decide for themselves. Doesn’t this exist already? Didn’t the Doom 9 folks or someone like that already do this? I have said that I am willing to do this from day one.

So let's do a shootout and end this silly conversation once and for all. Let's go...I am ready to provide content for the test

A fact is something you can prove. The only proof point you are offering is your opinion. Therefore, you are stating an opinion, not fact :).

I was not comparing movies. You said that VC-1 was less mature than MPEG-2. So surely it would have generated subpar results if it were less performant than MPEG-2. Yet the reviewer, despite trying, can't find codec artifacts in Phantom or a dozen other movies encoded with it before your title came out. At the same time, he had no trouble though finding artifacts in Chronos. Hard to integrate these things and arrive at the conclusion that you reached regarding VC-2 and MPEG-2.

I could list you a number of technical reasons why VC-1 is superior to MPEG-2:

1. Adaptive block size. MPEG-2 only has fixed block size.
2. More advanced entropy coder.
3. Finer motion estimation logic.
4. In-loop filter.
5. More advanced encoding control.

And so on. You could turn off all the advanced features of VC-1 and have it be very similar to MPEG-2, yet still a bit more efficient. Did your compressionist try that before you declared MPEG-2 better? I am sure not.

And how many HD titles you have produced so far? I seem to recall only two ;) :). Yet you have gone from MPEG-2 to VC-1 and now AVC.

VC-1 has improved substantially. But even day one, it produced better quality or Warner would not have bothered to use it . After all, they had access to plenty of MPEG-2 encoders, including the high quality one Sony used for BD titles. Recall that they could not wait to get back on VC-1 after having to use MPEG-2 for a short while. Is there someone here who thinks their quality went up with MPEG-2 and then back down with VC-1? No. And that is a company with in house encoding expertise which does not farm out work like you do.

I have passion around VC-1. I am up front about that and will make no apology for it. The guy who led the engineering of Porsche suspension has similar passion I am sure. I am very proud of the team who has pushed the state of the art in codec research and software engineering in that manner. Are you up front about your agenda regarding Microsoft and its technologies?

Oh I remember. You had said that your agenda is to produce the best picture quality possible. If that is your #1 concern, then go and take Stacey out to lunch and learn more about video compression and quality than you can imagine. Really, if you have to come here and ask which encoder to use, then I can’t emphasize enough how much you need to improve your relationship with the people who build such things for a living (see any other content distributor asking the same question here?). You owe that to your customers who buy this stuff as eye and ear candy and demand that you be best informed about the choices you make. Don’t for heaven’s sake expect Microsoft to come to you. The only person to gain something out of that is you, so you need to “go to the mountain”, not the other way around ;) :).

RBFilms

06-24-08, 12:03 AM

I guess they think I have money....they told me a few millin...unless I wanted Tiger Cub...I could get that for 300K. Are you talking about Tiger Cub or Tiger?

No, the price is more realistic, less than 500K $. Are you based on the story about first version of Tiger set-up for Technicolor for 1M $:)
NexCode is a next step after Tiger and priced more realistic, but still to expensive for most of the small studios.

Andrew

online

06-24-08, 12:18 AM

Here we go again...

amirm

06-24-08, 12:34 AM

FACT - You have no direct comparison between MPEG-2 and VC-1 from the early days with this film. Today, I know VC-1 would fare much better ... I know this for a fact … because it is more mature and better developed now then it was then …:).
If this is an example of what fact is, I am afraid to ask what you think fiction is!

DVD Forum, fully 3 years before either format came to market, did a double blind shoot out of codecs. MPEG-2 was bid by two companies. AVC I think by 4. And VC-1 of course by Microsoft. Tests were conducted at 7.7 mbit/sec, compared to uncompressed reference and MPEG-2 at 24 mbit/sec. VC-1 easily beat out all other codecs at the same 7.7 mbit/sec. Yes, this included MPEG-2. What's more, it matched the MPEG-2 reference at 24 mbit/sec on two clips I believe, and nearly matched it on the rest despite running at a third of the data rate. VC-1 would have never been accepted into HD DVD and later, BD, without beating MPEG-2. The blind test by the way, was conducted by many members of DVD forum from CE companies to studios.

Warner then, two years later, conducted its own private tests comparing MPEG-2 and AVC to VC-1 with the result being them standardizing on VC-1. They used the data rates appropriate for the format at this time with fresh “codec buster” titles. They made their own evaluation, not impacted by anyone else’s opinion.

Further, 50+ titles came out of "WMV-HD" which is the CBR/internet mode of VC-1. All of these titles fit on red laser DVDs and with the same capacity and bandwidth of standard DVD, managed to squeeze in 3X to 6X the number of pixels (720p and 1080p). MPEG-2 in the same situation would have no prayer of doing anything remotely close to this. Again, all of this predated Chronos production by 3+ years.

WMV-HD was also used for digital cinema presentation in early years (5+ before HD optical format launch). It was compared and criticized against film projection on 30 feet screens at events such as Sundance film festival where it was the only electronic format allowed for submission at the time.

Really, third-hand information like you have only goes so far. How can you possibly be aware of any and all comparisons Microsoft may have participated in to state things as strong as you have?

scaesare

06-24-08, 01:05 AM

I have...and will yet again for those who may have missed it.

{Many words about bitrate snipped}

Why do you continue to ignore the incorrect interlace mode choice for the encode?

For some reason you did not address it when Stacey originally brought it up. You did not address it when I asked you about it shortly thereafter. And you do not address it now.

Yet you draw conclusions from what appears to be a flawed test.

I do not understand this from somebody who professes to stand for video excellence. :confused:

Andrew_HD

06-25-08, 04:59 AM

I guess they think I have money....they told me a few millin...unless I wanted Tiger Cub...I could get that for 300K. Are you talking about Tiger Cub or Tiger?

There is no Tiger anymore. Nexcode is the new name and it's sold by French division of Thomson (as I understand they developed it) and price is actually about 250K $. There was some purchase in Europe recently for this price.

Andrew

RBFilms

06-25-08, 10:27 AM

RBFilms

06-25-08, 10:30 AM

That is not the same Tiger I am talking about ....

That is Tiger Cub ... of NexCode if you prefer ... which sells for 300K here in the US.

I am taking about the original Tiger encoder at Technicolor...which is the only one of its kind in the world. Tiger Cub / NexCode does not have all of the features or power of the Tiger system at Technicolor....at least that is my understanding ... which comes from a very good source.

There is no Tiger anymore. Nexcode is the new name and it's sold by French division of Thomson (as I understand they developed it) and price is actually about 250K $. There was some purchase in Europe recently for this price.

Andrew

benwaggoner

06-25-08, 10:32 AM

RBFilms

06-25-08, 10:34 AM

MPEG-2 sucks at low bit rates...everyone knows that ... so what is your point?

In our tests, MPEG-2 looked beter at 24mbps than the early versions of VC-1. I am not saying VC-1 sucked...just that it was not there yet. It looks much better now...but we prefer the Tiger AVC Encoder for our critical work.

I can make such confident statements because they are based on my own comparisons and personal experience. That is why I can easily make clear and concise statements ... you cannot discount my personal experience and testing...albeit, I can see you will continue to try...:)

If this is an example of what fact is, I am afraid to ask what you think fiction is!

DVD Forum, fully 3 years before either format came to market, did a double blind shoot out of codecs. MPEG-2 was bid by two companies. AVC I think by 4. And VC-1 of course by Microsoft. Tests were conducted at 7.7 mbit/sec, compared to uncompressed reference and MPEG-2 at 24 mbit/sec. VC-1 easily beat out all other codecs at the same 7.7 mbit/sec. Yes, this included MPEG-2. What's more, it matched the MPEG-2 reference at 24 mbit/sec on two clips I believe, and nearly matched it on the rest despite running at a third of the data rate. VC-1 would have never been accepted into HD DVD and later, BD, without beating MPEG-2. The blind test by the way, was conducted by many members of DVD forum from CE companies to studios.

Warner then, two years later, conducted its own private tests comparing MPEG-2 and AVC to VC-1 with the result being them standardizing on VC-1. They used the data rates appropriate for the format at this time with fresh “codec buster” titles. They made their own evaluation, not impacted by anyone else’s opinion.

Further, 50+ titles came out of "WMV-HD" which is the CBR/internet mode of VC-1. All of these titles fit on red laser DVDs and with the same capacity and bandwidth of standard DVD, managed to squeeze in 3X to 6X the number of pixels (720p and 1080p). MPEG-2 in the same situation would have no prayer of doing anything remotely close to this. Again, all of this predated Chronos production by 3+ years.

WMV-HD was also used for digital cinema presentation in early years (5+ before HD optical format launch). It was compared and criticized against film projection on 30 feet screens at events such as Sundance film festival where it was the only electronic format allowed for submission at the time.

Really, third-hand information like you have only goes so far. How can you possibly be aware of any and all comparisons Microsoft may have participated in to state things as strong as you have?

Golgot13

06-25-08, 11:06 AM

Hi at all,

I come back after made many BD project (in H264 and one in VC1...)
and Euro2008 (1/2final today and tomorrow and final will be this sunday).

Sony also has a new AVC Encoder coming out which I believe uses some of Tiger's technology. However, I am not sure which SDK it is built on. Does anyone know about this?

This is not true:
Tiger AVC is a old name. Andrew_HD said true now it's Nexcode (French company name ;) ).
MX1000, new encoder from Sony, was made by two japaneses guy without Nexcode SDK (I'm not sure there is a SDK).
Nexcode work on Linux OS (very nice OS...) but MX1000 use Windows2003 server OS.
I think your rumour, about Sony use SDK from Tiger, came that Sony Japan bought a Nexcode encoder Turnkey.
Maybe they will buy their technology and use it on Sony product (like MS and some software or video codec... :rolleyes:)
Or they will spy the Nexcode solution (reverse engineering like some companies...)

We have already determined that AVC is a better CODEC yielding superior results for our applications. Now I am looking for any information on the best AVC Encoder to use.

Professional opinions about these AVC Encoders would be appreciated...:)

You're not alone, many people know what I think about quality between AVC and VC1. :D
AVC is really better, Nexcode encoder can kill CinePSE/PeP (there is many preprocess filter).

I understand Richard you have access at some nice video footage. Can you share a video file
in 1920x1080@24P to make some test comparison ?
MSU don't have lot of HD video source to test codec or implementation of codec.
(Ben and Amirm, VC1/MS will be on MSU comparison this year ? Or you're afraid
because you know the result...)

(@ Andrew_HD, Poland was eliminated of Euro2008 quickly :(
and there was not England team ...
I hope to see Germany and Russian team on final: there is some guy with polski name (Podolski and Kolodin))

Golgot13

06-25-08, 11:08 AM

And Microsoft is also a significant participant in the H.264 patent pool.

I will soon put a patent to use the letter "p" and all book or magazin which use this letter
can be considered developped with my patent...

Golgot13

06-25-08, 11:34 AM

DVD Forum, fully 3 years before either format came to market, did a double blind shoot out of codecs. MPEG-2 was bid by two companies. AVC I think by 4. And VC-1 of course by Microsoft. Tests were conducted at 7.7 mbit/sec, compared to uncompressed reference and MPEG-2 at 24 mbit/sec. VC-1 easily beat out all other codecs at the same 7.7 mbit/sec. Yes, this included MPEG-2. What's more, it matched the MPEG-2 reference at 24 mbit/sec on two clips I believe, and nearly matched it on the rest despite running at a third of the data rate. VC-1 would have never been accepted into HD DVD and later, BD, without beating MPEG-2. The blind test by the way, was conducted by many members of DVD forum from CE companies to studios.

Yes the test was not open at really specialist company or OpenSource community....
You're update your website about VC1 with old news after many request from me.
(You like PSNR when it say VC1 is good...)

Further, 50+ titles came out of "WMV-HD" which is the CBR/internet mode of VC-1. All of these titles fit on red laser DVDs and with the same capacity and bandwidth of standard DVD, managed to squeeze in 3X to 6X the number of pixels (720p and 1080p). MPEG-2 in the same situation would have no prayer of doing anything remotely close to this. Again, all of this predated Chronos production by 3+ years.

About WMV-HD, I read MS bought a company to developp it (like NTFS format), it's true (?).

WMV-HD was also used for digital cinema presentation in early years (5+ before HD optical format launch). It was compared and criticized against film projection on 30 feet screens at events such as Sundance film festival where it was the only electronic format allowed for submission at the time.

Yes WMV-HD was a nice format, I made many test with it.
But you know that MPEG team need lot of time to finalize H264 AVC codec
(There is, since end of summer 2007, a new format H264 SVC)

Really, third-hand information like you have only goes so far. How can you possibly be aware of any and all comparisons Microsoft may have participated in to state things as strong as you have?

Yes, MS participate at some comparison if they're sure to win (?).
Why we don't see a comparison codec with VC1 on MSU or others ?
Why you don't want to participate at Codec Challenge on Doom9 with 24fps video?

Like many people, I need to see to believe (VC1's better than other codec)
I never see something from MS...

scaesare

06-25-08, 01:16 PM

I am sure I answered this but let me be clear about your question. What is the correct interlace mode?

Are you asking why we took 1080i source material and encoded it as 1080i?

We always try...and emphasize try...to stay in the native form,at of the source for all of our work...if it is 48/24 audio...it stays as 48/24...if it is 1080i...we always try to encode it as 1080i.

We believe in staying with the Native Format and that the playback equipment should handle and conversions if at all required. It is not always possible, but we try.

What is wrong with the "Interlace Mode" ... I am not clear on what the issue is. Please advise.

I am positive you have NOT answered this, as I attempted to follow up with you when it was originally posted by Stacey, and you didjn't address it.

When I reposted the issue to you shortly thereafter, you ignored it.

Amir just quoted it again verbatim a couple of days ago, and you didn't address it.

I honestly don't know if you are glossing over it on purpose, but one of THE key people who know what the VC1 encoder can do had this to say about the encode upon which you drew conclusions:

"I finally got around to looking at the bitstream on Nature's Journey and I am puzzled. It seems to be using the interlace frame encoding mode of PEP instead of the interlaced field mode. There are two interlaced modes, field and frame. Frame is used when there is no motion or very slow motion while field is used in normal motion. You can force an encode to be one, the other or both (auto). Normally we recommend the auto mode. A higher bitrate is required if you use the less effecient mode for the motion in the frame.

{point about bitrate snipped}

Truth be told, frame is the red headed step child. Our focus is always progressive mode first, followed by field and when time permits, frame."

There appears to ahve been an incorrect interlacing mode selected for the test. How that mistake was made (Technicolor's fault, MS's fault, simple miscimmunication, human error, disruption in the force, etc...) doesn't really matter. What does matter is that you are espousing your opinion based upon a flawed test.

What baffles me is why when this has been pointed out to you, it has been ignored. Your "DRS" charter definition on your website says:

"The Digital Reference Standard is your guarantee of R&B Films' commitment to the highest attainable quality in home entertainment. Each disc is mastered for the best possible reproduction of video and audio, routinely surpassing today's standards in our ongoing pursuit of the ultimate viewing experience....

R&B Films works closely with industry-leading engineers and facilities to create a digital master with video and audio fidelity rendered at its zenith. This devotion to detail continues through the authoring and replication processes to arrive at a finished disc capable of presenting the full aural and visual impact intended by its creators.

Why wouldn't you jump on a chance to correct a potentially significant flaw in your testing if it helped you achieve your stated goals?

amirm

06-26-08, 01:04 AM

MPEG-2 sucks at low bit rates...everyone knows that ... so what is your point?
I showed you that not only did MPEG-2 suck at low rates, but also underperforms at higher rates. How else do you think VC-1 nearly matched its quality at a third of the data rate? And once more, Warner tested it at HD DVD/BD data rates, not the original lower bit rates. Further, after being forced to use MPEG-2 for their titles for BD, they could not wait to get back to VC-1.

So it seems that you are data is at odds with people who have produced 100X more HD titles than you.

I can make such confident statements because they are based on my own comparisons and personal experience.
Did you encode the content? If not, then you are at the mercy of whoever did that work, without due involvement from Microsoft. And from Ben’s comments, and your confirmation of what was used, you relied on a different encoder than anyone else encoding titles.

That is why I can easily make clear and concise statements ... you cannot discount my personal experience and testing...albeit, I can see you will continue to try...:)
I can easily discount many of your statements. Sorry but I can. Remember calling Chronos HD DVD "interlaced" due to the same flags everyone else used for progressive content? How long did you argue here that HD DVD was an interlace format and not progressive? Seemed like you even fought with the reviewer as I quoted earlier.

Then we have Nature's Journey where you advertized the soundtrack to be 96Khz, only to have me test and find out it was 48Khz. From what I recall, you were not aware of the fact that the licensed stock music tracks only came in 48Khz.

Mind you, I am not discounting your years of experience in the field. I am sure you have a lot of expertise in many areas. But when it comes to new technology such as VC-1, I would think it makes sense to make fewer generalizations. Per my other note, you have only done two HD titles so even if one assumes your results are what you say they are, they cannot possibly be representative of merits of VC-1 vs AVC in general. If I someone encodes two movies with VC-1 and declares it better than AVC, you are going to believe that person? Especially if that person has zigzagged from MEPG-2 to VC-1 and now AVC?

Golgot13

06-26-08, 04:39 AM

I showed you that not only did MPEG-2 suck at low rates, but also underperforms at higher rates. How else do you think VC-1 nearly matched its quality at a third of the data rate? And once more, Warner tested it at HD DVD/BD data rates, not the original lower bit rates. Further, after being forced to use MPEG-2 for their titles for BD, they could not wait to get back to VC-1.

At higher bitrate (more than 20Mbps-25Mbps), VC1, H264 and MPEG2 codec give same result without video preprocessing.
Today CinePSE/PeP have a really nice preprocessing tool.
Don't forget I was the first who said VC1 is a codec for BD, it give a nice performance at higher bitrate and low CPU.
But today, all new GPU can decode H264...

Mind you, I am not discounting your years of experience in the field. I am sure you have a lot of expertise in many areas. But when it comes to new technology such as VC-1, I would think it makes sense to make fewer generalizations. Per my other note, you have only done two HD titles so even if one assumes your results are what you say they are, they cannot possibly be representative of merits of VC-1 vs AVC in general. If I someone encodes two movies with VC-1 and declares it better than AVC, you are going to believe that person? Especially if that person has zigzagged from MEPG-2 to VC-1 and now AVC?

If you're sure about VC1 and H264, you're welcome to show us in Doom9 forum.
Or you can participate at MSU codec comparison...
I'm sure this summer we will know which codec is the better.

Andrew_HD

06-26-08, 11:49 AM

(@ Andrew_HD, Poland was eliminated of Euro2008 quickly :(
and there was not England team ...
I hope to see Germany and Russian team on final: there is some guy with polski name (Podolski and Kolodin))

Yep:(

The joke is that Germany didn't score with Poland. Podolski was born in Poland and his family moved to Germany when has was 8- something like this:)

I'm wating for Germany -Russia final,

Andrew

Andrew_HD

06-26-08, 11:56 AM

That is not the same Tiger I am talking about ....

That is Tiger Cub ... of NexCode if you prefer ... which sells for 300K here in the US.

I am taking about the original Tiger encoder at Technicolor...which is the only one of its kind in the world. Tiger Cub / NexCode does not have all of the features or power of the Tiger system at Technicolor....at least that is my understanding ... which comes from a very good source.

I'm not sure if you're right. Was it set up by French division of Thomson?

Tiger was the first version and set up for Technicolor by Thomson for stupid money.
After that they've been working on it and if NexCode is different than it can be only better. I'm not 100% sure, but I don't think that you can do much more with Technicolor's Tiger than with NexCode- as I understand NexCode is next generation of Tiger.

Andrew

RobertR1

06-26-08, 01:24 PM

golgot,

What's so tough to understand that MS has nothing to gain on these "tests" you keep wanting them to enter. It does nothing for their business. Winning over e-Warriors means nothing as long as they're winning over studios. Did you read that post from Stacey Spears about another studio's codec shootout recently where VC1 won out and MS wasn't even involved? You see, that does a lot more for their business than trying to prove the "golgot13's" of the world wrong.

amirm

06-26-08, 02:31 PM

At higher bitrate (more than 20Mbps-25Mbps), VC1, H264 and MPEG2 codec give same result without video preprocessing.
While you have said the closest thing to something we would both agree with :), do you have an opinion as to why every studio has switched away from MPEG-2 to VC-1 or AVC? Even our vocal RBFilms has done that despite defending MPEG-2 initially. Are they all cofused?

