Can someone say if there is a simple way to burn these to a HD DVD 9, disc so that I can play them on my A1?

There are some nice test patterns there.

dr1394, I watched Park Run 1920x1080 clip (http://www.w6rz.net/parkrun1920_18mbps.ts) today (18Mbps).

I used the xport.exe to extract .mpv file, and watched it with Nero ShowTime. I am seeing a lot of artifacts. Since I am watching it on a laptop (Dell E1705, with 1920x1200 resolution), I paused and stepped each frame to make sure that LCD screen was not the limiting factor. The best I can describe is as part aliasing part macroblocks. I also see a bit of de-interlacing issues every now and then. Is the source interlaced?

I know, this is a demanding shot with a full horizontal pan, but I was expecting a bit more detail, and less artifacts. I am no expert, so I cannot describe this any better than this.

Is there anything else I can do? May be a different player? What is your suggestion?

PS: Thanks for the great collection of test patterns! Great work! I didn't pay attention to your video clips until trbarry mentioned it!
"Park Run" is very difficult material indeed. The original motivation behind creating those clips was to show that sub-sampled horizontal resolutions were not the bane on mankind as some would lead you to believe, but could actually provide better quality for tough material at ATSC and satellite bitrates. It also shows that even 18 Mbps is not enough for all material. In fact, the 23 Mbps version has some warts too.

It is interlaced. In fact, all the 1920x1080 clips on the VQEG site are interlaced, so they're not the best clips for determining film encoding quality (the main focus of this thread). They're also not long enough for a VBR algorithm to really do it's thing. However, they are real uncompressed content made publicly available for testing, so they are a bit special.

As for video quality assessment, I'm not a big fan of software decoders. Of course, HD-DVD and (eventually) Blu-ray players will get good HD decoders into the hands of hobbyists. However, I do like LCD displays, just because they are so unforgiving and represent a real display they consumers are actually using (rather than professional HDTV monitors which are pretty forgiving and not in anyone's home). At work, I use both.

The MPEG-2 HD Test Patterns website is pretty popular. I average about 30 Gigabytes of downloads a month. My provider upped the account bandwidth from 100 GB to 400 GB, so I decided to expose the links for the video quality clips on the main page.

Ron

Can someone say if there is a simple way to burn these to a HD DVD 9, disc so that I can play them on my A1?

There are some nice test patterns there.
See the "HD-DVD Authoring to DVD -/+ Media" thread.

http://www.avsforum.com/avs-vb/showthread.php?t=667462

In particular, see my post about getting around the problems with MF5 and the test pattern files.

http://www.avsforum.com/avs-vb/showthread.php?p=7530158&&#post7530158

I believe that some folks have successfully burned the patterns.

Ron

@benwaggoner et al.:

I hate to keep harping on this every few months, but...

Can the restrictions on codec settings for encoding VC-1 and H.264 for HD DVD (and Blu-Ray too!) finally be revealed?? Last time I asked about this I was (politely) informed that NDAs couldn't be broken etc. I really don't understand why this info is still "top secret", since you can now buy the actual HD DVD discs and players, and Sonic has finally released their (way too expensive) authoring app for both HD DVD and Blu-Ray. But, sure enough, I have googled and scoured the whole freakin' internet - and there is no more info out there then there was 6 months ago! Here are the kinds of questions that need answering:

Does HD DVD support High Profile H.264? ...and @ level 4.1?

If so, does HD DVD (and Blu-Ray, for that matter) support all High Profile features, like Custom Quanitzation Matrices, for example?

What precisely is the maximum peak bitrate for H.264 and VC-1 streams for HD DVD? I have seen conflicting info on this.

Do any of these codecs need a minumum bitrate specified?

What is the allowable (range of) GOP size for HD DVD? Blu-Ray?

Does HD DVD have a specific GOP structure (e.g. IBBPBBPBBPBB) that must be followed? How about Blu-Ray?

Can VC-1 or H.264 stream encoded at 720p24 or 1080p24 be muxed "as is" for HD DVD, or does it need to be "pulldowned" or whatever to 60i? How about for Blu-Ray?

Will HD DVD Forum or BDA ever intend to publish offical specs to answer these types of questions?

Bonus question: Will Windows Media Encoder Studio Edition definitely be able to encode HD DVD compliant VC-1 streams for later muxing?