Golgot13

06-26-08, 03:53 PM

Tiger was the first version and set up for Technicolor by Thomson for stupid money.
After that they've been working on it and if NexCode is different than it can be only better. I'm not 100% sure, but I don't think that you can do much more with Technicolor's Tiger than with NexCode- as I understand NexCode is next generation of Tiger.

No, Thomson encoder was developped in USA with a university (developpment)
and Technicolor (suggestion).
Today, if I understand, Nexcode a division of Thomson sell it and
continue the developpment (there is some nice last feature since Tiger).

@RobertR1
I don't read the post of Stacey.
Can you give me the link about it?
I prefer to have a codec challenge like on Doom9:
all people can participate and prouve what they say.
It's same on MSU, all participant (developper team) can prouve
they have the best solution (and codec)

RobertR1

06-26-08, 05:09 PM

@RobertR1
I don't read the post of Stacey.
Can you give me the link about it?
I prefer to have a codec challenge like on Doom9:
all people can participate and prouve what they say.
It's same on MSU, all participant (developper team) can prouve
they have the best solution (and codec)

That's weird because you actually replied to his post!: http://www.avsforum.com/avs-vb/showpost.php?p=14071238&postcount=446

Anyway, they have no incentive to win over e-warriors. They seem to be winning where it matter. Their (MS) prediction on here is that you'll see a lot more titles using VC1 as the year progresses. We'll all be here to see if they were right or wrong. In the meantime, stick with the codecs you prefer and quit wasting your time with this crusade.

manikin

06-26-08, 09:33 PM

That's weird because you actually replied to his post!: http://www.avsforum.com/avs-vb/showpost.php?p=14071238&postcount=446

Anyway, they have no incentive to win over e-warriors. They seem to be winning where it matter. Their (MS) prediction on here is that you'll see a lot more titles using VC1 as the year progresses. We'll all be here to see if they were right or wrong. In the meantime, stick with the codecs you prefer and quit wasting your time with this crusade.

Actually July 29th should be an interesting date and pretty telling about the future.
Doomsday is a day and date Universal release on Blu-ray. This is the first new release for Uni. No Video specs have been released as yet, audio is DTS-HD MA, If they hold true to their roots they stick with VC-1, otherwise it will leave only Warner as a major using VC-1. Fox, Sony, and Disney all appear to be sticking with AVC.

amirm

06-26-08, 09:38 PM

Actually July 29th should be an interesting date and pretty telling about the future.
Doomsday is a day and date Universal release on Blu-ray. This is the first new release for Uni. No Video specs have been released as yet, audio is DTS-HD MA, If they hold true to their roots they stick with VC-1, otherwise it will leave only Warner as a major using VC-1.
That kind of thing doesn't mean anything even if they use VC-1. Paramount doesn't do its own encoding and has no set policy one way or the other on codecs.

Fox, Sony, and Disney all appear to be sticking with AVC.
Sony is the only company you can rely on to use AVC at all cost.

amirm

06-26-08, 09:40 PM

Oops. I thought you were talking about Paramount :). Universal is in the same situation although they did standardize on VC-1. So if they choose other codecs consistently, that can mean a little :).

manikin

06-26-08, 09:50 PM

Oops. I thought you were talking about Paramount :). Universal is in the same situation although they did standardize on VC-1. So if they choose other codecs consistently, that can mean a little :).

Uni and VC-1 can be counted on at the same level as Warner and VC-1. So having them switch would be if not a WTF? at least a curiosity. Especially when this is their 1st Blu only title. Care to hint as what to expect? ;)

amirm

06-26-08, 09:57 PM

Uni and VC-1 can be counted on at the same level as Warner and VC-1.
Not at all. Warner has its own in house compression facilities where they have gained significant experience with the various tools. They have their workflow, etc. Uni like Paramount, has third-party companies doing the work for them. It is possible for BD they have decided to use a different shop and that shop may opt to use a different codec. Or Not.

So having them switch would be if not a WTF? at least a curiosity. Especially when this is their 1st Blu only title. Care to hint as what to expect? ;)
I have no information actually as to which directly they may be going. I just don't consider their choice nearly as newsworthy as Warner using something else.

RobertR1

06-26-08, 11:46 PM

Amir,

Let's get to a real question. How has retirement been? working on anything new?

amirm

06-27-08, 12:24 AM

Amir,

Let's get to a real question. How has retirement been?
Retirement is incredible. I have not felt this relaxed since I was a kid! No matter how much one loves his job, there is still some amount of stress that cannot be taken out of it.

working on anything new?
Have kicked around half a dozen ideas. Nothing concrete yet. Once something comes out of it, I will report back :).

tteich

06-27-08, 04:50 AM

[...]
(@ Andrew_HD, Poland was eliminated of Euro2008 quickly :(
and there was not England team ...
I hope to see Germany and Russian team on final: there is some guy with polski name (Podolski and Kolodin))
OT: Podolski was born in Poland. But his citizenship is Germany.
Russia vs. Germany or even Turkey would have been nice, however, since yesterday we know that the endgame will be Germany vs. Spain :)

Joe Bloggs

06-27-08, 05:46 AM

This thread is comparing AVC/H.264 with VC-1, but also talking about how they compare to Mpeg2. How would the TDVCodec video codec compare to the current Blu-ray codecs (and isn't it bad having codec in the name of the codec :)). Do you think this codec (or any other codecs) will actually get added to Blu-ray, and if so, do you know when?

Do you think the current codecs may get updated to handle new things like 3D properly (even though could probably work as they are, anyway, if the video was encoded as red+blue, or the secondary video thing was used). On a similar note, when do you think the specs and chips and codecs in the players will be updated to handle more other things that we want (increased rates/colour/res/maybe 3d) (the codecs themselves may already be able to handle some of these but maybe not the ones built into the players?)?

Sorry if slightly OT.

Tom McMahon

06-28-08, 03:43 PM

This thread is comparing AVC/H.264 with VC-1, but also talking about how they compare to Mpeg2.

It is mostly about comparing various H.264/AVC encoder implementations versus various SMPTE VC-1 encoder implementations.

amirm

06-28-08, 07:30 PM

How would the TDVCodec video codec compare to the current Blu-ray codecs (and isn't it bad having codec in the name of the codec :)).
I have not looked at this one much but thought it was some kind of extension to MPEG-2 for 3-D, and not a new coding (compression method).

Do you think this codec (or any other codecs) will actually get added to Blu-ray, and if so, do you know when?
No, they will not add any new codecs. Doing so obsoletes just about every player, will face formidable resistance from CE companies (threatens their patent position in existing codecs), etc.

Now for obscure optional stuff, maybe they add something. DVD for example, allows WMA Pro for optional audio.

Do you think the current codecs may get updated to handle new things like 3D properly (even though could probably work as they are, anyway, if the video was encoded as red+blue, or the secondary video thing was used).
Depending on the method, the codec needs no change. The impact is really the base "systems layer" which would have to specify the relationship between the two tracks, how the video output would work, etc.

On a similar note, when do you think the specs and chips and codecs in the players will be updated to handle more other things that we want (increased rates/colour/res/maybe 3d) (the codecs themselves may already be able to handle some of these but maybe not the ones built into the players?)?
While HD DVD was a going concern, we were working on extensions for 10-bit, and 4:2:2. Also did a bit of thinking about 3-D but it was out of reach. There were some other development elsewhere that I can't talk about.

Sorry if slightly OT.
Lot more useful conversation than "let's go out and have a fist fight" if you ask me :).

hellokeith

06-29-08, 03:13 AM

It is mostly about comparing various H.264/AVC encoder implementations versus various SMPTE VC-1 encoder implementations.

Tom,

From a workflow standpoint, do implementations follow the "you get what you pay for' rule? Can you go into a bit more detail?

Tom McMahon

06-29-08, 01:05 PM

Tom,

From a workflow standpoint, do implementations follow the "you get what you pay for' rule? Can you go into a bit more detail?

I think you get at most what you pay for. Certainly encoder vendors have leveraged a lot of what they learned from MPEG-2 encoder products. Then you also have Moore's law continuing to work for them in terms of RAM and general computing horsepower that they can apply to the H.264/AVC and VC-1 encoding problems.

But all of the vendors are still getting smarter about these respective codecs. There are a ton of new features and parts of the respective encoding toolkits that they are still learning how to exploit.

Some of the tools available to the encoder manufacturers only work well with certain types of content. So a tool that improves coding efficiency or quality with film (grain) may not exhibit much benefit with CGI. It's all over the map.

So saying one codec is better than another is pretty much meaningless. As we all found out in the varoius rounds of testing, it is extremely difficult to compare codecs, even with identical content and test conditions. Again, what you typically end up comparing is differerent encoder implementations as snapshotted at that point in their development as well as who is driving or steering or setting up each encoder.

Zodiaque

06-30-08, 12:05 AM

I could list you a number of technical reasons why VC-1 is superior to MPEG-2:

1. Adaptive block size. MPEG-2 only has fixed block size.
2. More advanced entropy coder.
3. Finer motion estimation logic.
4. In-loop filter.
5. More advanced encoding control.

Usefull at high/medium quantisation but not at low quantisation. The best technical encoding option at high bitrate are psy optimisation and MPEG2/VC1/H264 can implement these Psy optimisation based on first at all pre-process, Adaptative Quantisation (dark, luma, texturing masking), RDO psy optimisation ... etc.

DVD Forum, fully 3 years before either format came to market, did a double blind shoot out of codecs. MPEG-2 was bid by two companies. AVC I think by 4. And VC-1 of course by Microsoft. Tests were conducted at 7.7 mbit/sec, compared to uncompressed reference and MPEG-2 at 24 mbit/sec. VC-1 easily beat out all other codecs at the same 7.7 mbit/sec. Yes, this included MPEG-2. What's more, it matched the MPEG-2 reference at 24 mbit/sec on two clips I believe, and nearly matched it on the rest despite running at a third of the data rate. VC-1 would have never been accepted into HD DVD and later, BD, without beating MPEG-2. The blind test by the way, was conducted by many members of DVD forum from CE companies to studios.

Come on Amirm ... it's really ridiculous.

- VC1 or even H264 at 8Mbps never beat MPEG2 at 24 Mbps in all case. Perhaps MPEG2 with CBR and only IFrame "gop" but certainely not good MPEG2 encoder with full MP@HL functionality.
- 3 years old VC1 implementation is really poor VC1 codec.
- I make complete test with Casino Royal Movie at 6.9 Mbps (high grain level). Codec like x264 beat easely your last VC1 SDK in all case with the same pre-process. H264 is really superior to VC1 at "low bitrate".
- For really high bitrate (BD50, max at 40 Mbits) IMO it's really hard to difference between codec.

benwaggoner

06-30-08, 12:09 AM

Usefull at high/medium quantisation but not at low quantisation. The best technical encoding option at high bitrate are psy optimisation (like AQ for example) and MPEG2/VC1/H264 can implement these Psy optimisation.
What of those wouldn't be useful at low quantization (or, better stated, wouldn't be useful in achiving lower quantization than MPEG-2 at the same bitrate)?

The only one I can think of that gets less useful at higher bitrates is the loop filter, since it's proportional to quantization.

amirm

06-30-08, 01:20 AM

Come on Amirm ... it's really ridiculous.
- VC1 or even H264 at 8Mbps never beat MPEG2 at 24 Mbps in all case.
"All cases?" Who said "all cases?" I said in DVD forum tests, this is what happened:

"What's more, it matched the MPEG-2 reference at 24 mbit/sec on two clips I believe, and nearly matched it on the rest despite running at a third of the data rate."

That's two clips matching and others coming close.

Perhaps MPEG2 with CBR and only IFrame "gop" but certainely not good MPEG2 encoder with full MP@HL functionality.
These were three-pass encoded (VBR, and with "full MP@HL") with the option to hand tweak every frame if the supplier wanted to do that. There was not a person stupid enough to encoding at Iframe-only. These are not hobbyists.

Yes, all benchmarks are faulty and some are faultier than others. The DVD Forum tests were no different. But you all keep demanding independent tests, so here was one. Of course, no sooner than the results come out that you cry foul. This is why there is no point in participating in challenges like the one being called in this thread where there is no business value. If you win, as we did with DVD Forum, then some guy behind an alias shows up and cries foul without even knowing the test details, who provided the encoder, etc. If the results can be dismissed so easily, why waste time with people like this? Might it not be useful to first ask what encoder was used before making assumptions like you did?

- I make complete test with Casino Royal Movie at 6.9 Mbps (high grain level). Codec like x264 beat easely your last VC1 SDK in all case with the same pre-process. H264 is really superior to VC1 at "low bitrate".
So if we say 10 more times than the SDK that is given away for free is not the same as the professional tool used for HD encoding, will the message really sink in?

And where did you get uncompressed sources for Casino Royal? Let me guess, you took the already compressed material and compressed it? And that is supposed to be of what interest to people here?

Besides, who is to say your tests were right and your results are as you claim? Maybe if I looked at the output I would conclude that VC-1 is better. You didn’t just say that based on PSNR measurements, did you?
- For really high bitrate (BD50, max at 40 Mbits) IMO it's really hard to difference between codec.
Since BD with the above data rate is the only format of interest in this forum, can we conclude that you have nothing to add anymore to what folks like to know here then? If so, that would be the best possible way to end this thread.

trbarry

06-30-08, 01:34 AM

Uhhh, I guess that means MSFT is probably not going to participate in a Doom9 shootout?

As I wrote to Ben last fall in this same thread above (http://www.avsforum.com/avs-vb/showthread.php?p=12125545&highlight=discredit#post12125545):

Yes but when we last visited the Doom9 shootout the X264 implementation of the time appeared to be somewhat beating the VC-1 implementation of the time. You could choose to demonstrably win the next round if VC-1 was good enough. And I thought you promised to do that.

Or you could choose to discredit the contest.

- Tom

amirm

06-30-08, 01:45 AM

Uhhh, I guess that means MSFT is probably not going to participate in a Doom9 shootout?
It this because of what I said? If so, there should be no conclusion drawn from it as I don't work there. But if I were to bet, what you said that would be my conclusion.

Zodiaque

06-30-08, 08:35 AM

The only one I can think of that gets less useful at higher bitrates is the loop filter, since it's proportional to quantization.

Simply because advantage is statiquely less and less important at low quant level. Perceptual quality is like a(1-e-bitrate) function. At really high bitrate, high bitrate augmentation doesn't mean high quality improuvement.

Yes, all benchmarks are faulty and some are faultier than others. The DVD Forum tests were no different. But you all keep demanding independent tests, so here was one. Of course, no sooner than the results come out that you cry foul. This is why there is no point in participating in challenges like the one being called in this thread where there is no business value. If you win, as we did with DVD Forum, then some guy behind an alias shows up and cries foul without even knowing the test details, who provided the encoder, etc. If the results can be dismissed so easily, why waste time with people like this? Might it not be useful to first ask what encoder was used before making assumptions like you did?

Well in this case the DVD forum is completely ridiculous test.

So if we say 10 more times than the SDK that is given away for free is not the same as the professional tool used for HD encoding, will the message really sink in?

Well I have a recent CinevisionPSE implementation too and the result is not really better.

And where did you get uncompressed sources for Casino Royal? Let me guess, you took the already compressed material and compressed it? And that is supposed to be of what interest to people here?

Simply BD50 source: H264 encoding at 35 Mbps. IMO real uncompressed source -> encoding at 6.9 Mbps will produce equivalent result to H264 35 Mbps source -> reencoding at 6.9 Mbps simply because there are a really high quality difference between sources and encodings. And source is the same for all the codec.

Joe Bloggs

06-30-08, 08:36 AM

Let me guess, you took the already compressed material and compressed it? And that is supposed to be of what interest to people here?
Aren't some encodes done from compressed D5 HD masters? These might cause visible artefacts in the final encoding or otherwise lower the quality of the final encoding. Maybe all encodings/Blu-ray titles would be better if they were made from uncompressed original masters? (as well as, for film based stuff, a scan from the highest quality source available with a great film scanner at maybe 4k for 35mm (or more for original 65mm films). Maybe that would increase studio costs though.

Tom McMahon

06-30-08, 10:57 AM

Aren't some encodes done from compressed D5 HD masters? These might cause visible artefacts in the final encoding or otherwise lower the quality of the final encoding. Maybe all encodings/Blu-ray titles would be better if they were made from uncompressed original masters? (as well as, for film based stuff, a scan from the highest quality source available with a great film scanner at maybe 4k for 35mm (or more for original 65mm films). Maybe that would increase studio costs though.
Yes, Yes, Yes, and Yes. I and others who've spent a lot of time in this space can see D5 compression artifacts at 1+ Gbit even before they go through any of the other codecs.

scaesare

06-30-08, 11:10 AM

Well in this case the DVD forum is completely ridiculous test.

Please quantify your justification for this.

benwaggoner

06-30-08, 12:04 PM

Uhhh, I guess that means MSFT is probably not going to participate in a Doom9 shootout?
Oh, it's something I might do personally, depending on the shootout. Of course, any shootout can only prove anything relative to the scope of the shootout. The problem is people read way too much into them; conflating a web bitrate test with an optical bitrate test and that kind of thing, assuming there's some sort of easily quantifiable linear scale of "codec goodness."

I have to say I've been pretty disappointed by the 24p clip process. Lots of people clamored for it, but when I actually posted a rough cut of it and asked for feedback, I didn't get much, and very little from people who'd been complaining about not having it yet. Nor was there much interest or use from the complete 1080i clip I did post.

I still plan to finish it up, but the heat versus light ratio on these discussions certainly hasn't been motivating me to push anything else aside to focus on it. My main motivation for doing it is to get some interesting 24p film source content into the community for a variety of uses. It's a lot of work, and I'm really not inspired to do it just to enable someone (even myself!) to score points on forums.

benwaggoner

06-30-08, 12:10 PM

Simply because advantage is statiquely less and less important at low quant level. Perceptual quality is like a(1-e-bitrate) function. At really high bitrate, high bitrate augmentation doesn't mean high quality improuvement.
Sure. I don't know that I'm grasping your point, though. Isn't there a broad consensus that all decent codecs converge at high bitrates?

Well in this case the DVD forum is completely ridiculous test.

Well I have a recent CinevisionPSE implementation too and the result is not really better.
"Have" or "purchased"? PSE is absolutely not a tool that someone who hasn't been trained in can get optimum results out of. It's very much on the far end of the "high touch" spectrum of encoders. You past examples have all been with the command-line interface, which was deprecated ages ago and which hasn't been used in a single professional title.

Simply BD50 source: H264 encoding at 35 Mbps. IMO real uncompressed source -> encoding at 6.9 Mbps will produce equivalent result to H264 35 Mbps source -> reencoding at 6.9 Mbps simply because there are a really high quality difference between sources and encodings. And source is the same for all the codec.
That's a hypothesis you'll need to test and demonstrate before anyone's going to accept. Take the same uncompressed HD source, encode it to high bitrate H.264, and then to the final, versus going straight from the HD to the final.

benwaggoner

06-30-08, 12:11 PM

Yes, Yes, Yes, and Yes. I and others who've spent a lot of time in this space can see D5 compression artifacts at 1+ Gbit even before they go through any of the other codecs.
We've seen a very sudden shift in Hollywood post to HDCAM-SR for just this reason. I'd say SR has gone from 10% to 90% of the A-list titles in the last couple of years.

Zodiaque

06-30-08, 12:41 PM

"Have" or "purchased"? PSE is absolutely not a tool that someone who hasn't been trained in can get optimum results out of. It's very much on the far end of the "high touch" spectrum of encoders. You past examples have all been with the command-line interface, which was deprecated ages ago and which hasn't been used in a single professional title.

Well I know very well your soft. Now I use directly xml profil without Viewer. Anyway If you think that professional authoring companie use the complete possibility of PSE/PEP then you make a big error. In most case they make 2 pass encoding with default profil and without segment reencoding ... simply because they have no time or human ressource for make that.

benwaggoner

06-30-08, 12:50 PM

Well I know very well your soft. Now I use directly xml profil without Viewer. Anyway If you think that professional authoring companie use the complete possibility of PSE/PEP then you make a big error. In most case they make 2 pass encoding with default profil and without segment reencoding ... simply because they have no time or human ressource for make that.
I'm pretty confident that I've got better data on real-world PEP use than you do :). For the Hollywood titles we're talking about on this forum,

ALL are done using the Viewer interface, not manually editing XML. It'd be nearly impossible to make good segment reencoding decisions that way
I don't know of a single Hollywood title that shipped with just 2-pass encoding.Your desire to make your point has rather outstripped reality.