Thanks in advance if you can answer any or all of these questions!

lazyn00b

dr1394, I watched Park Run 1920x1080 clip (http://www.w6rz.net/parkrun1920_18mbps.ts) today (18Mbps).

I used the xport.exe to extract .mpv file, and watched it with Nero ShowTime. I am seeing a lot of artifacts. Since I am watching it on a laptop (Dell E1705, with 1920x1200 resolution), I paused and stepped each frame to make sure that LCD screen was not the limiting factor. The best I can describe is as part aliasing part macroblocks. I also see a bit of de-interlacing issues every now and then. Is the source interlaced?

I know, this is a demanding shot with a full horizontal pan, but I was expecting a bit more detail, and less artifacts. I am no expert, so I cannot describe this any better than this.

Is there anything else I can do? May be a different player? What is your suggestion?

PS: Thanks for the great collection of test patterns! Great work! I didn't pay attention to your video clips until trbarry mentioned it!

Freeze frame on 1080i typically looks bad on modern hi rez fixed pixel displays.

Anybody curious about the underlying uncompressed clip might want to view it deinterlaced to 1080p/60 YUY2. If you have Avisynth installed this can be done by downloading one of the uncompressed clips above from the VQEG site and using the following Avisynth script with my TomsMoComp filter to view it, though very likely not at full speed.


RawSource("g:\VCR\Uncompressed\Park Run 1080i\ParkRun_1080i.uyvy",1920,1080,"UYVY")
Top=AssumeFrameBased().ComplementParity()
Bottom=Top.SeparateFields.Trim(1,0).weave
Interleave(Top.TomsMoComp(1,5,1),Bottom.TomsMoComp(0,5,1))
return last


Of course be sure to change the file name to wherever. An updated TomsMoComp.dll filter can be found on the downloads section of my web site that will deinterlace this, though there are of course other choices. Just be sure to use something that keeps a bit more than 540 vertical pixels of detail.

Hopefully someone will use this to make some good 1080p60 clips with the most modern codecs. ;)

- Tom

In fact, all the 1920x1080 clips on the VQEG site are interlaced, so they're not the best clips for determining film encoding quality (the main focus of this thread). They're also not long enough for a VBR algorithm to really do it's thing. However, they are real uncompressed content made publicly available for testing, so they are a bit special.
So I've noticed. Are there ally 1080p24/25 sources along these lines we could use?

If not, I can certainly do a nice deinterlace, but I don't know how good a test that would be.

So, let me flip this around to the community - would would YOU want to see in downloable codec samples?

@benwaggoner et al.:

I hate to keep harping on this every few months, but...

Can the restrictions on codec settings for encoding VC-1 and H.264 for HD DVD (and Blu-Ray too!) finally be revealed?? Last time I asked about this I was (politely) informed that NDAs couldn't be broken etc. I really don't understand why this info is still "top secret", since you can now buy the actual HD DVD discs and players, and Sonic has finally released their (way too expensive) authoring app for both HD DVD and Blu-Ray. But, sure enough, I have googled and scoured the whole freakin' internet - and there is no more info out there then there was 6 months ago! Here are the kinds of questions that need answering:
I haven't talked to our legal about this recently, but the standards agreements are more restrictive than you might imagine. I'm not directly involved, but we certainly have people working to find the right way to share information to the public. You'll especially be seeing this with iHD. But I can answer a few of the below. We'll have a lot more details when we have public tools for VC-1 encoding for HD DVD, obviously.

Does HD DVD support High Profile H.264? ...and @ level 4.1?
Yes, High Profile has been made public.

Do any of these codecs need a minumum bitrate specified?By spec? Normally a minimum datarate is used in MPEG-2 encoders to get around VBR starving easy content, resulting in ringing in a MPAA graphic, for example. VC-1 doesn't have that problem.

Does HD DVD have a specific GOP structure (e.g. IBBPBBPBBPBB) that must be followed? How about Blu-Ray?VC-1 allows different B-frame patterns for HD DVD.

Can VC-1 or H.264 stream encoded at 720p24 or 1080p24 be muxed "as is" for HD DVD, or does it need to be "pulldowned" or whatever to 60i? How about for Blu-Ray?
HD DVD requires pulldown info flags to be added, but this doesn't require a recompression. You can also remote those flags from a HD DVD encoding stream to make a Blu-ray stream.

Will HD DVD Forum or BDA ever intend to publish offical specs to answer these types of questions?They have. They're just expensive :).