Now, maybe there are some long-tail European titles doing VC-1 in that mode, but I don't know why they'd be using PEP instead of something much simpler like Rhozet's Carbon Coder. It certainly don't represent how mainstream feature film discs are produced.

amirm

06-30-08, 01:08 PM

Well in this case the DVD forum is completely ridiculous test.
Well, please tell us how you think the test was done and what about it was "ridiculous" compared to your testing. Did you answer if you used PSNR and not a human in your evaluations?

Simply BD50 source: H264 encoding at 35 Mbps. IMO real uncompressed source -> encoding at 6.9 Mbps will produce equivalent result to H264 35 Mbps source -> reencoding at 6.9 Mbps simply because there are a really high quality difference between sources and encodings. And source is the same for all the codec.
Seemed like you missed the point then. Let's say there are two encoders. One softens the video and the other does not at 6.9 mbit/sec. If there are blocking artifacts or ringing in the compressed source, then the encoder which softens those artifacts, will be deemed to be more pleasant and garner higher PSNR results. The sharper encoder would preserve more of those artifacts which could be mistaken for it creating them. In addition, those artifacts have high frequency information in them which is harder to encode for the latter system. The other, gets rid of that detail so it has less work to do.

Note that source artifacts do not need to be visible to you to cause the above scenario. The codec sees those hard edges even if slight whether you do or not.

There is also the issue that some codecs cascade better than others. Depending on how the block sizes align or don't, you get artifacts which would not occur if you started with uncompressed material. For this reason, while D-5 artifacts can reduce fidelity of the encoded output, it tends to not be as much of an issue as starting with GOP-based encodings.

Finally, I can imagine that if you used an encoder which softens the image, whether you started with compressed or uncompressed, the results would be the same. And to this end, let me repeat that most people will pick an encoder with softer picture over one that is more detailed but preserves more of the source issues. And PSNR would be higher too for the former.

Golgot13

06-30-08, 04:53 PM

I'm pretty confident that I've got better data on real-world PEP use than you do :).

I'm not agree Ben. Today we don't have time and the price (in Europe) is down
(BD start at 5000$ in HDMV...). So lot of authoring companies don't lost time
with encoding process, they use default setting only !!!
eg. with big company: Sony DADC give the best price for authoring/encoding
but the quality is not present if the video source master is not enough for the encoding process...
I hope Sony DADC Europe will buy soon the last Sony encoder, MX1000, or the Nexcode solution
(or change to x264 ;) ) because BAE-V700 require a perfect source...

To finish with Ben and Amirm:
You don't want to prouve that VC1 from PeP/CinePSE is real the best solution ?
We, Zodiaque and me, receive a HD movie source with grain and we hope to see
your contribution on Codec Challenge in Doom9 forum.

benwaggoner

06-30-08, 06:18 PM

I'm not agree Ben. Today we don't have time and the price (in Europe) is down
(BD start at 5000K$ in HDMV...). So lot of authoring companies don't lost time
with encoding process, they use default setting only !!!
eg. with big company: Sony DADC give the best price for authoring/encoding
but the quality is not present if the video source master is not enough for the encoding process...
I hope Sony DADC Europe will buy soon the last Sony encoder, MX1000, or the Nexcode solution
(or change to x264 ;) ) because BAE-V700 require a perfect source...
Well, then you're judging VC-1 quality using workflow which the product you're using wasn't defined for. If you're making feature requests for a more automatic implementation, go ahead. But don't conflate your own frustrations with what viewers of Hollywood titles (the primary topic here at AVS) are actually seeing on professioanally encoded discs.

To finish with Ben and Amirm:
You don't want to prouve that VC1 from PeP/CinePSE is real the best solution ?
We, Zodiaque and me, receive a HD movie source with grain and we hope to see your contribution on Codec Challenge in Doom9 forum.
[/QUOTE]
And just like every time you say that, I'll ask you to actually provdie some feedback on the source clip in question. If you're at a loss, how's this for an easy first question:

http://forum.doom9.org/showpost.php?p=1134537&postcount=36

I'm not going to bother to make a clip for the sole purpose of counting coup on AVS; if I'm going to make the effort, I want to do something of broader use to the community. In particular, I'm not interested in doing a bunch of work, get results that are positive for VC-1, and then hearing a bunch of post-hoc objections as to why the clip wasn't a "good test." So if you really want to see something, your best way to get it is to engage in the process in a supportive manner.

Andrew_HD

06-30-08, 06:36 PM

I'm pretty confident that I've got better data on real-world PEP use than you do :). For the Hollywood titles we're talking about on this forum,

ALL are done using the Viewer interface, not manually editing XML. It'd be nearly impossible to make good segment reencoding decisions that way
I don't know of a single Hollywood title that shipped with just 2-pass encoding.Your desire to make your point has rather outstripped reality.

Now, maybe there are some long-tail European titles doing VC-1 in that mode, but I don't know why they'd be using PEP instead of something much simpler like Rhozet's Carbon Coder. It certainly don't represent how mainstream feature film discs are produced.

Ben, I'm not impressed with Hollywood A titles at all. I have to deal with old HD digital recordings and you know the best that good source is the most important thing for good encoding. There are nice encoding also and the most interesting they don't use DNR.

Hollywood has the best sources (or maybe not), but for some reason DNR is used on most of them- I don't understand for what reason ????

I've done one title with DNR (really noisy source- encoder could handle it properly even at 30Mbit)- spent 2 days playing with new Niagra to find the best settings, to compromise natural look with macroblocking free encoding. It was very tricky and I'm still not fully happy for the end result.

Some A titles (I'me based on screenshots from this forum) overuse DNR and look "ugly", others use it less, but still you can see it. In most of the cases, based on screnshots, sources seam to look fine and I don't see the reason to use DNR or at least not so strong.
I can suspect only one answer for this- strong DNR helps for encoding a lot, but it doesn't look natural! I would rather prefer few macroblockings than the "Hollywood-plastic" look.

Bottom line: Snell&Wilcox and other companies who makes DNR hardware don't have to worry about clients :)

Andrew

amirm

06-30-08, 08:55 PM

To finish with Ben and Amirm:
You don't want to prouve that VC1 from PeP/CinePSE is real the best solution ?
Do people eat rice in China? Really, you are three years late with such a challenge. VC-1 is a proven technology now and was the highest used codec for BD+HD DVD combined when I last saw the stats (late last year). As such, it needs no vote of confidence from doom9.

Benchmarks are most valuable when access to the technology is difficult. For example, you may want to see CPU benchmarks when a chip is pre-announced. But once the system is in the market and broadly available, people are free to use it with their own applications, with their own parameters, and in this case, viewers. They have no need for other people's criteria of what makes the best codec. To wit, I can't imagine a single BD content owner interested in how a codec does at 7 mbit/sec, yet that seems to have been a parameter you picked. I can understand that being a good rate to rip HD titles into but it is not a target parameter for the disc.

We, Zodiaque and me, receive a HD movie source with grain and we hope to see your contribution on Codec Challenge in Doom9 forum.
You will see no contributions from me. I have explained time and time again that you have created no motiavation for anyone to be invovled. Further, the continued tone of your posts is such a big turn off that even if there were some value, I would run far, far away from it.

Golgot13

07-01-08, 07:00 AM

Do people eat rice in China?

And, I love your reference when you don't want to participate.

Really, you are three years late with such a challenge. VC-1 is a proven technology now and was the highest used codec for BD+HD DVD combined when I last saw the stats (late last year).

Yes when you give tool for free, it's easy to see many product with your technology.

Benchmarks are most valuable when access to the technology is difficult.

No, when a benchmark give nice result for your codec you use it
(on MS website about VC1, we saw a reference at PSNR before update last year)

For example, you may want to see CPU benchmarks when a chip is pre-announced. But once the system is in the market and broadly available, people are free to use it with their own applications, with their own parameters, and in this case, viewers.

Professional want only a codec/tool which give the best quality with a resonable time.
H264 is a good solution (the best with some tool/software)

To wit, I can't imagine a single BD content owner interested in how a codec does at 7 mbit/sec, yet that seems to have been a parameter you picked. I can understand that being a good rate to rip HD titles into but it is not a target parameter for the disc.

Not agree, I think you don't know all VC1 HDDVD product in Europe,
because there is a title on HDDVD30 with many hour of HD video footage
and the VC1 average bitrate is low than 8Mbps :rolleyes: :rolleyes: :rolleyes:
To my mind, 7Mbps to 9Mbps will be used for HD footage and more than 12Mbps for movie stream.
But I think we will see more title with many lossless multichannel audio track, 2 or 3,
(or uncompressed audio track) so the average bitrate will low, less than 12Mbps
at this bitrate H264 kill VC1.

MovieSwede

07-01-08, 07:10 AM

Yes when you give tool for free, it's easy to see many product with your technology.

Like x264?

scaesare

07-01-08, 09:48 AM

And, I love your reference when you don't want to participate.

And I love how you ignore posts that call out your potential double standard. I ask again: On what basis did you dismiss DVD forum tests as "completely ridiculous"? (see POST 589 (http://www.avsforum.com/avs-vb/showpost.php?p=14189713&postcount=589))

Yes when you give tool for free, it's easy to see many product with your technology.

Aren't there free 264 products?

Not agree, I think you don't know all VC1 HDDVD product in Europe,
because there is a title on HDDVD30 with many hour of HD video footage
and the VC1 average bitrate is low than 8Mbps :rolleyes: :rolleyes: :rolleyes:
To my mind, 7Mbps to 9Mbps will be used for HD footage and more than 12Mbps for movie stream.
But I think we will see more title with many lossless multichannel audio track, 2 or 3,
(or uncompressed audio track) so the average bitrate will low, less than 12Mbps
at this bitrate H264 kill VC1.

Please. Just because one European release opts to put 12 hours of programming on a single disc, don't pretend like that's representative of mainstream authoring workflows.

Please see the disc specifications threads that user "benes" started in the appropriate forum sections here. Your bitrates here don't even represent a single percentage of what we are seeing on discs.

Golgot13

07-01-08, 12:26 PM

Like x264?

x264 is a encoder not a tool like PeP/CinePSE or CineVision.
x264 in 2006 was not compatible with HDDVD and BD specification.

And I love how you ignore posts that call out your potential double standard. I ask again: On what basis did you dismiss DVD forum tests as "completely ridiculous"? (see POST 589 (http://www.avsforum.com/avs-vb/showpost.php?p=14189713&postcount=589))

I know this because it was written on MS webpage about VC1.
But like Zodiaque we are not agree with some confidential codec's test
without best company, leader, of all codec...
About DVD Forum, do you know how much a company much to pay
to be on this association and be prevent about all test/meeting/seminar ?

Aren't there free 264 products?

Professional software: Not there is not...
Studio tool: No
Encoder: No (if a studio use x264 they must to pay a encoder royalty at "MPEG LA"),
so x264 is not free for a professional/commercial use :rolleyes:

Please. Just because one European release opts to put 12 hours of programming on a single disc, don't pretend like that's representative of mainstream authoring workflows.

Please see the disc specifications threads that user "benes" started in the appropriate forum sections here. Your bitrates here don't even represent a single percentage of what we are seeing on discs.

I didn't make this movie title, and I'm not sure you know the title
(12hour of HD video on HDDVD30 in VC1 it'll be horrible and less than 5.7Mbps for audio and video).
So the VC1 HDDVD title was a blockbuster and not a simple little movie.

To my mind, it's a example needed of futur BD title (lot of footage and bonus)
or BD title with many lossless multichannel audio track or many uncompress multichannel audio track.

bourke

07-01-08, 02:30 PM

Professional software: Not there is not...
Studio tool: No
Encoder: No (if a studio use x264 they must to pay a encoder royalty at "MPEG LA"),
so x264 is not free for a professional/commercial use :rolleyes:

Yes, we've gone over this many times before on these forums:

1. Everyone that uses VC-1 or h.264 implementations for commercial purposes must pay royalties to the MPEG Licencing Authority (about USD$0.02 per disc produced or something like that). Exactly the same as for MPEG-2 (about USD$0.02 cents per disc as well). This money payed to the MPEG L.A. is proportionately distributed between the pool of patent-holders.

2. Microsoft is a minor patent-holder for both VC-1 and h.264. Fact.

So no matter what encoder anyone uses for either of these advanced MPEG codec standards (VC-1 or h.264) - money still flows to Microsoft.

Deal with it.

Furthermore, Microsoft already provide (and were the first to provide!) a free encoding tool in July 2006 (Windows Media Encode Studio Edition, now Windows Media Encoder 9). That's right, before x264 was even compliant (or any good)!

Tom McMahon

07-01-08, 02:35 PM

Yes, we've gone over this many times before on these forums:

1. Everyone that uses VC-1 or h.264 implementations for commercial purposes must pay royalties to the MPEG Licencing Authority (about USD$0.02 per disc produced or something like that). Exactly the same as for MPEG-2 (about USD$0.02 cents per disc as well). This money payed to the MPEG L.A. is proportionately distributed between the pool of patent-holders.

2. Microsoft is a minor patent-holder for both VC-1 and h.264. Fact.

So no matter what encoder anyone uses for either of these advanced MPEG implementations (VC-1 or h.264) - money still flows to Microsoft.

Deal with it.

Thanks. I just can't imagine how people on this list ignore these facts and also the fact that they're comparing ENcoders, not codecs. Amazing to me. I hate to say ignorant, but they're ignorant and intentionally seem to remain so.

bourke

07-01-08, 03:01 PM

Well to be fair to people here, from what's been said in this thread most of the FUD seems to be coming from doom9...

Which is strange, because doom9 has a lot of very intelligent members.

I suppose it shows there is a vast difference between knowledge of programming and knowledge of codec standards. The two don't appear to go hand-in-hand.

Golgot13

07-01-08, 03:31 PM

1. Everyone that uses VC-1 or h.264 implementations for commercial purposes must pay royalties to the MPEG Licencing Authority (about USD$0.02 per disc produced or something like that). Exactly the same as for MPEG-2 (about USD$0.02 cents per disc as well). This money payed to the MPEG L.A. is proportionately distributed between the pool of patent-holders.

Encoder royalty is more expensive than decoder royalty.

About VC1 patent, strange it's a copy of H264 patent
and it was available after it :rolleyes::rolleyes:

And some company want to impose VC1 to touch lot of royalty.
eg: DVD royalties are many many hundred millions of dollars.

bourke

07-01-08, 05:39 PM

Encoder royalty is more expensive than decoder royalty.

Yes, but there are free encoders for both VC-1 and h.264.

And likewise, the more advanced encoding tools cost money for both VC-1 and h.264.

About VC1 patent, strange it's a copy of H264 patent
and it was available after it :rolleyes::rolleyes:

Perhaps that's because they improved upon it? I know that's what I would do!

benwaggoner

07-01-08, 06:07 PM

Encoder royalty is more expensive than decoder royalty.
Citation?

http://www.mpegla.com/avc/AVC_TermsSummary.pdf
http://www.mpegla.com/vc1/vc1_TermsSummary.pdf

About VC1 patent, strange it's a copy of H264 patent
and it was available after it :rolleyes::rolleyes:
Citation?

And some company want to impose VC1 to touch lot of royalty.
eg: DVD royalties are many many hundred millions of dollars.
Citation?

Listen, if you're so convinced about H.264 superiority, why don't you do an interesting test at a relevant scenario with the 1080i clip I posted a while ago. Same source. What's a good interlaced BD scenario? There's a ton of good interlaced concert videos out there in 1080i like the award-winning Dave Matthews concert. 20 Mbps average 40 Mbps peak (BD spec, of course)? What's interesting?

Doing and sharing some real encodes would give you a chance to improve your assertion/fact ratio.

amirm

07-01-08, 09:42 PM

About VC1 patent, strange it's a copy of H264 patent and it was available after it :rolleyes::rolleyes:
That is like asking why two Ford cars have similar radios!

Microsoft was a patent holder in H.264 and was a major force for why we went from $2.50 for MPEG-2 to a few cent royalty for AVC, and then VC-1. Without Microsoft's contributions to help create a more balanced program for all concerned, each BD player would have cost $7.50 for the three video codecs alone! Everyone, send me the donations on your way out the store :). And you Gogot, you own me twice as much for not doing an ounce of research before making such serious accusations :D.

scaesare

07-01-08, 09:49 PM

I know this because it was written on MS webpage about VC1.
But like Zodiaque we are not agree with some confidential codec's test
without best company, leader, of all codec...
About DVD Forum, do you know how much a company much to pay
to be on this association and be prevent about all test/meeting/seminar ?

My question was regarding what basis you had for dismissing the DVD forum tests as "ridiculous".

Your answer is that it was written on an MS webpage?

Or you are dismissing the test because it was not on a public forum?

Or because you have some specific knowledge of a company paying to be a DVD forum member and thereby "preventing" something?

Which is it? Or what other specific information do you have?

Or do you not have any direct knowledge, and are simply dismissing the tests out of supposition? To do so while at the same time insisting that tests on some Internet forum with rather arbitrary test criteria is a little disingenuous, no?

Professional software: Not there is not...
Studio tool: No
Encoder: No (if a studio use x264 they must to pay a encoder royalty at "MPEG LA"),
so x264 is not free for a professional/commercial use :rolleyes:

Therefore you assume they were not tested?

I didn't make this movie title, and I'm not sure you know the title
(12hour of HD video on HDDVD30 in VC1 it'll be horrible and less than 5.7Mbps for audio and video).
So the VC1 HDDVD title was a blockbuster and not a simple little movie.

To my mind, it's a example needed of futur BD title (lot of footage and bonus)
or BD title with many lossless multichannel audio track or many uncompress multichannel audio track.

I don't know the title. Nor does that change the fact that almost nobody in their right mind is authoring 12 hours of content to fit on a single disc. Just because one such project exists hardly makes it the proper criteria to judge the merits of an encoder implementation when the vast majority of titles will use a bitrate 6X that.

See:

Blu Ray Bitrate Thread (http://www.avsforum.com/avs-vb/showthread.php?t=760714)

and

HD DVD Bitrate Thread (http://www.avsforum.com/avs-vb/showthread.php?t=822245)

Your allegations have thus far not demonstrated much merit and your expectations for this doom9 test border on the absurd.

bourke

07-01-08, 11:01 PM

For those trumpeting the 'free codec' spiel, there is only one place for you to go - the Xvid forums.

A pity Xvid doesn't win in any recent performance reviews against h.264 or VC-1 codecs :-)

benwaggoner

07-02-08, 12:07 AM

For those trumpeting the 'free codec' spiel, there is only one place for you to go - the Xvid forums.

A pity Xvid doesn't win in any recent performance reviews against h.264 or VC-1 codecs :-)
Actually, I believe MPEG-4 part 2 still is patent encumbered.

I think you need to go back to H.263 or MPEG-1 to get a truly "free to implement" codec. And there's also Ogg Theora, of course, although I don't know how fully vetted against patents it has been.

The industry has amply demonstrated that the advantages of newer codecs outweigh their licensing costs.

Golgot13

07-02-08, 06:25 AM

About VC1 patent, strange it's a copy of H264 patent
and it was available after it
Citation?

From your link about AVC:

For (a) (1) branded encoder and decoder products sold both to end users and on an OEM
basis for incorporation into personal computers but not part of an operating system (a
decoder, encoder, or product consisting of one decoder and one encoder = “unit”),
royalties (beginning January 1, 2005) per legal entity are 0 - 100,000 units per year = no
royalty (this threshold is available to one legal entity in an affiliated group); US $0.20 per
unit after first 100,000 units each year; above 5 million units per year, royalty = US $0.10
per unit.

From your link about VC1:

For (a) (1) branded encoder and decoder products sold both to end users and on an OEM
basis for incorporation into personal computers but not part of an operating system (a
decoder, encoder, or product consisting of one decoder and one encoder = “unit”), royalties
(beginning January 1, 2006) per legal entity are 0 - 100,000 units per year = no royalty (this
threshold is available to one legal entity in an affiliated group); US $0.20 per unit after first
100,000 units each year; above 5 million units per year, royalty = US $0.10 per unit.

It's exactly same price and quantities (for MPEG2 it's completly different) and the number of
companies which have some patent part of each format are different.
There's only the date which changed

Listen, if you're so convinced about H.264 superiority, why don't you do an interesting test at a relevant scenario with the 1080i clip I posted a while ago. Same source. What's a good interlaced BD scenario? There's a ton of good interlaced concert videos out there in 1080i like the award-winning Dave Matthews concert. 20 Mbps average 40 Mbps peak (BD spec, of course)?