Bonus question: Will Windows Media Encoder Studio Edition definitely be able to encode HD DVD compliant VC-1 streams for later muxing?Not in the beta, but we announced at NAB that this will be added later this year.

And that is the key point I was trying to make. If you agree with this, then one has to also agree that BD at 18 Mbit/sec, will not be superior to D-VHS on every movie. So we must stop using D-VHS as a reference for BD. They are different animals.

Amir,

I think that's reflected in Sony's choice for it's intitial batch of 25g titles. Notice that more demanding material (Bridge over the River Kwai) is being held back for 50g BD.



Video data reduction (compression) is complex. To do well requires experienced people, monitoring, and analyzing. At this point in time it appears there is ample resources for doing MPEG-2. I can find very little resources for AVC/VC-1 at the professional level.

Despite that fact, WB seems to be doing an outstanding job with the image quality of their first-generation HD DVD authored discs using VC1.

I look forward to their 2nd wave of BD releases using VC1!

Thanks for the information: were you referring to HD-DVD-ROM or
to HD-DVD-VR in the answers you provided to the original post?

Any comment on syntax restrictions in the ROM or recordable formats?

It seems to me that with the computational load of HiDef and H.264 main or
high profile, a recordable format such as BD-RE or HD-DVD-VR can't practically
place limits on the syntax of the elementary streams which go to disk.
(with the exception of stuffing up to a minimum bit rate or introducing some header
or ancillary display or navigation information in the mux.) For example, imposing
a requirement that IDRs occur with certain frequency or some number of slices
per picture would require a full (real-time) decode and re-encode capability.

Also, can anyone comment on whether BD-RE will add H.264 support anytime
soon?

--gern

I haven't talked to our legal about this recently, but the standards agreements are more restrictive than you might imagine. I'm not directly involved, but we certainly have people working to find the right way to share information to the public. You'll especially be seeing this with iHD. But I can answer a few of the below. We'll have a lot more details when we have public tools for VC-1 encoding for HD DVD, obviously.


Yes, High Profile has been made public.

By spec? Normally a minimum datarate is used in MPEG-2 encoders to get around VBR starving easy content, resulting in ringing in a MPAA graphic, for example. VC-1 doesn't have that problem.

VC-1 allows different B-frame patterns for HD DVD.


HD DVD requires pulldown info flags to be added, but this doesn't require a recompression. You can also remote those flags from a HD DVD encoding stream to make a Blu-ray stream.

They have. They're just expensive :).

Not in the beta, but we announced at NAB that this will be added later this year.

Thanks for the information: were you referring to HD-DVD-ROM or
to HD-DVD-VR in the answers you provided to the original post?
Yes, I'm talking about movie discs.

Any comment on syntax restrictions in the ROM or recordable formats?Nope, since I don't know much of anything about the recordable formats.

So, let me flip this around to the community - would would YOU want to see in downloable codec samples?
Ben - I'd love to see some constant average bitrate comparisons at varying bitrates to highlight what each codec does well and where all of them have challenges. I would trust someone like Ron to work up an MPEG-2 sample given the same source file. We will still be short an AVC advocate, but AVC seems to mostly be MIA at this point, anyway.

I would also like it in a thread free of PS3/X360 talk, too, but that may be asking for too much these days. :D

Later,
Bill

Freeze frame on 1080i typically looks bad on modern hi rez fixed pixel displays.
Since I am watching this on a laptop, I am assuming that player converted it into progressive, even at freeze frame. However, not entirely successfully! Parts of the picture clearly showed deinterlacing artifacts, for sure.


Anybody curious about the underlying uncompressed clip might want to view it deinterlaced to 1080p/60 YUY2. If you have Avisynth installed this can be done by downloading one of the uncompressed clips above from the VQEG site and using the following Avisynth script with my TomsMoComp filter to view it, though very likely not at full speed.

I will try it. Thanks a lot!



dr1394, thanks for the clarification. By the way, Nero ShowTime worked great with H.264 trailers (King Kong, Silent Hill, etc), on my laptop, so I think this player is okay (much better than QuickTime7). By the way, even those trailers had problems in areas with smooth gradation, I think, but no other notable artifacts (apart from being a bit soft at times). I will try Tom's suggestion, and compare your encoding with uncompressed stream to get a better picture (no pun intended)! I don't have a hardware player yet. Waiting for the Samsung. Actually, that is one of the reasons I am so keen to understand MPEG2 limitations if any! :)

If not, I can certainly do a nice deinterlace, but I don't know how good a test that would be.