Yes, but first one all movie is in 24fps so 1080p test is more adapted .
Second one, 20Mbps is a high bitrate :) and in futur we will see more movie with a bitrate between 12Mbps-18Mbps
(there are some project with lot of HD footage and PiP video...).
Last if I make the VC1 encoding process many people (with you) will say that I'm used the best preset
(like on first round of Codec Challenge...).

Your allegations have thus far not demonstrated much merit and your expectations for this doom9 test border on the absurd.

I don't see a HDDVD title movie with a "fire biker" from european market.

bourke

07-02-08, 07:32 AM

From your link about AVC:

For (a) (1) branded encoder and decoder products sold both to end users and on an OEM
basis for incorporation into personal computers but not part of an operating system (a
decoder, encoder, or product consisting of one decoder and one encoder = “unit”),
royalties (beginning January 1, 2005) per legal entity are 0 - 100,000 units per year = no
royalty (this threshold is available to one legal entity in an affiliated group); US $0.20 per
unit after first 100,000 units each year; above 5 million units per year, royalty = US $0.10
per unit.

From your link about VC1:

For (a) (1) branded encoder and decoder products sold both to end users and on an OEM
basis for incorporation into personal computers but not part of an operating system (a
decoder, encoder, or product consisting of one decoder and one encoder = “unit”), royalties
(beginning January 1, 2006) per legal entity are 0 - 100,000 units per year = no royalty (this
threshold is available to one legal entity in an affiliated group); US $0.20 per unit after first
100,000 units each year; above 5 million units per year, royalty = US $0.10 per unit.

It's exactly same price and quantities (for MPEG2 it's completly different) and the number of
companies which have some patent part of each format are different.
There's only the date which changed

That's just the license pricing?! (Which mind you, I had already pointed out was the same!)

You claimed that they copied the actual patent itself!

signal2noise

07-02-08, 08:43 AM

About VC1 patent, strange it's a copy of H264 patent and it was available after it :rolleyes::rolleyes:
Which patent(s) specifically? For example, just by itself the MPEG LA pool (for each codec) includes dozens of essential patents from a great many patent holders.

VC-1: http://www.mpegla.com/vc1/vc1-patentlist.cfm

AVC: http://www.mpegla.com/avc/avc-patentlist.cfm

Golgot13

07-02-08, 09:40 AM

That's just the license pricing?! (Which mind you, I had already pointed out was the same!)

But for all other codecs (like MPEG2), formats (like ATSC or DVB) the price is different.
It's rare when there is exactly same price.

Which patent(s) specifically? For example, just by itself the MPEG LA pool (for each codec) includes dozens of essential patents from a great many patent holders.

VC-1: http://www.mpegla.com/vc1/vc1-patentlist.cfm

AVC: http://www.mpegla.com/avc/avc-patentlist.cfm

I said only it's same price of license but the number of patent and company which have this patent
are different: So to my mind, it's a copy ....

scaesare

07-02-08, 09:55 AM

{SNIP}

I don't see a HDDVD title movie with a "fire biker" from european market.

I see you continue to ignore requests for justification of your claims dismissing test by a recognized body (the DVD Forum), while at the same time demanding justification from others regarding their participation, or lack thereof, in tests on an internet forum with anonymous participants (doom9).

Your credibility is gone, IMO.

benwaggoner

07-02-08, 09:59 AM

I said only it's same price of license but the number of patent and company which have this patent
are different: So to my mind, it's a copy ....
Do either of us have any idea what point you're trying to make?

A similar group of companies working under the same licensing authority come up with similar license terms for similar technologies in a similar timeframe. I think it's a feature, not a bug that they're the same. It's certianly a feature if you're trying to minimize patent costs; as Amir said, prices get better when you have two technologies to choose between instead of just one.

Wouldn't you complaining about any differences in licenses if they did exist?

Anyway, you've randomized us all enough at this point. If you want to keep talking about this, I request you actually propose a test, make availble some source, offer some feedback on the 24p test clip, or do something other than just making a bunch of erroneous or irrelevant assertions.

If you can't be troubled to actually help on the 24p clip, surely you can do something with the 1080i one. That's being used on far more Blu-ray discs than any of the scenarios you've proposed as been advantageous for H.264.

signal2noise

07-02-08, 10:12 AM

But for all other codecs (like MPEG2), formats (like ATSC or DVB) the price is different.
It's rare when there is exactly same price.

I said only it's same price of license but the number of patent and company which have this patent
are different: So to my mind, it's a copy ....

You’re referring to the patent licensing terms rather than the patents themselves.

I’m not sure that I understand your concern. It seems to me that being commercially competitive dictates similar licensing strategies (there are a few differences between AVC and VC-1). It may also simplify the administration for all involved, etc., etc., etc.

Golgot13

07-02-08, 10:27 AM

If you can't be troubled to actually help on the 24p clip, surely you can do something with the 1080i one. That's being used on far more Blu-ray discs than any of the scenarios you've proposed as been advantageous for H.264.

But Ben, it's hard to have your "movie grain" on 1080i to prouve that VC1 do the best retention.
See your post on Doom9, you explain that on ED there is not movie grain, that VC1
and PeP/CinePSE is the best solution for true movie (with "grain").

And you did a demo at NAB about this, VC1 can preserve more detail like "movie grain" than H264.

Tom McMahon

07-02-08, 10:46 AM

From your link about AVC:

For (a) (1) branded encoder and decoder products sold both to end users and on an OEM
basis for incorporation into personal computers but not part of an operating system (a
decoder, encoder, or product consisting of one decoder and one encoder = “unit”),
royalties (beginning January 1, 2005) per legal entity are 0 - 100,000 units per year = no
royalty (this threshold is available to one legal entity in an affiliated group); US $0.20 per
unit after first 100,000 units each year; above 5 million units per year, royalty = US $0.10
per unit.

From your link about VC1:

For (a) (1) branded encoder and decoder products sold both to end users and on an OEM
basis for incorporation into personal computers but not part of an operating system (a
decoder, encoder, or product consisting of one decoder and one encoder = “unit”), royalties
(beginning January 1, 2006) per legal entity are 0 - 100,000 units per year = no royalty (this
threshold is available to one legal entity in an affiliated group); US $0.20 per unit after first
100,000 units each year; above 5 million units per year, royalty = US $0.10 per unit.

It's exactly same price and quantities (for MPEG2 it's completly different) and the number of
companies which have some patent part of each format are different.
There's only the date which changed

Yes, but first one all movie is in 24fps so 1080p test is more adapted .
Second one, 20Mbps is a high bitrate :) and in futur we will see more movie with a bitrate between 12Mbps-18Mbps
(there are some project with lot of HD footage and PiP video...).
Last if I make the VC1 encoding process many people (with you) will say that I'm used the best preset
(like on first round of Codec Challenge...).

I don't see a HDDVD title movie with a "fire biker" from european market.

Just a side note - realize that not all patent holders have joined, and that in order to ship a given unit you might have to pay not only MPEG-LA but another as well.

benwaggoner

07-02-08, 11:22 AM

But Ben, it's hard to have your "movie grain" on 1080i to prouve that VC1 do the best retention.
See your post on Doom9, you explain that on ED there is not movie grain, that VC1
and PeP/CinePSE is the best solution for true movie (with "grain").

And you did a demo at NAB about this, VC1 can preserve more detail like "movie grain" than H264.
Then provide feedback on the 24p source! Have you been banned from Doom9 or something?

Anyway, if you're so convinced about the superiority of H.264, then you should do even better on an no-grain interlaced test, which doesn't play to VC-1's strengths. Let's pick a realistic scenario, and see what happens.

Really, it's time to actually contribute something other than misplaced assertions and excuses.

Golgot13

07-02-08, 12:38 PM

Then provide feedback on the 24p source! Have you been banned from Doom9 or something?

I'm work too, Ben. And me I didn't dissapear 7month after said I will give a video.
I put the HD video on ftp share at Zodiaque, because like your video on Doom9
it's hard to share a uncompressed video on the web.

Anyway, if you're so convinced about the superiority of H.264, then you should do even better on an no-grain interlaced test, which doesn't play to VC-1's strengths. Let's pick a realistic scenario, and see what happens.

Really, it's time to actually contribute something other than misplaced assertions and excuses.

You're very strange, you talked only about Movie, movie grain, quality look...
Now everything are forgotten...

benwaggoner

07-02-08, 01:53 PM

I'm work too, Ben. And me I didn't dissapear 7month after said I will give a video.
I put the HD video on ftp share at Zodiaque, because like your video on Doom9
it's hard to share a uncompressed video on the web.

You're very strange, you talked only about Movie, movie grain, quality look...
Now everything are forgotten...
Lame.

Propose a test with available content, or provide feedback on my draft content.

You're very free complaining about how I invest my time without showing any willingness to invest your own.

Golgot13

07-02-08, 04:09 PM

Lame.

Propose a test with available content, or provide feedback on my draft content.

You're very free complaining about how I invest my time without showing any willingness to invest your own.

Same about you on "Codec Challenge" thread...
It's hard to download your content. My content is share today
for some specific user and soon at public (2min of video from movie source with grain).

I send you a PM here you to give the link.
I hope to see your encoded file from last VC1 encoder developpment.

Zodiaque

07-03-08, 09:10 AM

My question was regarding what basis you had for dismissing the DVD forum tests as "ridiculous".

Speak about that with codec enginer on doom9 if you want:
- Ateme (H264)
- Mainconcept (H264)
- x264 (H264)

H264 is the most efficient codec (better theorical specification than VC1) but the ratio quality between MPEG2 and H264 is not at 3:1 (like in the DVD forum test). If DVDForum say "VC1 at 7 Mbps produce same quality than MPEG2 at 24 Mbps" then the DVD forum test is completely ridiculous test. It's really simple to prove that with available 1080p source.

MovieSwede

07-03-08, 09:24 AM

If DVDForum say "VC1 at 7 Mbps produce same quality than MPEG2 at 24 Mbps" then the DVD forum test is completely ridiculous test. It's really simple to prove that with available 1080p source.

Its not that simple since the effectivness of codecs depends on the type of the content.

So if something works in scenario A, will maybe not work in scenario B.

So in best case you can show that there a situations were the codec isnt 3X as effective.

RBFilms

07-03-08, 09:25 AM

VC-1 at 7Mbps does not come close to anything else at 24Mbps...we have looked ta low bit rate encodes ... they just do not compare. I would have to agree that is a flawed test and a statement that cannot be trusted.

Speak about that with codec enginer on doom9 if you want:
- Ateme (H264)
- Mainconcept (H264)
- x264 (H264)

H264 is the most efficient codec (better theorical specification than VC1) but the ratio quality between MPEG2 and H264 is not at 3:1 (like in the DVD forum test). If DVDForum say "VC1 at 7 Mbps produce same quality than MPEG2 at 24 Mbps" then the DVD forum test is completely ridiculous test. It's really simple to prove that with available 1080p source.

Zodiaque

07-03-08, 09:29 AM

Its not that simple since the effectivness of codecs depends on the type of the content.

So if something works in scenario A, will maybe not work in scenario B.

So in best case you can show that there a situations were the codec isnt 3X as effective.

Well if your scenario is small 25 frames fade scene then you can produce 1:10 ratio between H264 and MPEG2. But you don't have 1:3 ratio in general case or for complete movie.

scaesare

07-03-08, 09:29 AM

Speak about that with codec enginer on doom9 if you want:
- Ateme (H264)
- Mainconcept (H264)
- x264 (H264)

H264 is the most efficient codec (better theorical specification than VC1) but the ratio quality between MPEG2 and H264 is not at 3:1 (like in the DVD forum test). If DVDForum say "VC1 at 7 Mbps produce same quality than MPEG2 at 24 Mbps" then the DVD forum test is completely ridiculous test. It's really simple to prove that with available 1080p source.

So the DVD forum test is "ridiculous" in your opinion because:

A) You have knowledge of the test methodology and judgement criteria and know they were flawed?

-OR-

B) You don't believe the results?

If the former, then what was the flaw?

If the latter, then the hypocrisy of golgot13's assertions here is self evident.

Incidentally, the short-lived WMV-HD move releases from a few years back had ABR's in the 5-8 range that certainly look far-better then some MPEG2 stuff in the 20's I've seen. Not quite an apples-to-apples comparison, but I don't have a hard time believing that the advanced codecs can be 2-3X as efficient as MPEG2.

RBFilms

07-03-08, 09:31 AM

That is not the case.

Here are the details my engineer gave me for the Chronos encodes:

--------

To answer your question regarding Chronos, here are my findings:

Yes, the source was 1080p. From that source a 1080p encode is created
for Blu-ray and a 1080i variant is created for HD DVD.

Explanations:
HD DVD encodes video in "Program Streams" and uses field syntax with 30
frame per second timing (effectively 60 interlaced fields per second,
a.k.a. 60i) for both 30 fps interlace and 24 fps progressive content.
So for 1080 source content at 24 (or 23.976) progressive frames per
second HD DVD encodes have "repeat field flags" set to create the 60i
Program Stream required for HD DVD muxing. Essentially a 3:2 pulldown
scheme is used effectively creating a 1080i encoded stream for HD DVD.

Some decoders (i.e. set top players) can ignore the "flags" to output
the video back to its original progressive frame rate.

For Blu-ray the video is encoded in "Transport Streams" and allows for
frame syntax video encodes so that 1080p (23.976fps) content can be
compressed and muxed as is (maintaining the native resolution/frame
timing attributes as were in the source).

Side Note: Many Blu-ray Disc players will process a 1080i video, decode
it, and deinterlace it so that the output to the display is 1080p. This
is the case for Natures Journey (which was a 1080i source).

Unfortunately, an analysis of the disc indicates the encoding paramters were incorrect.

scaesare

07-03-08, 09:34 AM

VC-1 at 7Mbps does not come close to anything else at 24Mbps...we have looked ta low bit rate encodes ... they just do not compare. I would have to agree that is a flawed test and a statement that cannot be trusted.

Zodiaque has said "3X". I have typically seen the advanced codecs described as "2-3X" as efficient.

Yet his example that you repeat here is almost 3.5X.

I wouldn't have a hard time believing that a 7Mbps VC-1/AVC encode could compete quite favorably with a 14-21Mb MPEG2 encode.

scaesare

07-03-08, 09:37 AM

That is not the case.

Here are the details my engineer gave me for the Chronos encodes:

--------

To answer your question regarding Chronos, here are my findings:

Yes, the source was 1080p. From that source a 1080p encode is created
for Blu-ray and a 1080i variant is created for HD DVD.

Explanations:
HD DVD encodes video in "Program Streams" and uses field syntax with 30
frame per second timing (effectively 60 interlaced fields per second,
a.k.a. 60i) for both 30 fps interlace and 24 fps progressive content.
So for 1080 source content at 24 (or 23.976) progressive frames per
second HD DVD encodes have "repeat field flags" set to create the 60i
Program Stream required for HD DVD muxing. Essentially a 3:2 pulldown
scheme is used effectively creating a 1080i encoded stream for HD DVD.

Some decoders (i.e. set top players) can ignore the "flags" to output
the video back to its original progressive frame rate.

For Blu-ray the video is encoded in "Transport Streams" and allows for
frame syntax video encodes so that 1080p (23.976fps) content can be
compressed and muxed as is (maintaining the native resolution/frame
timing attributes as were in the source).

Side Note: Many Blu-ray Disc players will process a 1080i video, decode
it, and deinterlace it so that the output to the display is 1080p. This
is the case for Natures Journey (which was a 1080i source).

Have you READ Stacey's comments that I and Amir have quoted back to you several times??

It has nothing to do with repeat flags in the stream for playback.

It has to do with the encoder parameters set to use an incorrect processing mode for the ENCODING.

Go back and look at what he said.

Zodiaque

07-03-08, 09:39 AM

Incidentally, the short-lived WMV-HD move releases from a few years back had ABR's in the 5-8 range that certainly look far-better then some MPEG2 stuff in the 20's I've seen. Not quite an apples-to-apples comparison, but I don't have a hard time believing that the advanced codecs can be 2-3X as efficient as MPEG2.

Well you make comparison between high quality software VC1 VBR encoding and low quality hardware CBR MPEG2 encoding. Perhaps that DVD forum make the test like that ... :eek:

scaesare

07-03-08, 09:41 AM

Well you make comparison between high quality software VC1 VBR encoding and low quality hardware CBR MPEG2 encoding. Perhaps that DVD forum make the test like that ... :eek:

Um... how do you know what MPEG2 I was comparing to?

Is this how you judged the DVD forum tests? By assuming?

trbarry

07-03-08, 09:42 AM

So the DVD forum test is "ridiculous" in your opinion because:

A) You have knowledge of the test methodology and judgement criteria and know they were flawed?

-OR-

B) You don't believe the results?

If the former, then what was the flaw?

If the latter, then the hypocrisy of golgot13's assertions here is self evident.

Incidentally, the short-lived WMV-HD move releases from a few years back had ABR's in the 5-8 range that certainly look far-better then some MPEG2 stuff in the 20's I've seen. Not quite an apples-to-apples comparison, but I don't have a hard time believing that the advanced codecs can be 2-3X as efficient as MPEG2.

Even when you can't do an apple-to-apples comparison because someone else owns the relevant apples you can still do statistical comparisons. A lot of things have been compressed since that DVD forum test was done and there is a general though very sloppy opinion of how good the relative codecs are. And, without claiming proof or objectivity, I completely disbelieve that HD in 7 mbps VC1 is as good as 24 mbps MPEG-2 (on average) for a randomly chosen source.

VC1 is certainly a better codec specification than MPEG-2. I just refuse to believe it is THAT much better. And it's even harder to believe it was better back when the test was conducted. So while I wouldn't brand that test unfairly 'ridiculous' I also don't give it excessive credibility. We've just had too much time since then to see too many counter examples.

- Tom

scaesare

07-03-08, 09:59 AM

Even when you can't do an apple-to-apples comparison because someone else owns the relevant apples you can still do statistical comparisons. A lot of things have been compressed since that DVD forum test was done and there is a general though very sloppy opinion of how good the relative codecs are. And, without claiming proof or objectivity, I completely disbelieve that HD in 7 mbps VC1 is as good as 24 mbps MPEG-2 (on average) for a randomly chosen source.

VC1 is certainly a better codec specification than MPEG-2. I just refuse to believe it is THAT much better. And it's even harder to believe it was better back when the test was conducted. So while I wouldn't brand that test unfairly 'ridiculous' I also don't give it excessive credibility. We've just had too much time since then to see too many counter examples.

- Tom

Please see my subsequent post.... I'm not sure where Zodiaque got the 3X number (and then uses almost 3.5X as his example)...

But ultimately my point is wanting to know what basis Golgot13 has for dimissing a set of structured testing as "ridiculous" while demanding participation in an interweb forum test with anonymous partcipants?

Zodiaque

07-03-08, 10:21 AM

Please see my subsequent post.... I'm not sure where Zodiaque got the 3X number (and then uses almost 3.5X as his example)...

Well 7.7/24 = 0.32 in fact ... but I make simply a little rounding. Anyway 3:1, 3.2:1 or 3.5:1 don't change anything. Speak about that with codec enginer if you want. You can expect 2:1 ratio in best case with the best avalaible implementation for 1080p encoding at high bitrate. And it's particulary true for high complexity source.

scaesare

07-03-08, 10:27 AM

Well 7.7/24 = 0.32 in fact ... but I make simply a little rounding. Anyway 3:1, 3.2:1 or 3.5:1 don't change anything.

OK, seeing as how you are being detail oriented now, the comment was:

What's more, it matched the MPEG-2 reference at 24 mbit/sec on two clips I believe, and nearly matched it on the rest despite running at a third of the data rate.

As has been mentioned, what's in the clip is significant factor. And what is being claimed is that in some cases is does NOT match.

So I ask again: What aspect of the test regimen makes them ridiculaous?

Or is this a blind condemnation based on simply not wanting to believe the conclusions?

Zodiaque

07-03-08, 10:46 AM

So I ask again: What aspect of the test regimen makes them ridiculaous?

Simply the ratio quality. You can't expect that and by far in the general case. It's possible to find example with 1:10 ratio (with fade H264 will use wpred), anyway "I can't claim that H264 is 10X better that MPEG2". You have never 3:1 ratio in average or for complete movie, particulary at high bitrate and for grainy source like in real world BD encoding.

scaesare

07-03-08, 11:17 AM

Simply the ratio quality. You can't expect that and by far in the general case. It's possible to find example with 1:10 ratio (with fade H264 will use wpred), anyway "I can't claim that H264 is 10X better that MPEG2". You have never 3:1 ratio in average or for complete movie, particulary at high bitrate and for grainy source like in real world BD encoding.

Dude. Read what was written:

DVD Forum, fully 3 years before either format came to market, did a double blind shoot out of codecs. MPEG-2 was bid by two companies. AVC I think by 4. And VC-1 of course by Microsoft. Tests were conducted at 7.7 mbit/sec, compared to uncompressed reference and MPEG-2 at 24 mbit/sec. VC-1 easily beat out all other codecs at the same 7.7 mbit/sec. Yes, this included MPEG-2. What's more, it matched the MPEG-2 reference at 24 mbit/sec on two clips I believe, and nearly matched it on the rest despite running at a third of the data rate. VC-1 would have never been accepted into HD DVD and later, BD, without beating MPEG-2. The blind test by the way, was conducted by many members of DVD forum from CE companies to studios.

Warner then, two years later, conducted its own private tests comparing MPEG-2 and AVC to VC-1 with the result being them standardizing on VC-1. They used the data rates appropriate for the format at this time with fresh “codec buster” titles. They made their own evaluation, not impacted by anyone else’s opinion.

For DVD use, VC-1 won at the 7.7 rate. It also managed to comapre well to a couple of MPEG2 clips at 24. Not all. Not even for a single movie. I don't see claims of anybody suggesting that was the intent of that test. Therefore to call it "ridculous" because it doesn't necessarily hold for a different set of criteria is disingenous at best.

You seem to ignore that it was mentioned that a DIFFERENT test was conducted with "data rates appropriate for the format at this time", and VC-1 was chosen. Perhaps not at 3:1 datarates compared to MPEG2, but that was never claimed for tests specifically geared towards HD disc authoring.

Seriously... "common wisdom" has always been that VC-1 and AVC are "2-3X" more efficient than MPEG2. Do you really dispute that?

benwaggoner

07-03-08, 01:24 PM

Same about you on "Codec Challenge" thread...
It's hard to download your content. My content is share today
for some specific user and soon at public (2min of video from movie source with grain).

I send you a PM here you to give the link.
I hope to see your encoded file from last VC1 encoder developpment.
Okay, I've downloaded the source. It'd be interesting to encode, although not a comprehensive test, as it's mainly long shots, and doesn't include credits, motion graphics, or transitions that we'll see a lot of in a real movie. It's also pretty short, which won't really stress rate control. But certainly worth an intial test.

It'd be good to get some kind of explicit license or copyright statement for the clip as well. I'd certainly need that before I could distribute any derivatives from it. If we can get that, I'll put up a simple web version of it so everyone can see what we're talking about.

So, other readers? Want to define a Blu-ray scenario or scenarios that you'd be interested in seeing tested with this clip.

Zodiaque

07-03-08, 04:57 PM

Seriously... "common wisdom" has always been that VC-1 and AVC are "2-3X" more efficient than MPEG2. Do you really dispute that?

It's false too. VC1 is not 2-3X more efficient than MPEG2. Even H264 is not 2-3X more efficient than MPEG2.

For make this comparison Microsoft use objective test (OPSNR) and encoding at 24 Mbps with MPEG2 (with good implementation) produce always by far better result than VC1 at 12 Mbps in all my test.

Well now it's time for simple example ...

benwaggoner

07-03-08, 05:04 PM

It's false too. VC1 is not 2-3X more efficient than MPEG2. Even H264 is not 2-3X more efficient than MPEG2.

For make this comparison Microsoft use objective test (OPSNR) and encoding at 24 Mbps with MPEG2 (with good implementation) produce always by far better result than VC1 at 12 Mbps in all my test.
I think it's really too imprecise a measurement. 2-3X more efficient? Very scenario specific.

I could certainly make a 20 Mbps VC-1 encode that looks just as good as a 60 Mbps MPEG-2 encode, but that's probably not what people meant :). Even a 2x difference at Blu-ray bitrates and within Blu-ray constraints seems a stretch; MPEG-2 isn't that bad a codec by any means.

Now, at 100-500 Kbps at "anything goes" web streaming specs, yeah, I bet VC-1 could spank MPEG-2 at 1/3rd the bitrate, maybe more. MPEG-2 looks good at higher rates, but quality starts falling off a cliff once QP starts climbing up, without any kind of in-loop deblocking to keep from propagating errors forward.

Zodiaque

07-03-08, 05:30 PM

Now, at 100-500 Kbps at "anything goes" web streaming specs, yeah, I bet VC-1 could spank MPEG-2 at 1/3rd the bitrate, maybe more. MPEG-2 looks good at higher rates, but quality starts falling off a cliff once QP starts climbing up, without any kind of in-loop deblocking to keep from propagating errors forward.

Yes but here it's "BD area forum" and it's really simple for me to prove that MPEG2 will be always better than VC1/H264 at 3:1 ratio in this area.

benwaggoner

07-03-08, 05:37 PM

Yes but here it's "BD area forum" and it's really simple for me to prove that MPEG2 will be always better than VC1/H264 at 3:1 ratio in this area.
Didn't I just agree with you :)?

Although you tempt me with the "always" - if sufficiently motivated, I'm I sure I could make a clip where this wasn't so - a nightclub scene with a lot of strobes and shadow detail comes to mind.

amirm

07-03-08, 06:10 PM

I have given multiple opportunity to people to ask what the DVD forum test criteria was, yet it seems they are more interested in assertions without that knowledge. So here are a few tidbits of interest:

1. The tests were double blind. So please don't tell me what you can tell apart without being in the same situation. Look at the surprising results here in double blind audio test that Steve ran and that of Xylon. If I tell you one clip is 7 mbit/sec and the other 24, I have already biased the results. So unless you have participated in blind tests and are then able to do differently, your opinion is not relevant to DVD Forum tests.

2. The original clip was also scored. It did not rank a perfect score! I don't remember the exact number but I think its score was 90% of the quality of the original! Why is this? Because people in a blind test don't want to think they are making a mistake and score down even the original reference as having artifacts.

3. VC-1 I think managed to outscore the original on some clip! Once more, in blind tests there is statistical variance and sometimes, a compressed image may look a bit better than the original (e.g. to the extent it removes noise).

4. Many viewers were used. So the results I mentioned was the average. If the people who are objecting to the results had been participant, it is possible that their score would be different. But then again, they could have done worse than average just the same.

5. As is evident, mathematical measures were not used. Instead, real humans scored things. This is the way to do tests, as opposed to synthetic numbers which folks on Doom9 seem to worship.

As I mentioned, every tests is flawed. But to know how, you need to first read up on what was done and then form an opinion.

trbarry

07-03-08, 07:14 PM

Was the DVD Forum 7.7/24 mbps test just at DVD (720x480) resolution then? That would change my opinion somewhat. I thought it was HD.

- Tom

benwaggoner

07-03-08, 07:32 PM

Was the DVD Forum 7.7/24 mbps test just at DVD (720x480) resolution then? That would change my opinion somewhat. I thought it was HD.

They were definitely HD.

diogen

07-03-08, 07:44 PM

I have given multiple opportunity to people to ask what the DVD forum test criteria was, yet it seems they are more interested in assertions without that knowledge. So here are a few tidbits of interest...Interesting...
Anything else you can disclose?
What were those clips about? CG? Grainy film? Who prepared them?
Is this type of test common for codec comparison, i.e. the original gets rated?
Knowing what you know today, were they representative of hidef movie encoding?
Changes in codecs since then are of evolutionary or revolutionary nature?

Amir, you also mentioned at one point the there was a second shoutout.
What has changed between the two? Did it have the same participants? Same rating system?
How different was the outcome?

Diogen.

amirm

07-03-08, 08:14 PM

Interesting... Anything else you can disclose?
Whatever this aging brain can remember :).

What were those clips about? CG? Grainy film? Who prepared them?
Movies were supplied by the studios. Warner, Sony, etc. They were live action other than Toy Story from what I recall.

Is this type of test common for codec comparison, i.e. the original gets rated?
In double blind tests, there is always a "control" that is rated this way. That is to determine the degree of error. If the original achieves 90% quality, then we know that if another clip achieves 90%, then it is achieving transparency. Or if it is at 85%, then it is almost at the same level.

My wife was a lab technician doing blood work, etc. At the start of the day, they would run sugar water, or something like it through the machine before the real samples. That way, they knew if the machine was working. By the same token, if you find that most of the time someone can't tell the original, you can choose to ignore the results of that person.

Knowing what you know today, were they representative of hidef movie encoding?
It think it was a reasonable test. Yes, the data rate was pretty low for 1080p content (average rate was 7 mbit/sec -- with peak, it was allowed to average 7.7 mbit/sec). But it did show if a codec had a tendency to soften for example when pushed.

Changes in codecs since then are of evolutionary or revolutionary nature?
You mean since the tests were conducted? If so, then they can be considered some of each. The decoders are fixed so nothing architecturally can be changed. On the other hand, I know that we came up with pretty smart enhancements in the VC-1 encoder since then.

Amir, you also mentioned at one point the there was a second shoutout. What has changed between the two? Did it have the same participants? Same rating system? How different was the outcome?

Diogen.
The second round was created because people thought there was some cheating going on. In the first test, you encoded the content, decoded it, and sent them back uncompressed material. So if you wanted to cheat, you could just sent the source back :eek: :D. In the second round, you had to provide the compressed bitstream with instructions/tools on how to regenerated the uncompressed output. They also put a cap on overal bandwidth. That is, the average still had to be 7 mbit/sec but you could not set the peak too high, or keep the codec at high enough bitrate throughout the clip as to cause the overall bitrate to exceed 7.7 mbit/sec.

The participants were the same as was the setup. Rating system changed though. Instead of a single metric of quality, there were others for “noise” and “sharpness.” The latter were added because some codecs were artifact free, but softened the image.

amirm

07-03-08, 08:16 PM

Was the DVD Forum 7.7/24 mbps test just at DVD (720x480) resolution then? That would change my opinion somewhat. I thought it was HD.

- Tom
Nothing remotely close to 7.7 mbit/sec is needed using advanced codecs for SD content. Recall that WMV-HD titles were auto-encoded, using constant bitrates. DVD Forum tests allowed VBR encoding and hand tuning although the latter was not necessarily used.

diogen

07-03-08, 09:23 PM

scaesare

07-03-08, 10:12 PM

It's false too. VC1 is not 2-3X more efficient than MPEG2. Even H264 is not 2-3X more efficient than MPEG2.

For make this comparison Microsoft use objective test (OPSNR) and encoding at 24 Mbps with MPEG2 (with good implementation) produce always by far better result than VC1 at 12 Mbps in all my test.

Well now it's time for simple example ...

Well, as Ben has pointed out, there will obviously be variability. That having been said, DVD encodes ~350K pixels/frame at ~6Mbps ABR.

VC-1/AVC can encode 6X as many pixels (~2mil) at 2-3X the ABR (12-18Mbps). There's the "common wisdom" part of it around here. No doubt it's not linear, and if you crank the bitrate for MPEG2 up high enough there are diminishing returns.

But I certainly bet that a 6Mb advanced codec encode can comnpete quite favorably with MPEG2 at 12, but perhaps not necessarily 18.

But who cares if it's not even 2X? Or 1.5X?

To dismiss out-of-hand the results of a test designed to satisfy one criteria, just because they don't satisfy an entirely DIFFERENT set of criteria is what Golgot13 did, and it simply weakens his credibility.

None of which really adds to the discussion here, I might add.

trbarry

07-03-08, 10:14 PM

Nothing remotely close to 7.7 mbit/sec is needed using advanced codecs for SD content. Recall that WMV-HD titles were auto-encoded, using constant bitrates. DVD Forum tests allowed VBR encoding and hand tuning although the latter was not necessarily used.

Right. But another comment above made me doubt my memory of this, and I wasn't there. If it had really been just DVD resolutions then I could easily see how the bit rate was effectively saturated on MPEG-2 and 7.7 VC1 could match it in a double blind test.

But apparently that isn't what happened. Just checking.

- Tom

diogen

07-03-08, 10:32 PM

To dismiss out-of-hand the results of a test designed to satisfy one criteria, just because they don't satisfy an entirely DIFFERENT set of criteria is what Golgot13 did, and it simply weakens his credibility.After the Elefant Dream encode competition on doom9, the "gurus" there made a few conclusions:
-x264 was the best;
-MPEG-2 was better than expected;
-VC1 can't compete with x264, isn't much better than MPEG-2 and is XviD/DivX (MPEG-4 part 2) level at best.
The PSNR/SSIM numbers supported these conclusions.

What really amazes me when Golgot13 references a paper with ~1Mbps encodes to claim AVC superiority over VC1.
I haven't seen first-time posters do this neither on doom9 nor here. Let alone people claiming to be pros in the field...

Diogen.

signal2noise

07-03-08, 10:37 PM

Interesting...
What were those clips about? CG? Grainy film? Who prepared them?
90 second film clips at 1920 x 1080 taken from:

Harry Potter and the Sorcerer’s Stone (Warner)
Monsters, Inc. (Disney)
Dick Tracy (Disney)
Seven (New Line/Warner)
Titan A.E. (Fox)
Stuart Little 2 (Sony)

Corellianrogue

07-03-08, 11:03 PM

benwaggoner

07-03-08, 11:32 PM

After the Elefant Dream encode competition on doom9, the "gurus" there made a few conclusions:
-x264 was the best;
-MPEG-2 was better than expected;
-VC1 can't compete with x264, isn't much better than MPEG-2 and is XviD/DivX (MPEG-4 part 2) level at best.
The PSNR/SSIM numbers supported these conclusions.
Although I think the VC-1 encodes were perceptually, not PSNR tuned. I think Zambelli later offered up a PSNR encode which was maybe 1.5 better, even though it looked worse.

I probably should revisit that clip with our latest encoder implementations. We've had quite a few rounds of tuning since then.

Still, Elephant's Dream is almost canonically the kind of content that H.264 will have the biggest advantage with; completely grain free, static shots, and with banding that a nice strong loop filter will tend to smooth out :).

For more typical film and video content, VC-1 dramatically outperforms both MPEG-4 pt 2 and MPEG-2. One of the biggest complaints about MPEG-4 pt 2 was how little it actually improved compression efficiency over MPEG-2 in the first place, which is why it wasn't ever adopted for broadcast or HDM.

benwaggoner

07-03-08, 11:34 PM

Isn't x264 an upgraded version of h264? So it's a bit pointless of people to compare it to VC-1 in regards to HD DVD and Blu-Ray since neither can use x264, lol!
x264 is an implementation of the H.264 standard, more accurately. And one of the best H.264 implementations, in my opinion, certainly for the lower bitrates I've tested it for.

That said, I don't know if there's ever been a released version capable of producing Blu-ray complaint streams; I've certainly never heard of a real-world title being encoded with it. The Blu-ray specs, like all optical media specs, can be remarkably finicky if followed to the letter.

amirm

07-03-08, 11:56 PM

Thanks, Amir.

We heard more than once from different parties here on AVS that setting up a codec shoutout is not only hard and expensive, but no matter the outcome, it will never be conclusive. Did this realization come after those two rounds? If not, why was it organized in the first place?
I have been participating in video/audio tests of this sort for more than a decade (going back to when I started at Microsoft 11 years ago). One quickly finds out the these tests are subjective to start. And have so many parameters that are essentially impossible to level out. Let me give an example.

In low bit rates (< 300 kbits/sec) one can optimize the codec for less artifacts and less detail or more artifacts and more detail. And here I am talking about the very same codec. Then you ask people which one is better and some say the first and other the second. Again, this is using the exact same codec. If one cannot get consensus when using the same codec, what hope there is when there are so many parameters involved?

Think of another ramification of above. What if the default parameters are different in above in that way and the person doesn’t know how to tune them or doesn’t want to? A report was posted recently on Jan Ozer’s test. Jan used to insist that every encoder had to be used in their default mode as to make things “fair.” We would try to explain that was not fair until we were blue in the face but he would not accept it. I realize that from his point of view that was the right thing to do, but from the point of the view of the people who build the codecs, not.

Since all stidios are in the same business and are having the same goals in terms of quality of the output, why do they still do internal comparisons and no external are considered useful?
The advantage of internal tests is that they can use their content and their staff who approves quality. So using the above example, to the extent they prefer a softer picture (I am making this up), then the external tests which were tilted toward a preference for sharper pictures would not necessarily be useful to them. Of course, this is a made up example. It is possible that the external test had more rigor and as such, was more correct.

You mentioned Sony as the provider of the clip(s). Wasn't they interested in keeping MPEG-2 only on BD? Or has this changed by then?
Let me start by saying that while I am sure of Warner providing clips, I am less sure of Sony. WRT to your question though, turns out this doesn’t matter. All the companies at that time were active in these tests. For example, the host company which provided the evaluation facility was Panasonic – one of the three core companies in BD. We had just joined DVD Forum so I can’t explain the dynamics leading up to that point accurately or explain why they would have had any interest here.

What studio was the driving force behing BD adopting other codecs?
None assuming I am not confusing my sense of timing here :). At the time more codecs were picked, the only studio supporting BD was Sony/MGM I believe. Warner/Paramount/Uni where on the HD DVD side and Fox/Disney undecided. Sony and Panasonic were very much against advanced codecs and had put together a paper and shootout at >20 mbit/sec showing that AVC was not transparent to the source but MPEG-2 was.

Golgot13

07-04-08, 08:46 AM

Hi,

few minutes of my work time for Ben:

That said, I don't know if there's ever been a released version capable of producing Blu-ray complaint streams; I've certainly never heard of a real-world title being encoded with it. The Blu-ray specs, like all optical media specs, can be remarkably finicky if followed to the letter.

No true, I made many disc with x264 encoder (thanks Zodiaque) in HDDVD and BD (available on market).
No problems on any HDDVD or BD player (for each format), only that some verifier software
don't check the video header(no warning, strange!!!). And some muxer don't support well bad SEI
(thanks at Trahald for his patcher).

Okay, I've downloaded the source. It'd be interesting to encode, although not a comprehensive test, as it's mainly long shots, and doesn't include credits, motion graphics, or transitions that we'll see a lot of in a real movie. It's also pretty short, which won't really stress rate control. But certainly worth an intial test.

Yes but it's a movie source, a nice US guy share me it for test.

It'd be good to get some kind of explicit license or copyright statement for the clip as well. I'd certainly need that before I could distribute any derivatives from it. If we can get that, I'll put up a simple web version of it so everyone can see what we're talking about.

There is only one restriction to use the video, we must mention that "the clip is from SMOTHER by Luis Gispert".

So, other readers? Want to define a Blu-ray scenario or scenarios that you'd be interested in seeing tested with this clip.

We compare the codec quality, so we can use exactly same bitrate than on "Codec Chalenge" thread.
With same HDDVD restriction, because it will compatible with BD specification (some different header data only):
18 Mbps (Max at 28.0 Mbps) for video stream with +/- 0.5 % for bitrate tolerance
12 Mbps (Max at 24.0 Mbps) for video stream with +/- 0.5 % for bitrate tolerance
6 Mbps (Max at 20.0 Mbps) for video stream with +/- 0.5 % for bitrate tolerance

benwaggoner

07-04-08, 10:25 AM

No true, I made many disc with x264 encoder (thanks Zodiaque) in HDDVD and BD (available on market).
Can you share which titles?

No problems on any HDDVD or BD player (for each format), only that some verifier software don't check the video header(no warning, strange!!!). And some muxer don't support well bad SEI (thanks at Trahald for his patcher).
Well, the bar for compliance isn't "works in most players" but "meets the spec exactly." For example, I know lots of players can work with single-slice H.264 in BD, but the spec requires three slices, so that's what a compliant encoder has to do.

In any comparative test, we need to exactly follow the published specifications.

Yes but it's a movie source, a nice US guy share me it for test.

There is only one restriction to use the video, we must mention that "the clip is from SMOTHER by Luis Gispert".
Good, thanks. Any chance we can get some kind of actual documentation on the copyright waiver from him or something?

We compare the codec quality, so we can use exactly same bitrate than on "Codec Chalenge" thread.
With same HDDVD restriction, because it will compatible with BD specification (some different header data only):
18 Mbps (Max at 28.0 Mbps) for video stream with +/- 0.5 % for bitrate tolerance
12 Mbps (Max at 24.0 Mbps) for video stream with +/- 0.5 % for bitrate tolerance
6 Mbps (Max at 20.0 Mbps) for video stream with +/- 0.5 % for bitrate tolerance
Does anyone still care about HD DVD specs anymore? Seems like we should be doing Blu-ray at this point.