So, let me flip this around to the community - would would YOU want to see in downloable codec samples?Ben, if you could, could you provide both encodings, please? One using original, and one using your deinterlace! :)

And, you can post the uncompressed deinterlaced material here so dr1394 or someone else can have a go at it.

Of course, if you have some other 24p source material that can be published, by all means, do so!! :)

So, after wading through this thread for what seems like weeks now, would I be correct in concluding that for the vast majority of us (without the calibrated 1080p systems), we *might* not notice a difference. Now that is funny.

I haven't talked to our legal about this recently, but the standards agreements are more restrictive than you might imagine. I'm not directly involved, but we certainly have people working to find the right way to share information to the public. You'll especially be seeing this with iHD. But I can answer a few of the below. We'll have a lot more details when we have public tools for VC-1 encoding for HD DVD, obviously.

Thanks for the help! Based on your comments and what I've seen elsewhere, it sounds like HD DVD is less restrictive in terms of encoding. If Blu-Ray forces B-frames where you don't need them, I think it could have a small, but bad, effect on quality.

Based on your comments and what I've seen elsewhere, it sounds like HD DVD is less restrictive in terms of encoding. If Blu-Ray forces B-frames where you don't need them,
Not that Ben will mind your conclusion :D , but I think as far as frames are concerned, VC-1 on HD DVD will be same as on BD (Ben, please correct me if I am wrong. I need to double check). Pulldown flags are HD-DVD specific (as opposed to VC-1 specific), though, as Ben said.

Not that Ben will mind your conclusion :D , but I think as far as frames are concerned, VC-1 on HD DVD will be same as on BD (Ben, please correct me if I am wrong. I need to double check). Pulldown flags are HD-DVD specific (as opposed to VC-1 specific), though, as Ben said.
If you're asking, yes BD is also flexible on B-frames. We're generally able to convert a HD DVD compatible VC-1 bitstream into a BD compatible one with a little bit flipping, but without recompression. Mainly just removing the pulldown flags.

So I've noticed. Are there ally 1080p24/25 sources along these lines we could use?

If not, I can certainly do a nice deinterlace, but I don't know how good a test that would be.

So, let me flip this around to the community - would would YOU want to see in downloable codec samples?

I don't personally know of any good 1080p uncompressed sources that would allow unrestricted distribution. I'd sort of forgotten the VQEG clips were 1080i when I'd first suggested them.

It might be most appropriate to just leave them in 1080i and at least have some samples of how well the new VC1 can handle interlaced HD. After all, there will probably be some sports on HD DVD's eventually. Or some HD Net stuff.

Wasn't Bikini Destinations 1080i? ;)

- Tom

I don't personally know of any good 1080p uncompressed sources that would allow unrestricted distribution. I'd sort of forgotten the VQEG clips were 1080i when I'd first suggested them.

It might be most appropriate to just leave them in 1080i and at least have some samples of how well the new VC1 can handle interlaced HD. After all, there will probably be some sports on HD DVD's eventually. Or some HD Net stuff.

The problem is that real-time 1080i playback on computers is difficult to control, and variable between systems, since you've got real-time deinterlacing added to the mix.

I think I might just be able to reuse the Island trailer again, when the Dreamworks licensing person gets back from vacation. I'd like to take another whack at it - I got the DPX decode off a little originally, so it came out too bright.

If you're asking, yes BD is also flexible on B-frames. We're generally able to convert a HD DVD compatible VC-1 bitstream into a BD compatible one with a little bit flipping, but without recompression. Mainly just removing the pulldown flags.

Thanks. And, yes, regarding pulldown flags that is what I understood. Thanks for confirming.

Regarding 1080i source, I can understand your reluctance, but as Tom mentioned I think it will be a good test for us (viewers) to look at VC-1's capability with interlaced source.

If you are not ready for 1080i yet, it is allright. Please give us something else!

Ben, there are some uncompressed 1080p test clips available here:

f*p://ftp.ldv.e-technik.tu-muenchen.de/pub/test_sequences/1080p/

Ben, there are some uncompressed 1080p test clips available here:

f*p://ftp.ldv.e-technik.tu-muenchen.de/pub/test_sequences/1080p/
Downloading now!