Anyone else have any particular tests they'd like to see relevant to the current market?

diogen

07-04-08, 10:45 AM

Although I think the VC-1 encodes were perceptually, not PSNR tuned. I think Zambelli later offered up a PSNR encode which was maybe 1.5 better, even though it looked worse.I think so.
And then everything got buried in the discussion of what the vbv_buffer (?) in the spec under NDA was...
Zambelli in the mean time disappeared...

I remember looking at the VC1 PSNR table. Zambelli's encode score was essentially defined by 2-3 parts of 10 seconds or less, IIRC.
I probably should revisit that clip with our latest encoder implementations. We've had quite a few rounds of tuning since then.I like the quanfiable nature of PSNR/SSIM.
I think there should be an in-the-face example of "perfect PSNR - crappy picture" and "crappy PSNR - good picture" before those measuring sticks become outdated...

Diogen.

benwaggoner

07-04-08, 12:26 PM

I think so.
And then everything got buried in the discussion of what the vbv_buffer (?) in the spec under NDA was...
Zambelli in the mean time disappeared...
Yeah, he switched jobs a while ago, and has been working on some huge projects.

I remember looking at the VC1 PSNR table. Zambelli's encode score was essentially defined by 2-3 parts of 10 seconds or less, IIRC.
I like the quanfiable nature of PSNR/SSIM.
I love the quantifiable nature of PSNR/SSIM as well. However, it's so lovable it makes it tempting to test for what's easy to measure instead of test for what you actually want :).

PSNR is great to have just to validate that codec changes haven't done something pathological. We absolutely have features in our encoder (Differential Quantization and Adaptive Deadzone in particular) that we know help perceptual quality in double-blind measurements, but hurt PSNR. But we try to introduce those features deliberately and carefully.

I think there should be an in-the-face example of "perfect PSNR - crappy picture" and "crappy PSNR - good picture" before those measuring sticks become outdated...
Noisy content would be a classic example. An encoder could have a very different per-pixel grain distribution while preserving the same grain texture, and wind up with a lousy PSNR since the pixels would be so different. Has anyone ever run a PSNR on VP6/VP7? They have a noise synthesis postprocessing filter that should show this behavior.

For more typical content, I don't know how often you'd see as big a night and day difference as you're looking for. Perceptual quality certainly correlates with both PSNR and SSIM, it just isn't a 1.0 correlation. It'd say the same overall quality could vary in PSNR by 2-3 dB at least, probably more in edge cases.

SSIM is better than PSNR overall (since it weights smooth regions over noisy regions), but is still tuned for a CRT-style perceptually uniform gamma, and so doesn't weight problems in the low-luma range which are a big problem on modern LCD and plasma displays.

trbarry

07-04-08, 01:34 PM

...
At the time more codecs were picked, the only studio supporting BD was Sony/MGM I believe. Warner/Paramount/Uni where on the HD DVD side and Fox/Disney undecided. Sony and Panasonic were very much against advanced codecs and had put together a paper and shootout at >20 mbit/sec showing that AVC was not transparent to the source but MPEG-2 was.

Yes, and in my opionion it was the competing force of Microsoft/VC-1/HD DVD that managed to drive Sony and BD from MPEG-2 to AVC. For this we are all eternally grateful. ;-)

If they eventually instead end up with VC-1 that would be perfectly fine also. But sticking with MPEG-2 would have been horrid.

- Tom

trbarry

07-04-08, 01:43 PM

I think so.
And then everything got buried in the discussion of what the vbv_buffer (?) in the spec under NDA was...
Zambelli in the mean time disappeared...

I remember looking at the VC1 PSNR table. Zambelli's encode score was essentially defined by 2-3 parts of 10 seconds or less, IIRC.
I like the quanfiable nature of PSNR/SSIM.
I think there should be an in-the-face example of "perfect PSNR - crappy picture" and "crappy PSNR - good picture" before those measuring sticks become outdated...

Diogen.

Whether to trust either metric vs visual inspection is one of those religious issues we will never get everybody to agree on. So, whether we have consensus or not, there will effectively be two prizes, one for best metrics and one for best subjective video quality. I like to look at both.

In an ideal world some entry would win both both categories but be we can't count on that and it has been pointed out it is possible to optimize most codecs for either. So we might as well accept that.

- Tom

diogen

07-04-08, 11:09 PM

Noisy content would be a classic example. An encoder could have a very different per-pixel grain distribution while preserving the same grain texture, and wind up with a lousy PSNR since the pixels would be so different.
Whether to trust either metric vs visual inspection is one of those religious issues we will never get everybody to agree on.Sounds good.
A grainy enough source would be able to (unlike ED) show a difference in PSNR/SSIM when going up in bitrate above 12Mbps.
And maybe even offer a chance to show why PSNR isn't everything.

Diogen.

Corellianrogue

07-05-08, 10:05 AM

No true, I made many disc with x264 encoder (thanks Zodiaque) in HDDVD and BD (available on market).
No problems on any HDDVD or BD player (for each format), only that some verifier software
don't check the video header(no warning, strange!!!). And some muxer don't support well bad SEI
(thanks at Trahald for his patcher).

x264 is compatible with HD DVD & Blu-Ray?! :eek: I'm sure the studios don't want people to know that, lol! Then again I'm guessing that they can't play MKV files (which most x264 files seem to be stored in) though? (I know the Xbox360 can't yet.)

Zodiaque

07-06-08, 02:59 AM

x264 is compatible with HD DVD & Blu-Ray?! :eek: I'm sure the studios don't want people to know that, lol! Then again I'm guessing that they can't play MKV files (which most x264 files seem to be stored in) though? (I know the Xbox360 can't yet.)

X264 is a H264 codec. mkv is a container. You must use BD authoring software with x264.

RBFilms

07-06-08, 06:21 AM

I am not clear on the question then.

Can you repost or send me in a PM?

Which title are we talking about? Nature's Journey?

Have you READ Stacey's comments that I and Amir have quoted back to you several times??

It has nothing to do with repeat flags in the stream for playback.

It has to do with the encoder parameters set to use an incorrect processing mode for the ENCODING.

Go back and look at what he said.

RBFilms

07-06-08, 06:29 AM

scaesare

07-06-08, 10:00 AM

Are you talking about CBR verses VBR?

The claim is incorrect if that is the case. I read that comment and posted my reply to the CBR verses VBR.

There is NO gain and NO advantage to VBR based on how we handled the encode ... we maxxed out the encode (26Mbps for HD and 37Mbps for BD) at the highest possible bit rate and there was nowhere that VBR could have gone beyond that for the HD-DVD version. The BD still had headroom ... maybe around 4Mbps to 5Mbps we could have played with. At 37Mbps ... using VC-1 .... we would not have gained much more by pushing it any further .... most people claim they can see little difference between the 26Mbps and 37Mbps encodes ... albeit, others have said there is a difference.

This may not be the most efficient way if you are looking to save space ... that would be your only valid argument.

However, there is no argument as to quality of the encode as there was no more headroom for VBR ... we already gave the encode MAXIMUM AVAILABLE bit rate.

I have confirmed this with numerous encoding engineers to assure that there would have been no advantage to a VBR encode. This would have only been the case if we did a lower bit rate encode and had headroom for peaks. There was no headroom left at the rate we encoded ... and we ran the entire program at maximum bit rate.

If it was a different question or comment ... then please clarify.

It was not only the CBR-like nature of the encode. It was the parameter fed to the encoder regarding linterlacing:

I finally got around to looking at the bitstream on Nature's Journey and I am puzzled. It seems to be using the interlace frame encoding mode of PEP instead of the interlaced field mode. There are two interlaced modes, field and frame. Frame is used when there is no motion or very slow motion while field is used in normal motion. You can force an encode to be one, the other or both (auto). Normally we recommend the auto mode. A higher bitrate is required if you use the less effecient mode for the motion in the frame.

The encode is virtually CBR, which can be wasteful in the bitrate department. On one hand, the easy scenes make a look a tad better, but the more complex scenes may suffer.
Truth be told, frame is the red headed step child. Our focus is always progressive mode first, followed by field and when time permits, frame.

(Original post HERE (http://www.avsforum.com/avs-vb/showthread.php?p=11979429#post11979429))

The frame-progressive mode appears to have been selected in error here, and it is the least "devleoped" of the modes.

benwaggoner

07-06-08, 11:08 AM

Are you talking about CBR verses VBR?

The claim is incorrect if that is the case. I read that comment and posted my reply to the CBR verses VBR.

There is NO gain and NO advantage to VBR based on how we handled the encode ... we maxxed out the encode (26Mbps for HD and 37Mbps for BD) at the highest possible bit rate and there was nowhere that VBR could have gone beyond that for the HD-DVD version. The BD still had headroom ... maybe around 4Mbps to 5Mbps we could have played with. At 37Mbps ... using VC-1 .... we would not have gained much more by pushing it any further .... most people claim they can see little difference between the 26Mbps and 37Mbps encodes ... albeit, others have said there is a difference.

This may not be the most efficient way if you are looking to save space ... that would be your only valid argument.

However, there is no argument as to quality of the encode as there was no more headroom for VBR ... we already gave the encode MAXIMUM AVAILABLE bit rate.

I have confirmed this with numerous encoding engineers to assure that there would have been no advantage to a VBR encode. This would have only been the case if we did a lower bit rate encode and had headroom for peaks. There was no headroom left at the rate we encoded ... and we ran the entire program at maximum bit rate.
While in theory there's no difference between a VBR encode where ABR=PBR and a CBR encode, in practice there often is. For example, plenty of encoders are 1-pass only for their CBR modes, while offering 2-pass for the VBR modes. 2-pass typically offers more efficient rate control due to lookahead, scene, fade, and flash detection, and other stuff where it has perfect ability to "look into the future."

PEP used to have a CBR mode, but since it was such a rare method to use for HDM, we eventually dropped it since we'd not had any real customer feedback on how well it was working. I don't know of any title that ever shipped using it, particularly since the ability to do segment reencoding flexibly pretty much makes your encode VBR anyway.

Zodiaque

07-06-08, 12:40 PM

Well in fact BD and HDDVD can't really use pure VBR. It's a constrained VBR with max_bitrate and buffer. More than average bitrate is close to max bitrate and less encoding is really VBR.

@ Ben

I will make very little test. I will use VC1 SDK or CinevisionPSE. What is your recommanded setting for really fine grainy source at 12 Mbps. Be carefull average quantizer will be relatively high even at 12 Mbps with this source. This profil is good for that or not? What are your recommended setting?

@REM -----------------------------------------------------------
@REM
@REM Profil BluRay 1080p23.976 2 passes extra high quality
@REM
@REM -----------------------------------------------------------

@REM Source file name (suffit de mettre la source ici)
set E_SRC=hjhjhhjjh.avs

@REM Set of bitrates (ici le bitrate)
set E_BR=12000

@REM Set of max bitrates (ici le bitrate max)
set MAX_BR=40000

@REM Set of Buffer (ici le buffer)
set BUF_BR=3750000

@REM Profil (ici le nom des fichiers de sortie)
AVS2ASF.exe -i %E_SRC% -o azerty.vc1 -rate %E_BR% -peakrate %MAX_BR% -vbv %BUF_BR% -framerate 23.976 -ratecontrol 3 -profiletype 2 -maxkeydist 24 -bframes 1 -bdeltaqp 1 -adaptiveGOP -keyPop 1 -inloop 1 -overlap 1 -complexity 5 -motionsearchlevel 2 -mesearchmethod 1 -mbcost 1 -mvcost 1 -mvrange 4 -adaptivequant 1 -dquantoption 1

pause

benwaggoner

07-06-08, 02:19 PM

Well in fact BD and HDDVD can't really use pure VBR. It's a constrained VBR with max_bitrate and buffer. More than average bitrate is close to max bitrate and less encoding is really VBR.
Sure. NOTHING can really use pure VBR in the abstract, since you can make no guarnatees about max decode complexity.

I will make very little test. I will use VC1 SDK or CinevisionPSE. What is your recommanded setting for really fine grainy source at 12 Mbps. Be carefull average quantizer will be relatively high even at 12 Mbps with this source. This profil is good for that or not? What are your recommended setting?

@REM Source file name (suffit de mettre la source ici)
set E_SRC=hjhjhhjjh.avs

@REM Set of bitrates (ici le bitrate)
set E_BR=12000

@REM Set of max bitrates (ici le bitrate max)
set MAX_BR=40000

@REM Set of Buffer (ici le buffer)
set BUF_BR=3750000

@REM Profil (ici le nom des fichiers de sortie)
AVS2ASF.exe -i %E_SRC% -o azerty.vc1 -rate %E_BR% -peakrate %MAX_BR% -vbv %BUF_BR% -framerate 23.976 -ratecontrol 3 -profiletype 2 -maxkeydist 24 -bframes 1 -bdeltaqp 1 -adaptiveGOP -keyPop 1 -inloop 1 -overlap 1 -complexity 5 -motionsearchlevel 2 -mesearchmethod 1 -mbcost 1 -mvcost 1 -mvrange 4 -adaptivequant 1 -dquantoption 1

pause

That's using the VC-1 Encoder SDK, which doesn't include all the latest PEP tweaks. I'm not sure about the parameter mapping for that utility, so I'll just decribe the settings. For this content at 12 Mbps, I'd probably look at something like:

Turn off bdeltaqp; that'll put it in the automatic mode which is normally more efficient

Turn off keyPOP. It often doesn't help much; only turn it on if you see actual popping

Turn off Overlap unless you're actually seeing block artifacts; it softens the image some.

I'd probably turn off DQuant as well; the VC-1 Encoder SDK version is more tuned to 15+ Mbps bitrates. The PEP version does a much better job at lower bitrate encodes.

Are those Adaptive Motion Match Method and Motion Search Range? Those are the modes you want to use.

If you want to make it crazy slow for probably no perceptible increase in quality, you can aways set numthreads to 1 :). The lower complexity entropy decode in VC-1 allows single-slice encoding on BD. It'd be rare to make any real difference unless you're encoding a pogo-stick championship or something.

benwaggoner

07-06-08, 03:13 PM

MovieSwede

07-06-08, 03:35 PM

Ben is there any free VC1 encoding tools that can produce BD compatible streams?

benwaggoner

07-06-08, 03:45 PM

Ben is there any free VC1 encoding tools that can produce BD compatible streams?
Yes, the AVS2ASF metioned above should be able to do it. It uses the free "Pro" version of the VC-1 Encdoer SDK.

Zodiaque

07-06-08, 04:06 PM

I just ran a quick test using the settings above in Rhozet's Carbon Coder (VC-1 Encoder SDK based app), and they looked pretty awesome.

I'm not sure why you were concerned about high quants. The highest QP macroblock I found in the whole thing was 5. The whole thing looked great I thought. It's really not that challenging a clip, other than the grain texture.

I'm running a test via PSE 2.1 right now, and we'll see how much of a difference that makes.

Well the source will be available. If you want try to obtain better result with your best setting.

Turn off bdeltaqp; that'll put it in the automatic mode which is normally more efficient

Imply generaly higher quant for bframe ... I think

Turn off Overlap unless you're actually seeing block artifacts; it softens the image some.

Yes more sharp, more grain retention ... but really more blocking at high quant level.

not sure why you were concerned about high quants. The highest QP macroblock I found in the whole thing was 5. The whole thing looked great I thought. It's really not that challenging a clip, other than the grain texture

It's a high entropy source. With really fine grain. Encoding at 12 Mbps produce relative high average quant for VC1.

Zodiaque

07-06-08, 04:16 PM

Yes, the AVS2ASF metioned above should be able to do it. It uses the free "Pro" version of the VC-1 Encdoer SDK.

Yes work very well and will produce really good quality for BD25 or BD50 encoding. Moreover AVS2ASF is really simple to use in CLI mode.

Zodiaque

07-06-08, 05:17 PM

Partial result: VC1 vs H264

http://jfl1974.free.fr/Darknight/Darknight.vc1
http://jfl1974.free.fr/Darknight/Darknight.264

I use a special x264 build for grain retention. Like you can see there are no possible comparison here. x264 is the best and by far ... and for grain retention too.

next update:
- lossless source
- MPEG2 encoding at 24 Mbps
- VC1 and H264 encoding at 8 Mbps

30XS955 User

07-06-08, 05:24 PM

^^ Did you post the original clip too?

benwaggoner

07-06-08, 05:54 PM

Imply generaly higher quant for bframe ... I think
Actually not. When you don't specify a BDeltaQP, it'll be dynamic. For most content it'll still be 1, but the encoder can make it 0 or 2 based on the amount of motion.

Yes more sharp, more grain retention ... but really more blocking at high quant level.
Sure, but even the older VC-1 Encoder SDK implementation is only going up to a peak QP of 5 even in B-frames, not really in the high quant range at all for this clip.

It's a high entropy source. With really fine grain. Encoding at 12 Mbps produce relative high average quant for VC1.
What quants are you seeing? I tried to replicate your settings exactly, and I'm getting quite low quants.

benwaggoner

07-06-08, 06:10 PM

benwaggoner

07-06-08, 08:21 PM

Partial result: VC1 vs H264

http://jfl1974.free.fr/Darknight/Darknight.vc1

Looks corrupt. It's crashing PSE Viewer for me when I try to play it, and File Properties has nonsensical data.

I use a special x264 build for grain retention. Like you can see there are no possible comparison here. x264 is the best and by far ... and for grain retention too.

If you're using a custom build, can I as well :)?

30XS955 User

07-06-08, 08:40 PM

Looks corrupt. It's crashing PSE Viewer for me when I try to play it, and File Properties has nonsensical data.

If you're using a custom build, can I as well :)?

I thought builds had to remain within BD spec.

amirm

07-06-08, 09:35 PM

I thought builds had to remain within BD spec.

You can tune an encoder for a specific clip. That would degrade the performance elsewhere but if no one is looking....

benwaggoner

07-06-08, 09:49 PM

I thought builds had to remain within BD spec.
Yes, indeed. All proposed files should pass validation and mux with Blu-print, and follow the BD spec to the letter.

Given the complexity of validation, I worry about a custom build of an encoder in general, since by definition it wouldn't have gotten tons of formal testing.

lgans316

07-06-08, 10:00 PM

- MPEG2 encoding at 24 Mbps
- VC1 and H264 encoding at 8 Mbps

How will these clips fair on 50"+ or PJ screens ?

MovieSwede

07-07-08, 12:15 AM

One question would be what was the source for the encode?

PaulGo

07-07-08, 12:27 AM

I viewed both encodes several times. I view it on a 24" Samsung monitor at full 1080p resolution. To me the AVC (H264) encode does look better. I really don't know how to describe the difference except it just had a bit more "pop". The VC-1 encode just looked a bit flat. I realize this does not settle anything because people have different preferences and by tweaking the encodes different results may be obtained, but based on these samples I definitely prefer AVC.

MovieSwede

07-07-08, 12:42 AM

RobertR1

07-07-08, 12:44 AM

PaulGo

07-07-08, 01:14 AM

What did you use to calibrate your monitor?

I do not have any test equipment, but I do have many video calibration and greyscale images. Any viewing is subjective but I viewed both clips using the same monitor settings. I have a Nvidia 7950GT video card (DVI) and I viewed the clips with PowerDVD8.

benwaggoner

07-07-08, 01:15 AM

But if you encode something from an already compressed source
you can/will give one of the encoders an advantage it maybe wouldnt have
if it were compressed from the orginal uncompressed source.
Yes. For cases where you aren't cropping/scaing, reencoding in the same codec often yields best results when the same codec is used for source and output, since the same kinds of filtering would already have been applied.

benwaggoner

07-07-08, 01:16 AM

I do not have any test equipment, but I do have many video calibration and greyscale images. Any viewing is subjective but I viewed both clips using the same monitor settings. I have a Nvidia 7950GT video card (DVI) and I viewed the clips with PowerDVD8.
I've seen cases where the same system may decode different codecs in different color spaces.

It'd be good if we could have some color bars also calibrated with the same settings to make sure we're getting the same end-to-end luma levels and such.

Zodiaque

07-07-08, 01:21 AM

What's the source?

Darknight trailer ...

What bitrate are you targeting here?

12 Mbps.

What quants are you seeing? I tried to replicate your settings exactly, and I'm getting quite low quants.

Source will be available.

Looks corrupt. It's crashing PSE Viewer for me when I try to play it, and File Properties has nonsensical data.

Encoding with SDK. This encoding work with PSE viewer for me. Mux correctly with Scenarist. Esa don't detect error. No buffer underflow.

If you're using a custom build, can I as well ?

It's special patched build. This x264 build kill all the other codec or H264 implementation for grain retention like you can see ... ;-)

MovieSwede

07-07-08, 01:29 AM

Darknight trailer ...

Was the dark knight trailer in .mov?

Zodiaque

07-07-08, 02:56 AM

Partial result: VC1 vs H264 vs MPEG2

http://jfl1974.free.fr/Darknight/Darknight_x264_12Mbps.264
http://jfl1974.free.fr/Darknight/Darknight_SDK_12Mbps.vc1
http://jfl1974.free.fr/Darknight/Darknight_HCEnc_24Mbps.m2v

- H264 with x264 special build
- VC1 with MS VC1 SDK encoder
- MPEG2 with HCEnc

next update:

- lossless source
- VC1 and H264 encoding at 8 Mbps

Annexe:

@REM -----------------------------------------------------------
@REM
@REM Profil BluRay 1080p23.976 extra high quality
@REM
@REM -----------------------------------------------------------

@REM Source file name (suffit de mettre la source ici)
set E_SRC=hjhjhhjjh.avs

@REM Set of quality (ici la qualité 1-50)
set E_BR=22

@REM Set of max bitrate (ici le bitrate max)
set MAX_BR=40000

@REM Set of Buffer (ici le buffer)
set BUF_BR=30000

@REM Set credit (frame de début du générique)
set CRE_FR=201560

@REM Set end credit (frame de fin du générique)
set END_FR=207442

@REM Profil

x264.exe --threads auto --thread-input --keyint 24 --min-keyint 1 --crf %E_BR% --vbv-maxrate %MAX_BR% --vbv-bufsize %BUF_BR% --mvrange 511 --merange 16 --level 4.1 --bframe 3 --b-pyramid --b-rdo --bime --weightb --ref 3 --mixed-refs --direct auto --deblock -2:-2 --ipratio 1.10 --pbratio 1.10 --partitions "all" --8x8dct --me "umh" --subme 7 --trellis 2 --no-fast-pskip --no-dct-decimate --aud --nal-hrd --sar 1:1 --cqmfile Sagittaire.cfg --aq-strength 1.00 --aq-mode 2 --zone %CRE_FR%,%END_FR%,b=0.33 --progress --pass 1 --stats "stat.log" -o 1080p_Q1.264 %E_SRC%

pause

@REM -----------------------------------------------------------
@REM
@REM Profil BluRay 1080p23.976 2 passes extra high quality
@REM
@REM -----------------------------------------------------------

@REM Source file name (suffit de mettre la source ici)
set E_SRC=hjhjhhjjh.avs

@REM Set of bitrates (ici le bitrate)
set E_BR=12000

@REM Set of max bitrates (ici le bitrate max)
set MAX_BR=40000

@REM Set of Buffer (ici le buffer)
set BUF_BR=3750000

@REM Profil (ici le nom des fichiers de sortie)
AVS2ASF.exe -i %E_SRC% -o azerty.vc1 -rate %E_BR% -peakrate %MAX_BR% -vbv %BUF_BR% -framerate 23.976 -ratecontrol 3 -profiletype 2 -maxkeydist 24 -bframes 1 -adaptiveGOP -keyPop 1 -inloop 1 -overlap 1 -complexity 5 -motionsearchlevel 2 -mesearchmethod 1 -mbcost 1 -mvcost 1 -mvrange 4 -adaptivequant 1 -dquantoption 1

pause

*INFILE d:\mes dossiers\codec\x264\hjhjhhjjh.avs
*OUTFILE d:\mes dossiers\codec\hcenc\azerty.m2v
*BITRATE 24000
*MAXBITRATE 40000
*FRAMES 0 3042
*PROFILE best
*AUTOGOP 18
*AQ 2
*DC_PREC 8
*PROGRESSIVE
*BIAS 25
*LASTIFRAME
*MPEGLEVEL MP@HL
*MATRIX qlb
*COLOUR 1

MovieSwede

07-07-08, 03:08 AM

Zodiaque

Did you use a 264 source for the encode test?

Zodiaque

07-07-08, 03:27 AM

Zodiaque

Did you use a 264 source for the encode test?

lossless logarith encoding with special pre-process (Darknoise filtering).

Moreover 264 source don't change anything if tou break the original partition structure with resize, crop, convolution, DCT filtering, artificial noise ... etc etc etc. Don't change anything too if your 264 source is extremely high quality source (like BD50) and if you make reencoding at very lower bitrate (like BD9 backup).

MovieSwede

07-07-08, 03:31 AM

No but using a 264 source you giving the VC1 encode both disadvantages from each codec/encoder and non of the benefits.

Zodiaque

07-07-08, 03:43 AM

No but using a 264 source you giving the VC1 encode both disadvantages from each codec/encoder and non of the benefits.

Well it's ridiculous. VC1, H264 or VC1 source don't change anything. x264 will win even if the source is VC1 from HDDVD.

mov source
http://jfl1974.free.fr/images/386_mov.PNG

x264 encoding from my lossless source
http://jfl1974.free.fr/images/386_x264.PNG

x264 at 12 Mbps outperform apple at 10 Mbps ...

MovieSwede

07-07-08, 03:45 AM

Were did you get access to a lossless source?

Zodiaque

07-07-08, 04:06 AM

Were did you get access to a lossless source?

Well it's not the subject of this test:

- x264 outperform last VC1 SDK build for this source.
- VC1 at 12 Mbps is not better than MPEG2 at 24 Mbps

I can use multiple other source to prove that. I can use very clean or noisy source for make that. Darknight source will be available but it's more than 2 Gb source. I think that it's clear demo now ...

MovieSwede

07-07-08, 04:27 AM

To validate the results, its would be nice to know how the lossless source was created.

Zodiaque

07-07-08, 04:36 AM

To validate the results, its would be nice to know how the lossless source was created.

No ... codec can't choose the source. Anyway if you want other source I can make that too. But result will be always the same.

1) Lossless, MPEG2, H264 or VC1 source don't change anything ... x264 actualy produce always the best result in all case at high/medium quant. With trailer or complete movie. In 1 pass and without fastidious reencoding zone like for cinevisionPSE or PEP.

2) VC1 and H264 never produce equivalent quality to MPEG2 with 1:3 ratio for bitrate if you use the best available implementation. And I'am a H264 fan (particulary x264 fan).

MovieSwede

07-07-08, 04:49 AM

No ... codec can't choose the source. Anyway if you want other source I can make that too. But result will be always the same.

1) Lossless, MPEG2, H264 or VC1 source don't change anything ... x264 actualy produce always the best result in all case at high/medium quant. With trailer or complete movie. In 1 pass and without fastidious reencoding zone like for cinevisionPSE or PEP.

So you claiming that

Lossless - VC1
Lossless - H264 - VC1

Isnt a major difference?

I agree that if you have a lossless encode as a source, it wouldnt matter. But if you have used a lossy 264 encode, then will the x264 have serious advantages for the encode.

If we gonna make this scientificly, both the source and the data about the encoding must be accessable to everyone, so the same result can be repeatable.

Zodiaque

07-07-08, 05:08 AM

So you claiming that

Lossless - VC1
Lossless - H264 - VC1

Isnt a major difference?

No ...

Lossless -> H264 -> H264
Lossless -> H264 -> VC1

If H264 is really high quality (30 or 35 Mbps from BD50 for example) and reencoding low/medium quality (BD9 backup) that doesnt' change anything even if you compare with direct encoding from lossless source.

I agree that if you have a lossless encode as a source, it wouldnt matter. But if you have used a lossy 264 encode, then will the x264 have serious advantages for the encode.

Well I can choose high qualiy VC1 source if you want. The result will be exactly the same. I doubt seriousely that H264 source advantage H264 encoding. Anyway it's really simple to break the H264 properties with pre-process.

If we gonna make this scientificly, both the source and the data about the encoding must be accessable to everyone, so the same result can be repeatable.

Well I upload the source actually ... but it's more than 2 Gb source.
http://jfl1974.free.fr/Darknight/Darknigh.part03.rar

MovieSwede

07-07-08, 05:10 AM

Well I upload the source actually ... but it's more than 2 Gb source.
http://jfl1974.free.fr/Darknight/Darknigh.part03.rar

Thank you.

RBFilms

07-07-08, 06:20 AM

My lab agrees that while it technically is VBR ... but it is as close to CBR as you cna get ... since we limited the minimum encode rates and set the mximum encode rates very close together.

There you go ... so it is VBR ... acting sort of like CBR. Either way, the OP complaining is obviously unaware of all this ... and can now let it go as they are incorrect to say we did anything wrong. We clearly did not.

I am getting their field / frame question answered next.

While in theory there's no difference between a VBR encode where ABR=PBR and a CBR encode, in practice there often is. For example, plenty of encoders are 1-pass only for their CBR modes, while offering 2-pass for the VBR modes. 2-pass typically offers more efficient rate control due to lookahead, scene, fade, and flash detection, and other stuff where it has perfect ability to "look into the future."

PEP used to have a CBR mode, but since it was such a rare method to use for HDM, we eventually dropped it since we'd not had any real customer feedback on how well it was working. I don't know of any title that ever shipped using it, particularly since the ability to do segment reencoding flexibly pretty much makes your encode VBR anyway.

scaesare

07-07-08, 08:50 AM

scaesare

07-07-08, 08:55 AM

scaesare

07-07-08, 09:04 AM

No ...

Lossless -> H264 -> H264
Lossless -> H264 -> VC1

If H264 is really high quality (30 or 35 Mbps from BD50 for example) and reencoding low/medium quality (BD9 backup) that doesnt' change anything even if you compare with direct encoding from lossless source.

Well I can choose high qualiy VC1 source if you want. The result will be exactly the same. I doubt seriousely that H264 source advantage H264 encoding. Anyway it's really simple to break the H264 properties with pre-process.

While the source is no doubt visually "high quality", these are perceptually lossy codecs. They rely on human visual sensitivty at 24fps to allow them to throw away 98% of the data and yet still deliver an acceptable motion picture.

However, an encoder is not subject to the same limitations as our visual system is, and sees each individual frame at non-realtime speed. Thus you are asking a (re)encoder to do work with FAR less original source data. Furthermore things that we may NOT see (block edge boundries, filtering effects, motion vector artifacts, chroma subsampling, etc...) are things th encoder is forced to digest.

"Stacking" encodes is a great way to decide how you want to rip your HD discs for b!tt0rr3nt archive, but it's not a very good way to compare encoder implementations.

Zodiaque

07-07-08, 09:31 AM

However, an encoder is not subject to the same limitations as our visual system is, and sees each individual frame at non-realtime speed. Thus you are asking a (re)encoder to do work with FAR less original source data. Furthermore things that we may NOT see (block edge boundries, filtering effects, motion vector artifacts, chroma subsampling, etc...) are things th encoder is forced to digest.

No it's not the case if intermediate encoding is really low quantizer encoding (BD50) and if reencoding is high quantizer encoding (BD9). In this case make BD9 from BD50 or LossLess source will produce equivalent visual result. It's really easy to prove that with metric test for example. It's like make DivX from lossy source or DVD source. DivX at ~1000 Mbps and 720*304 will be not better if you use HD lossless source, HD source or DVD source.

So you are comparing a custom tweaked build against the SDK tool that's a few revsions behind?

Source coming soon. Ben will use all the available vc1 custom build. Anyway there are no doubt for the result because I use in practice insane setting for the vc1 codec with high quality pre-process for the source (Dark-noise filtering) for all codec.

Is anybody really targeting 8Mbps for HD disc use? I really don't care what that will prove.

Well DVDForum make test at 24 Mbps and 8 Mbps. I just use exactly the same bitrate. But I don't find the same result ... never. I can use many real source for prove that.

I'am ready for the Challenge ... it's really easy for me to prove that. Choose your source. I have for example high quality source like Matrix (HDDVD in VC1) with high grain level if you want. Use VC1 source for x264 will be not a big desavantage.

sperron

07-07-08, 10:08 AM

Ben, do you have any comments on why US lionsgate AVC release of Rambo contains more detail then the German Warner VC-1 release? Here's a link to a mouseover comparison: http://horn.hdtvtotal.com/hdtvtotal/scripts/waggle/waggle.php?hd=http://chidragon.thedessie.com/bdcomp/rambocomp/rambo_de.png&sd=http://chidragon.thedessie.com/bdcomp/rambocomp/rambo_us.png

The stats: Blu-ray - Germany - Warner - 17.97Mbps VC-1 vs Blu-ray - USA - Lionsgate - 26.82Mbps AVC

Many have long suspected that Warner filters thier releases which would appear to be confirmed in this comparison unless the VC-1 encoder somehow caused this.

Zodiaque

07-07-08, 10:29 AM

The stats: Blu-ray - Germany - Warner - 17.97Mbps VC-1 vs Blu-ray - USA - Lionsgate - 26.82Mbps AVC

There are many filter in CinevisionPSE (noise, darknoise ...) and the codec itself can soften the picture. But you can't compare VC1 at 18 Mbps and H264 at 27 Mbps because H264 encoding use 50% bitrate more than VC1. H264 at 12 Mbps encoding will be certainely not better than VC1 at 18 Mbps.

benwaggoner

07-07-08, 10:45 AM

benwaggoner

07-07-08, 11:00 AM

Well DVDForum make test at 24 Mbps and 8 Mbps. I just use exactly the same bitrate. But I don't find the same result ... never. I can use many real source for prove that.
Bear in mind that the DVD Forum tests predated (and were a major motivator for) high profile.

Both VC-1 and H.264 have dramatically improved since that testing; I'm certainly glad we didn't wind up having to use 7.7 Mbps of that era's VC-1 implementation for HD media!

benwaggoner

07-07-08, 11:06 AM

AVS2ASF.exe -i %E_SRC% -o azerty.vc1 -rate %E_BR% -peakrate %MAX_BR% -vbv %BUF_BR% -framerate 23.976 -ratecontrol 3 -profiletype 2 -maxkeydist 24 -bframes 1 -adaptiveGOP -keyPop 1 -inloop 1 -overlap 1 -complexity 5 -motionsearchlevel 2 -mesearchmethod 1 -mbcost 1 -mvcost 1 -mvrange 4 -adaptivequant 1 -dquantoption 1
Hmmm.

You definitely don't want DQuant on at 8 Mbps with the VC1ESDK. Is that mvrange the adaptive one? That's normally your best bet.

Still downloading the source, so I won't have any detailed opinions on that until probably next week (I'm leaving for a trip in a couple of hours).

benwaggoner

07-07-08, 11:08 AM

Well I upload the source actually ... but it's more than 2 Gb source.
http://jfl1974.free.fr/Darknight/Darknigh.part03.rar
This looks like it's just a single 200 MB chunk. Is there more?

Zodiaque

07-07-08, 11:56 AM

Bear in mind that the DVD Forum tests predated (and were a major motivator for) high profile.

Both VC-1 and H.264 have dramatically improved since that testing; I'm certainly glad we didn't wind up having to use 7.7 Mbps of that era's VC-1 implementation for HD media!

Well I think that DVDForum test is simply not serious test ...

This looks like it's just a single 200 MB chunk. Is there more?

There are 12 rar files ...

You definitely don't want DQuant on at 8 Mbps with the VC1ESDK. Is that mvrange the adaptive one? That's normally your best bet

Just AQ for IFrame here. mvrange is adaptative. This profil is insane setting for VC1 SDK (mesearch, hadamard, rdo ...).

@ Ben

What do you think about 12 Mbps Darknight x264 encoding. Really impressive for fine grain retention, isn't it?
I make that with the same quality level at 7 Mbps for Casino Royal (high grain level retention).

sspears

07-07-08, 12:46 PM

Well I think that DVDForum test is simply not serious test ...

The golden eyes of Hollywood participated. People like Don Eklund were present. The content, as I recall, included: Se7en, Dick Tracy, Harry Potter, Monsters Inc., Stuart Little 2 and Titan AE.

It was always played back in split screen with the source on one side and the compressed CODEC on the other. Multiple display technologies were used.

Zodiaque

07-07-08, 01:39 PM

The golden eyes of Hollywood participated. People like Don Eklund were present. The content, as I recall, included: Se7en, Dick Tracy, Harry Potter, Monsters Inc., Stuart Little 2 and Titan AE.

It was always played back in split screen with the source on one side and the compressed CODEC on the other. Multiple display technologies were used.

Well perhaps good eyes but bad MPEG2 encoding ... old and crappy wmv-hd implementation at 7.7 Mbps better than 24 Mbps MPEG2 encoding? It's simply ridiculous. Even today really good implementation like VC1 SDK can't fight with MPEG2 at 3:1 ratio.

amirm

07-07-08, 01:56 PM

Well perhaps good eyes but bad MPEG2 encoding ...
Those MPEG-2 encodes at 7.7 mbit/sec beat AVC at the same rate ;). So they couldn't have been that bad!

old and crappy wmv-hd implementation at 7.7 Mbps better than 24 Mbps MPEG2 encoding?
"Better?" No one said anything about better or even the same across the board. If we have this much trouble communicating (I have corrected you all three times on this), heaven help us when it comes to more complicated topics!

And how is MPEG-2 after a decade+ not “old and crappy” regardless of encoder?

It's simply ridiculous. Even today really good implementation like VC1 SDK can't fight with MPEG2 at 3:1 ratio.
What is ridiculous is that not only have you not spent a minute understanding the DVD test scenario, but you don't even read the things people tell you about it. Right now, your idea of you being the one encoding VC-1 is more laughable than anything DVD Forum has ever done. At least in that test, the proponents of each technology did their own work. You are not even using the same tool any Hollywood title is encoded in HD DVD/BD for heaven’s sake. So clean up your own house before you throw rocks at someone else’s.

Zodiaque

07-07-08, 03:38 PM

amirm

07-07-08, 04:20 PM

Well I make encoding at 2:1 ratio for MPEG2/VC1. And even at 2:1 VC1 not matched the MPEG-2.
By all means do all the tests you want. But none of that has any bearing on DVD Forum results unless you follow the same procedure to the letter, including double-blind tests with controls, large number of viewers, multiple display types, and mix of golden eye viewers and industry experts.

As I have said before, every benchmark is flawed. Some are more flawed than others. Reading your posts, right now, I consider your tests more flawed than DVD Forum.

I will speak about with codec developper (VC1, MPEG2 and H264) and we will see if DVDforum test is ridiculous or not.
You don't need to talk to a codec developer. You need to talk to someone who understands how to conducts tests of this sort. That may or may not be the same as a codec developer.

I'am afraid to know already the answer.
Oh that is a big surprise ;).

Anyway I suspect Microsoft to use heavy pre-process to prove that VC1 is the best.
This right there, is what we call "shooting from the hip." ;) :). The second round of DVD forum tests was conducted because they suspected the same. As such, in that round, they stipulated that use of any preprocessing would disqualify the participant. Further, they required all the steps to reproduce the submitted bitstream so that they could verify if anyone cheated. You better believe that the people who lost out went and did the verification with the steps Microsoft provided. And found nothing to complain about VC-1.
Well MPEG2 progress always.
You called WMV/VC-1 old yet seem to think MPEG-2 is staying young and fresh. “This right there, is what we call a double standard. :D”

My prefered MPEG2 implementation is libavcodec. Produce really terrible result at low bitrate.
And that is the problem with tweaked implementations. Like you are using now to test AVC. No Hollywood studio wants to play with different encoders for different situations. VC-1 performs great whether Ben is creating Silverlight 720p content at less than DVD MPEG-2 SD data rates or BD’s peak rate. That is the real trick. Running a marathon and sprint just as well….

With this implementation I obtain result close to VC1 at high bitrate for PSNR.
And I should care about PSNR why? Do people have PSNR meters at home that they watch or movies? If PSNR was valid, we wouldn’t have compressionists encoding DVDs/BDs.

For make ratio quality equivalence Microsoft use PSNR ... I prove that these ratio are completely false with libavcodec ...
Microsoft’s technology did not getting into BD/HD DVD because of PSNR. It got in there because real humans judged it superior to MPEG-2 and AVC. However, MPEG-2 and AVC got into the standard because folks like you, had the bias in favor of it regardless of their performance at the time. So you want to go after somebody, go after the people who forced three codecs to be mandatory in BD when one would do. There is a much bigger sin than Microsoft showing PSNR, using MPEG test clips, that VC-1 performs better than MPEG-2.
Anyway you have source now for prove that VC1 is better than H264 and close to MPEG2 at 1:3 ratio. I'am now really curious to see that.
You don’t have a real source. It is precompressed material and if it comes from a BD source, no company would ever want to download it, lest they encourage breaking copy protection.

Finally, keep in mind that no one in this forum makes decisions on which codec to use for a Hollywood title. If you have a story to tell, go and tell the studios. Telling people here won’t do you any good.

vsv

07-07-08, 06:25 PM

Is anybody really targeting 8Mbps for HD disc use?