We will still be short an AVC advocate, but AVC seems to mostly be MIA at this point, anyway.Bill, I do not know the status AVC via HD-DVD. But I can tell you that Dish Network, DirecTV and some DVB-S services chose AVC (MPEG-4) to complement their use of MPEG-2.

Downloading now!
Excellent - I think this test is better conducted using 1080p sources.

Excellent - I think this test is better conducted using 1080p sources.
They're not all downloaded, but it looks like about 90 seconds of individual shots. What would be interesting in a test?

Just encode the clips in alphabetical order?
Add some cross dissolves and fades?
Add some scrolling credits?

The only problem with adding effects and such is that I'd need to reencode from the .yuv. But I could do a new .yuv file for opening and closing credits.

Source is 25p, but can reclock easily as there's no audio. Better off with test encodes at 1080p25, or 1080p23.976?

IMO, If you want to go all out then the source material should be from the Sony HDCam SR (http://www.hdtvmagazine.com/articles/2006/02/hd_feature_film.php) series.

Ben, There is a version of DisplayMate (http://www.displaymate.com/infodmmm.html) that has motion that may or may not have any use for codec testing. Unfortunately it requires specific hardware and OS to use the motion section. I have the DisplayMate for Windows version.

IMO, If you want to go all out then the source material should be from the Sony HDCam SR (http://www.hdtvmagazine.com/articles/2006/02/hd_feature_film.php) series.

Ben, There is a version of DisplayMate (http://www.displaymate.com/infodmmm.html) that has motion that may or may not have any use for codec testing. Unfortunately it requires specific hardware and OS to use the motion section. I have the DisplayMate for Windows version.
Getting contnet to test with is easy. The trick is content that is avaiable both to us and to other codec vendors, so everyone can agree it's an open, fair test.

Getting contnet to test with is easy. The trick is content that is avaiable both to us and to other codec vendors, so everyone can agree it's an open, fair test.

Along that line, mayb it would be better not to have too many fades or other special effect since it would make it harder for anyone else to duplicate with other codecs. (unless you also posted the uncompressed)

- Tom

Here is one layman's comments on the HDCam SR material (http://www.avsforum.com/avs-vb/showthread.php?p=7630097&&#post7630097) via Blu-ray Disc.

Bonus question: Will Windows Media Encoder Studio Edition definitely be able to encode HD DVD compliant VC-1 streams for later muxing?

No, the beta version cannot create HD or BD complaint streams.

Here is one layman's comments on the HDCam SR material (http://www.avsforum.com/avs-vb/showthread.php?p=7630097&&#post7630097) via Blu-ray Disc.

Could HDCam SR material be put on HD DVD? Could HDCam SR material be put on HDD?

Ben, there are some uncompressed 1080p test clips available here:

f*p://ftp.ldv.e-technik.tu-muenchen.de/pub/test_sequences/1080p/


Here is an example avisynth script to open these:

RawSource("F:\rush_hour.yuv",1920,1080,"I420")
AddBorders(0,0,0,8)

I had to add 8 pixels to the height to keep my encoder gui from complaining, YMMV.
You can get the rawsource.dll to put in your avisynth plugins folder from http://www.avisynth.org/warpenterprises/

I'm tinkering now with x264 to see what settings are best for rush_hour.yuv. This clip has lots of nice film grain and some "heat wave" blur that should make for a nice challenge. I am encoding at High Profile @ Level 4.1... will report back later.

UPDATE: Wow, this clip is brutal. I get very nice results at 18 Mbps (no ringing, no blocks, lots of detail), but true "nose 6 inches from the screen" transparency seems to take about 32 Mbps! :eek: The difference is preservation of the tiny film grain. Using 1 ref, 3 adaptive b-frames, GOP from 25 to 250, all high profile options, level 4.1, plus Sharktooth's EQM AVC-HR matrix: 2-threaded encoding happens at about 1 fps on my dual core Pentium D 820.

Could HDCam SR material be put on HD DVD? Could HDCam SR material be put on HDD?Sure. And any one of the three codecs could be used. EDIT- I assume by HDD you mean Hard Disk Drives? In the HD-SDI (semi raw form) mode it will use drive space in a hurry. For an end user (consumer) just think of it as another HD source.