Yes. Read about RED-RAY - 120 min 4K on DVD9 (http://www.reduser.net/forum/forumdisplay.php?f=38)

Or this:

Blu-ray DVD creation for the masses

By Matt Armstrong
July 3, 2008 – 7:00 am

One of the more interesting tidbits to come out of the series of panels at the CineGear Expo came from Larry Jordan of Digital Production Buzz. He related the findings of Bruce Nazarian, president of the DVD Association, who has found a way to burn standard DVD-5 or DVD-9 discs with the current red laser technology using the H.264 codec to play in high def in Blu-ray players. Obviously this is huge news for those that want to master to high def but avoid the cost of AACS licensing as well as costly Blu-ray burners. But the true breakthrough would be in replication, truly opening up high-def delivery for the indie filmmaker. Not to worry, Nazarian, according to Jordan, is currently working on replication tests and should come out with his findings soon. Stay tuned! (http://www.studiodaily.com/blog/?p=580)

RBFilms

07-07-08, 07:10 PM

Going on a year later, it may be time to let all this go. However, in the interest of assuring that ALL information and FULL DISCLOSURE on this topic has been provided for those that must know, here is the FINAL ANSWER regarding the encoding choices that we made for Nature's Journey.

The comments below also back up my original statements regarding the fact that the VC-1 CODEC was not mature enough to use for CHRONOS. It was better by the time we got around to producing Nature's Journey, but still not where it is today. VC-1 has improved since that time, but we still plan to use Tiger AVC for encoding all projects moving forward. There will be no VC-1 titles coming out of our Studio unless it is a title we acquire completed from a Producer with all the Encoding and authoring work already completed.

Also, as I understand from conversation with my compression technician, "Certain" people on the AVS thread did not (and still do not) have all the facts surrounding this particular project. At one point our compression technician even asked their Microsoft contact to inform these "Certain" individuals why Natures Journey was encoded the way it was. It's sad to see that these so called "experts" have spent so much time debating over production specifics for which they have no clairvoyance whatsoever as to the actual project details & constraints.

In summary, I still do not understand why folks assume they know everything about our projects from its source videography / content and film transfers to its rendered file delivery, to how it should or should not have been encoded ... this is all a mystery to me.

In summary, the Natures Journey project was very well handled and massaged down to every last bit. For what it's worth, I believe this title will stand the test of time regardless of any forum critics who presume to be "all knowing".

************************************************************ *******

Here is the response from one of our lead High Definition VC1 compression technicians who actually spent "many" hours on the Natures Journey project (both HD DVD and Blu-ray Disc encode versions):

-----

"At the time that Natures Journey was encoded (last year), the 1st Generation Samsung Blu-ray Disc Players had issues with Frame/Field Auto Mode VC1 encodes. This was what Microsoft had told us. So we tried encoding it in Field only mode. The Field only mode had issues in trying to re-encode shots where the re-encodes ended up looking worse than the originals. We consulted Microsoft again and they had no solution at the time, and so they requested us to send them the file and XML. Moving forward we ran some tests with encoding it in Frame only
mode which allowed for the re-encodes to work and we were able to fix the QP spikes and other re-encode parameters which were (up to that point) failing the Blu-ray Disc Verifiers.

This was very difficult source material to encode and it took a lot of re-encoding and some source redeliveries to make the picture look good and to pass the Verifiers. Some of the difficulties were caused by the codec itself trying to deal with the source material. For instance, the source had a lot of recorded in Macro blocking that proved difficult for the codec to handle. In addition, because of the challenges with the source material, we had to carefully manage the re-encode settings in order to avoid buffer allocation errors that were being flagged by the verifiers.

Yes, technically it was encoded in VBR because at the time the encoder would not do a true CBR which is what was requested to maximize the bits across the entire disc. So in order to achieve this, the difference between the Average bitrate and the Maximum bitrate were kept extremely narrow thus making it virtually a CBR encode at a very high average.

The VC1 encoder has come a long way since last year, and now the current version is a lot more superior than it was back then. Nonetheless, we are still reluctant to use the Field only mode as it still tends to give us problems with the re-encodes. Also, Microsoft has told us that Frame/Field auto mode has been fixed for use in the 1st Generation Blu-ray Disc players, but we still haven't found the best results from using this auto mode."

************************************************************ ************

It was not only the CBR-like nature of the encode. It was the parameter fed to the encoder regarding linterlacing:

(Original post HERE (http://www.avsforum.com/avs-vb/showthread.php?p=11979429#post11979429))

The frame-progressive mode appears to have been selected in error here, and it is the least "devleoped" of the modes.

Zodiaque

07-07-08, 07:21 PM

Partial result: VC1 vs H264 vs MPEG2

http://jfl1974.free.fr/Darknight/Darknight_x264_12Mbps.264
http://jfl1974.free.fr/Darknight/Darknight_SDK_12Mbps.vc1
http://jfl1974.free.fr/Darknight/Darknight_HCEnc_24Mbps.m2v

- H264 with x264 special build
- VC1 with MS VC1 SDK encoder
- MPEG2 with HCEnc

The source

http://jfl1974.free.fr/Darknight/
Source with high quality dark noise filtering pre-process.

next update:

- VC1 and H264 encoding at 8 Mbps

Annexe:

@REM -----------------------------------------------------------
@REM
@REM Profil BluRay 1080p23.976 extra high quality
@REM
@REM -----------------------------------------------------------

@REM Source file name (suffit de mettre la source ici)
set E_SRC=hjhjhhjjh.avs

@REM Set of quality (ici la qualité 1-50)
set E_BR=22

@REM Set of max bitrate (ici le bitrate max)
set MAX_BR=40000

@REM Set of Buffer (ici le buffer)
set BUF_BR=30000

@REM Set credit (frame de début du générique)
set CRE_FR=201560

@REM Set end credit (frame de fin du générique)
set END_FR=207442

@REM Profil

x264.exe --threads auto --thread-input --keyint 24 --min-keyint 1 --crf %E_BR% --vbv-maxrate %MAX_BR% --vbv-bufsize %BUF_BR% --mvrange 511 --merange 16 --level 4.1 --bframe 3 --b-pyramid --b-rdo --bime --weightb --ref 3 --mixed-refs --direct auto --deblock -2:-2 --ipratio 1.10 --pbratio 1.10 --partitions "all" --8x8dct --me "umh" --subme 7 --trellis 2 --no-fast-pskip --no-dct-decimate --aud --nal-hrd --sar 1:1 --cqmfile Sagittaire.cfg --aq-strength 1.00 --aq-mode 2 --zone %CRE_FR%,%END_FR%,b=0.33 --progress --pass 1 --stats "stat.log" -o 1080p_Q1.264 %E_SRC%

pause

@REM -----------------------------------------------------------
@REM
@REM Profil BluRay 1080p23.976 2 passes extra high quality
@REM
@REM -----------------------------------------------------------

@REM Source file name (suffit de mettre la source ici)
set E_SRC=hjhjhhjjh.avs

@REM Set of bitrates (ici le bitrate)
set E_BR=12000

@REM Set of max bitrates (ici le bitrate max)
set MAX_BR=40000

@REM Set of Buffer (ici le buffer)
set BUF_BR=3750000

@REM Profil (ici le nom des fichiers de sortie)
AVS2ASF.exe -i %E_SRC% -o azerty.vc1 -rate %E_BR% -peakrate %MAX_BR% -vbv %BUF_BR% -framerate 23.976 -ratecontrol 3 -profiletype 2 -maxkeydist 24 -bframes 1 -adaptiveGOP -keyPop 1 -inloop 1 -overlap 1 -complexity 5 -motionsearchlevel 2 -mesearchmethod 1 -mbcost 1 -mvcost 1 -mvrange 4 -adaptivequant 1 -dquantoption 1

pause

*INFILE d:\mes dossiers\codec\x264\hjhjhhjjh.avs
*OUTFILE d:\mes dossiers\codec\hcenc\azerty.m2v
*BITRATE 24000
*MAXBITRATE 40000
*FRAMES 0 3042
*PROFILE best
*AUTOGOP 18
*AQ 2
*DC_PREC 8
*PROGRESSIVE
*BIAS 25
*LASTIFRAME
*MPEGLEVEL MP@HL
*MATRIX qlb
*COLOUR 1

Zodiaque

07-07-08, 07:26 PM

Microsoft’s technology did not getting into BD/HD DVD because of PSNR. It got in there because real humans judged it superior to MPEG-2 and AVC. However, MPEG-2 and AVC got into the standard because folks like you, had the bias in favor of it regardless of their performance at the time. So you want to go after somebody, go after the people who forced three codecs to be mandatory in BD when one would do. There is a much bigger sin than Microsoft showing PSNR, using MPEG test clips, that VC-1 performs better than MPEG-2.

Well you are sure for that ... :rolleyes:

http://www.avsforum.com/avs-vb/showthread.php?p=11189339#post11189339

Your internal test are simply ridiculous too ... and I can prove that when you want ... it's really simple with objective test.

|--------------|---------|---------|----------|---------|---------|
| Codec | PProc | Bitrate | Size | OPSNR | SSIM 2 |
|--------------|---------|---------|----------|---------|---------|
| MPEG2 | No | 5999 | 479233 | 43.08 | 82.55 |
| VC-1 | No | 5994 | 479405 | 44.15 | 85.85 |
| H264 | No | 6007 | 479766 | 46.35 | 89.54 |
|--------------|---------|---------|----------|---------|---------|
| MPEG2 | No | 12001 | 958666 | 46.38 | 91.27 |
| VC-1 | No | 11987 | 957575 | 47.30 | 92.25 |
| H264 | No | 12008 | 959329 | 49.11 | 94.18 |
|--------------|---------|---------|----------|---------|---------|
| MPEG2 | No | 18002 | 1438146 | 47.17 | 93.26 |
| VC-1 | No | 18005 | 1438379 | 49.00 | 94.60 |
| H264 | No | 18009 | 1438862 | 50.53 | 95.74 |
|--------------|---------|---------|----------|---------|---------|
SSIM 0: Lumimask Off
SSIM 1: Lumimask On (Original Lumimask)
SSIM 2: Lumimask On (One2Tech Patch)

bobgpsr

07-07-08, 07:29 PM

re: Final Answer on Nature's Journey Encodes

I confused. How did Phantom of the Opera look so good if there were "issues with Frame/Field Auto Mode VC1 encodes"? Or was this only a Blu-ray problem with early Samsung BD players?

Don't understand the "re-encode" thingy. Could this be explained please?

sspears

07-07-08, 08:06 PM

I confused. How did Phantom of the Opera look so good if there were "issues with Frame/Field Auto Mode VC1 encodes"? Or was this only a Blu-ray problem with early Samsung BD players?

Frame/Field Auto is for interlaced content. Phantom used the Progressive encoding mode.

amirm

07-07-08, 08:13 PM

Well you are sure for that ... :rolleyes:
Yes, I am sure that DVD Forum and BDA did not pick VC-1 because of PSNR scores. ;) Seems like you read every other sentence of anyone who answers you. Address this point for me. Tell me why you are not upset that both MPEG-2 and AVC lost in blind evaluations yet they got picked to be mandatory in the format. If you want to cry foul, you can’t be selective.

Your internal test are simply ridiculous too ...
So what else is new? Another test you know nothing about, but are sure that it is ridiculous. I guess it was magic that managed to stuff 1080p content on red laser DVDs some 3+ years ago, in the same space as MPEG-2 SD. Not VC-1.
and I can prove that when you want ... it's really simple with objective test.
Now we are getting someplace. So you admit that objective results can be cooked yet you and your friend keep asking for shootouts using the same type of measure?

bobgpsr

07-07-08, 08:45 PM

Frame/Field Auto is for interlaced content. Phantom used the Progressive encoding mode.Ah! So Nature's Journey truly is shot with interlaced video cameras and not in progressive or with film. Now I remember the comments at the time that led me to skip this purchase.

I found this quote (http://www.avsforum.com/avs-vb/showthread.php?p=11959753#post11959753)from TheLion:
This release is softer and less detailed (-> due to the used "prosumer" cameras) than even mediocre 35mm sourced content.

Wasn't there a lot of work done later on VC-1 to make it work well with rock concert video? Perhaps it was too late for Nature's Journey .

scaesare

07-07-08, 10:29 PM

Yes. Read about RED-RAY - 120 min 4K on DVD9 (http://www.reduser.net/forum/forumdisplay.php?f=38)

Or this:

Blu-ray DVD creation for the masses

By Matt Armstrong
July 3, 2008 – 7:00 am

One of the more interesting tidbits to come out of the series of panels at the CineGear Expo came from Larry Jordan of Digital Production Buzz. He related the findings of Bruce Nazarian, president of the DVD Association, who has found a way to burn standard DVD-5 or DVD-9 discs with the current red laser technology using the H.264 codec to play in high def in Blu-ray players. Obviously this is huge news for those that want to master to high def but avoid the cost of AACS licensing as well as costly Blu-ray burners. But the true breakthrough would be in replication, truly opening up high-def delivery for the indie filmmaker. Not to worry, Nazarian, according to Jordan, is currently working on replication tests and should come out with his findings soon. Stay tuned! (http://www.studiodaily.com/blog/?p=580)

OK... I guess I had to spell out the context of this discussion: for movie encodes.

Red ray is intriguing, but outside the specs of what's being discussed.

And home footage burning to DVD media is not what studios producing feature films will target.

scaesare

07-07-08, 10:37 PM

Going on a year later, it may be time to let all this go. However, in the interest of assuring that ALL information and FULL DISCLOSURE on this topic has been provided for those that must know, here is the FINAL ANSWER regarding the encoding choices that we made for Nature's Journey.

The comments below also back up my original statements regarding the fact that the VC-1 CODEC was not mature enough to use for CHRONOS. It was better by the time we got around to producing Nature's Journey, but still not where it is today. VC-1 has improved since that time, but we still plan to use Tiger AVC for encoding all projects moving forward. There will be no VC-1 titles coming out of our Studio unless it is a title we acquire completed from a Producer with all the Encoding and authoring work already completed.

Also, as I understand from conversation with my compression technician, "Certain" people on the AVS thread did not (and still do not) have all the facts surrounding this particular project. At one point our compression technician even asked their Microsoft contact to inform these "Certain" individuals why Natures Journey was encoded the way it was. It's sad to see that these so called "experts" have spent so much time debating over production specifics for which they have no clairvoyance whatsoever as to the actual project details & constraints.

In summary, I still do not understand why folks assume they know everything about our projects from its source videography / content and film transfers to its rendered file delivery, to how it should or should not have been encoded ... this is all a mystery to me.

In summary, the Natures Journey project was very well handled and massaged down to every last bit. For what it's worth, I believe this title will stand the test of time regardless of any forum critics who presume to be "all knowing".

************************************************************ *******

Here is the response from one of our lead High Definition VC1 compression technicians who actually spent "many" hours on the Natures Journey project (both HD DVD and Blu-ray Disc encode versions):

-----

"At the time that Natures Journey was encoded (last year), the 1st Generation Samsung Blu-ray Disc Players had issues with Frame/Field Auto Mode VC1 encodes. This was what Microsoft had told us. So we tried encoding it in Field only mode. The Field only mode had issues in trying to re-encode shots where the re-encodes ended up looking worse than the originals. We consulted Microsoft again and they had no solution at the time, and so they requested us to send them the file and XML. Moving forward we ran some tests with encoding it in Frame only
mode which allowed for the re-encodes to work and we were able to fix the QP spikes and other re-encode parameters which were (up to that point) failing the Blu-ray Disc Verifiers.

This was very difficult source material to encode and it took a lot of re-encoding and some source redeliveries to make the picture look good and to pass the Verifiers. Some of the difficulties were caused by the codec itself trying to deal with the source material. For instance, the source had a lot of recorded in Macro blocking that proved difficult for the codec to handle. In addition, because of the challenges with the source material, we had to carefully manage the re-encode settings in order to avoid buffer allocation errors that were being flagged by the verifiers.

Yes, technically it was encoded in VBR because at the time the encoder would not do a true CBR which is what was requested to maximize the bits across the entire disc. So in order to achieve this, the difference between the Average bitrate and the Maximum bitrate were kept extremely narrow thus making it virtually a CBR encode at a very high average.

The VC1 encoder has come a long way since last year, and now the current version is a lot more superior than it was back then. Nonetheless, we are still reluctant to use the Field only mode as it still tends to give us problems with the re-encodes. Also, Microsoft has told us that Frame/Field auto mode has been fixed for use in the 1st Generation Blu-ray Disc players, but we still haven't found the best results from using this auto mode."

************************************************************ ************

Richard, thanks for closing this loop.

Interesting that there were quite a few VC1 discs available at this time already, and yet this issue hasn't seemed to creep up elsewhere. Now that you know it was problem with first-generation hrdware that kept your enocding tech from using the best mode of the encoder, are you interested in re-evaluating?

Also, can you expand on the issue of "re-encoding" the source material that apparently already had macro-clocks that were difficult to encode? What was the source? D5?

Thanks.

-Steve

PS- For the record, I don't believe Stacey EVER claimed to know the inside scoop on why this was encoded the way it was... he only disclosed what he found when examining the file.

RBFilms

07-08-08, 12:12 AM

You are welcome...and it is closed.

The source was .TIFF Files...fed directly to the encoder.

I am sold on Tiger AVC for our work...we have completed our testing and evaluation with the latest versions ... and that is what we have chosen....at least for now or until something better comes along.

Thank you.

Richard, thanks for closing this loop.

Interesting that there were quite a few VC1 discs available at this time already, and yet this issue hasn't seemed to creep up elsewhere. Now that you know it was problem with first-generation hrdware that kept your enocding tech from using the best mode of the encoder, are you interested in re-evaluating?

Also, can you expand on the issue of "re-encoding" the source material that apparently already had macro-clocks that were difficult to encode? What was the source? D5?

Thanks.

-Steve

PS- For the record, I don't believe Stacey EVER claimed to know the inside scoop on why this was encoded the way it was... he only disclosed what he found when examining the file.

RBFilms

07-08-08, 01:36 AM

If you read the comments..it is still an issue.

Richard, thanks for closing this loop.

Interesting that there were quite a few VC1 discs available at this time already, and yet this issue hasn't seemed to creep up elsewhere. Now that you know it was problem with first-generation hrdware that kept your enocding tech from using the best mode of the encoder, are you interested in re-evaluating?

Also, can you expand on the issue of "re-encoding" the source material that apparently already had macro-clocks that were difficult to encode? What was the source? D5?

Thanks.

-Steve

PS- For the record, I don't believe Stacey EVER claimed to know the inside scoop on why this was encoded the way it was... he only disclosed what he found when examining the file.

benwaggoner

07-08-08, 03:27 AM

Wasn't there a lot of work done later on VC-1 to make it work well with rock concert video? Perhaps it was too late for Nature's Journey .
That was all about handling flashes/strobes in the NiN title. However, it was 30p IIRC, so still would have been progressive encoding.

Golgot13

07-08-08, 05:07 AM

You don’t have a real source. It is precompressed material and if it comes from a BD source, no company would ever want to download it, lest they encourage breaking copy protection.

Zodiaque, you can use a second video, from me, it's uncompressed video from HDCAM SR (Ben have it).
About copy protection breaking, it's simple way to finish the comparison because somebody can not or
don't want to give a good video source (1920x1080@24P with movie grain).

One of the more interesting tidbits to come out of the series of panels at the CineGear Expo came from Larry Jordan of Digital Production Buzz. He related the findings of Bruce Nazarian, president of the DVD Association, who has found a way to burn standard DVD-5 or DVD-9 discs with the current red laser technology using the H.264 codec to play in high def in Blu-ray players. Obviously this is huge news for those that want to master to high def but avoid the cost of AACS licensing as well as costly Blu-ray burners. But the true breakthrough would be in replication, truly opening up high-def delivery for the indie filmmaker. Not to worry, Nazarian, according to Jordan, is currently working on replication tests and should come out with his findings soon. Stay tuned!

Hi Serguey,
I can answer that this format "AVCHD" can not work on all BD player.
I make many test with replicator, this format (AVCHD) is a BD9 with a modification of
*.bdmv file, a specific LeadIn (DVDR LeadIn and datarate at 30Mbps) and no AACS.
Many BD player which support BD9 don't play it because it don't support BD9 without AACS
and many BD player don't play BD9... (there is only Panasonic, Sony and some Pioneer BD player which play it fine).

But may be with last format "AVCRec" (BD on DVDR: Camcorder or BD+Tuner recorder),
we will have more BD player which will support it (it's like DVD-Video on CD).