Here is an example avisynth script to open these:
UPDATE: Wow, this clip is brutal. I get very nice results at 18 Mbps (no ringing, no blocks, lots of detail), but true "nose 6 inches from the screen" transparency seems to take about 32 Mbps! :eek: The difference is preservation of the tiny film grain. Using 1 ref, 3 adaptive b-frames, GOP from 25 to 250, all high profile options, level 4.1, plus Sharktooth's EQM AVC-HR matrix: 2-threaded encoding happens at about 1 fps on my dual core Pentium D 820.
I really wish I could share the H.264 encoding specs for HD DVD... You might want to download the 30-day trial of Main Concept H.264, and steal the settings it uses for its HD DVD and Blu-ray compatibilty mode. That'll help you get legal GOP sizes and such.

A GOP length of 250 is definitely well beyond what's legal :).

Question for the crowd - do you want to see a test using all the clips together, or each individual clip. I was assuming the former, as that'll make rate control for 2-pass VBR matter a lot more, which is critical for optical discs. I'm rendering out a test right now which is basically all the clips in alphabetical order, at 23.976 fps.

Question for the crowd - do you want to see a test using all the clips together, or each individual clip. I was assuming the former, as that'll make rate control for 2-pass VBR matter a lot more, which is critical for optical discs. I'm rendering out a test right now which is basically all the clips in alphabetical order, at 23.976 fps.

Definitely all in one.

- Tom

Sounds great. Thanks, Ben!

Yes, please!

UPDATE: Wow, this clip is brutal. I get very nice results at 18 Mbps (no ringing, no blocks, lots of detail), but true "nose 6 inches from the screen" transparency seems to take about 32 Mbps! :eek: The difference is preservation of the tiny film grain. Using 1 ref, 3 adaptive b-frames, GOP from 25 to 250, all high profile options, level 4.1, plus Sharktooth's EQM AVC-HR matrix: 2-threaded encoding happens at about 1 fps on my dual core Pentium D 820.
I've got a full encode going, but I'm having a strange problem where some of the frames are coming out in reverse order. Are you seeing that? I'm not sure if it's a problem with the source, or the YUV reader I'm using.

I'm reminded how much YUV annoys me as a format ;). Someone should really convert these into IYUV .AVI files and then zip them up.

Other than that, quality @ 16 Mbps average is really very good throughout, but I don't want to post until I've got all the frames playing in the right order...

Here's the MPEG-2 version:

http://www.w6rz.net/1080p25.zip

30 Mbps peak, 18 Mbps average. M=3, N=15. Actual peak was about 28.5 Mbps. My encoder found the "Rush Hour" clip to be pretty easy. The "Riverbed" clip was definitely the most difficult followed by the "Tractor" clip and the "Blue Sky" clip.

The ftp site was a bit of a PITA. It kept aborting transfers. "Blue Sky" (217 of 250) and "Tractor" (690 of 761) are a few frames short.

http://img253.imageshack.us/img253/7323/1080pbits3rn.jpg (http://imageshack.us)

http://img335.imageshack.us/img335/7537/1080pq6et.jpg (http://imageshack.us)

http://img288.imageshack.us/img288/5858/1080pbps7ni.jpg (http://imageshack.us)

Ron

This thread is kind of exciting, which I didn't expect.

Go to the following thread for a sample encoded at 18 Mbps VBR MPEG-2:

http://www.avsforum.com/avs-vb/showthread.php?p=7675419#post7675419

screencap:
http://img323.imageshack.us/img323/9013/elephantsdream6br.th.jpg (http://img323.imageshack.us/my.php?image=elephantsdream6br.jpg)


Kai

Go to the following thread for a sample encoded at 18 Mbps VBR MPEG-2:



tkx kai,

Is there a high bit-rate VC-1 version for download we can compare it to?

I think Real shot will be more relevant then CGI for comparision sake. I just finished downloading Ron encoding. I can say that digital artifact are minimal. Yet, the shots lacks details. I went into d/l original uncompressed and noticed the problem is from the source. tkx ron for your efforts. looking forward for vc-1 version

Yet, the shots lacks details. I went into d/l original uncompressed and noticed the problem is from the source. tkx ron for your efforts. looking forward for vc-1 version
I agree. The "Pedestrian Area" and "Rush Hour" clips were kinda greyish and hazy. The "Tractor" clip seemed to lack resolution on the tilled ground that the tractor was traversing. I guess the "Sunflower" clip was meant to the resolution clip, but there's a bit too much camera movement.

Ron

tkx kai,

Is there a high bit-rate VC-1 version for download we can compare it to?

I think Real shot will be more relevant then CGI for comparision sake. I just finished downloading Ron encoding. I can say that digital artifact are minimal. Yet, the shots lacks details. I went into d/l original uncompressed and noticed the problem is from the source. tkx ron for your efforts. looking forward for vc-1 version

not unless Ben can encode one for us :)

Here is a lossless screenshot from rush_hour showing the original (on the left) compared to 18 Mbps VBR H.264. The center of each frame has been cropped, but NOT resized or screwed with in any way. First screenshot is of I-frame, next is of the following B-frame. Looks good, but film grain is rubbed out of B-frame. I will try to tweak x264 later to get better results.

http://www.electronicfiles.net/files/1813/rush_hour_I.png

http://www.electronicfiles.net/files/1813/rush_hour_B.png

not unless Ben can encode one for us :)
I'm working at it, but I'm getting weird frame reversals in the .YUV file. Is no one else seeing that?

Wow. Awesome!

I'm working at it, but I'm getting weird frame reversals in the .YUV file. Is no one else seeing that?

No, I just watched the original .yuv file (via my avisynth script) very closely a few times to make sure. It stutters a couple times, probably because my hard drive isn't fast enough, but it never jumps back.

I really wish I could share the H.264 encoding specs for HD DVD... You might want to download the 30-day trial of Main Concept H.264, and steal the settings it uses for its HD DVD and Blu-ray compatibilty mode. That'll help you get legal GOP sizes and such.

A GOP length of 250 is definitely well beyond what's legal :).

Hi Ben, I installed the Mainconcept encoder briefly and checked out the presets for HD DVD and Blu-ray. The default GOP length is "33" for both, a strange number. This seems quite small! I also looked at the docs for Apple's HD DVD encoding app: they show .5 seconds as the minimum and 5 seconds as the maximum. That would be a range 12 frames to 120 frames for film content. Is a range of 12 to 120 kosher in your opinion?

lazyn00b,
film grain is also mostly removed in the I-frame. This lead to a loss of detail as well
That's not a good, because it makes everything look flat

Kai

No, I just watched the original .yuv file (via my avisynth script) very closely a few times to make sure. It stutters a couple times, probably because my hard drive isn't fast enough, but it never jumps back.
Bummer (well, I'm glad it's working for you).

I should re-download - maybe there was a glitch here? I've also asked the Studio Edition devs to download it themselves and see what happens.

Hi Ben, I installed the Mainconcept encoder briefly and checked out the presets for HD DVD and Blu-ray. The default GOP length is "33" for both, a strange number. This seems quite small! I also looked at the docs for Apple's HD DVD encoding app: they show .5 seconds as the minimum and 5 seconds as the maximum. That would be a range 12 frames to 120 frames for film content. Is a range of 12 to 120 kosher in your opinion?
120 is waaay too high.

DVD was 0.5 sec, if I might offer a strong hint :). Since the real issue is memory utilization, GOP length for hardware systems is typcially definced by time instead of frame count.

I just did some encoding tests with the Mainconcept MPEG encoder (version 1.5.1), and that is the worst encoder I have seen in quite some time. Slow as he** too.

Kai

I just did some encoding tests with the Mainconcept MPEG encoder (version 1.5.1), and that is the worst encoder I have seen in quite some time. Slow as he** too.

MPEG-2? I can think of some much slower, and much worse, but yes, it's not a top tier encoder at this point.

Yeah, MPEG-2.

I've been doing some test encodes with ExpertHD, Procoder and Mainconcept among others, and Procoder was the clear winner quality-wise.

As for the speed: I mixed that up with ExpertHD :)

Everything was done on the same machine:
ProCoder was 6.2 times real time, MainConcept was 6.6 time RT, ExpertHD was north or 20 times RT :)

Kai

MPEG-2? I can think of some much slower, and much worse, but yes, it's not a top tier encoder at this point.
What would be a top-tier, consumer available encoder? Or are those two things incompatible?

Yeah, MPEG-2.

I've been doing some test encodes with ExpertHD, Procoder and Mainconcept among others, and Procoder was the clear winner quality-wise.

As for the speed: I mixed that up with ExpertHD :)

Everything was done on the same machine:
ProCoder was 6.2 times real time, MainConcept was 6.6 time RT, ExpertHD was north or 20 times RT :)

ProCoder pretty much rocks. For professional use, the $5000 version of it called Carbon (from Rhozet) is even more awesome for transcoding and integration. FYI, they were demoing their VC-1 support at NAB.

I consulted on the design of that product for a number of years, before I joined Microsoft.

Go to the following thread for a sample encoded at 18 Mbps VBR MPEG-2:

http://www.avsforum.com/avs-vb/showthread.php?p=7675419#post7675419



Am I the only one seeing motion being not smooth here. Tried this at 50Hz.

nataraj: that's due to them not being experienced with real film material.

Some of their pans and "camera" movements are simply too fast for 24 fps causing strobing effects.

Here is some excellent info on strobing that ought to have been a must-read for the animators/compositors of Elephants Dream.

And btw: the movie is 24.000 fps, so try to play it back at 24, 48 or 72 Hz

Kai

And btw: the movie is 24.000 fps, so try to play it back at 24, 48 or 72 Hz

I think I tried at 60, 48 and 50. But I'll try again.

Here's the MPEG-2 version:

http://www.w6rz.net/1080p25.zip

OK, so I was able to improve the preservation of film grain on rush_hour by using the following command line for x264 (1st & 2nd pass at 18 Mbps VBR):

--pass 1 --bitrate 18000 --stats "D:\TMP\rush_hour.stats" --level 4.1 --keyint 50 --min-keyint 12 --bframes 3 --b-pyramid --filter -2,-1 --subme 1 --analyse none --vbv-maxrate 28000 --me dia --threads 2 --thread-input --cqmfile "C:\eqm_avc_hr.cfg" --progress --no-psnr --output NUL "D:\TMP\rush_hour.avs"

--pass 2 --bitrate 18000 --stats "D:\TMP\rush_hour.stats" --level 4.1 --keyint 50 --min-keyint 12 --ref 3 --mixed-refs --no-fast-pskip --bframes 3 --b-pyramid --b-rdo --bime --weightb --filter -2,-1 --subme 6 --trellis 1 --analyse p8x8,b8x8,i4x4,i8x8 --8x8dct --vbv-maxrate 28000 --threads 2 --thread-input --cqmfile "C:\eqm_avc_hr.cfg" --progress --no-psnr --output "D:\TMP\rush_hour_vbr18000k_2.mp4" "D:\TMP\rush_hour.avs"

GOP size was from 0.5 to 2 seconds, max bitrate was set to 28 Mbps. Notice that I lowered the loop filter strength to -2 and the threshold to -1: this seemed to help a lot with the film grain. I feel like I've made all the tweaks I can without substancially worsening the time to encode or creating new problems; the second pass is already 0.75 fps!

Then I tried the same clip with the new WMV AP (VC-1) codec that was just released with WMP 11. I used 2 pass VBR at 18 Mbps constrained to 28 Mbps max bitrate, with 2 seconds between keyframes and "Compression Optimization Type = 1". Second pass was also dog slow with this codec, taking several minutes.

The verdict? Both codecs performed really well - not frame by frame transparency by any stretch, but good enough so that if you were watching the encoded clips in real-time you would not think anything was amiss. After watching both clips over and over... I have to say I think the WMV AP (VC-1) clip is slightly superior (with a lot less tweaking, too). Amir will be pleased after the scorn I heaped on Microsoft for not releasing this codec sooner ;)

If anybody wants to see the two clips I encoded, I'd be happy to upload them to someone's server.

Then I tried the same clip with the new WMV AP (VC-1) codec that was just released with WMP 11. I used 2 pass VBR at 18 Mbps constrained to 28 Mbps max bitrate, with 2 seconds between keyframes and "Compression Optimization Type = 1". Second pass was also dog slow with this codec, taking several minutes.

The verdict? Both codecs performed really well - not frame by frame transparency by any stretch, but good enough so that if you were watching the encoded clips in real-time you would not think anything was amiss. After watching both clips over and over... I have to say I think the WMV AP (VC-1) clip is slightly superior (with a lot less tweaking, too). Amir will be pleased after the scorn I heaped on Microsoft for not releasing this codec sooner ;)

Whoo! I can't tell you how glad I am for the codec to be in the hands of people who don't need to get a signoff from legal to release content. I know it's been a lot of me asserting stuff around here, but the proof is in the pudding.

One setting that Compression Optimization Type=1 doesnt' turn on that we use for HD DVD is integer chroma search. You do that via:

Motion Search Level=-2

There's a great thread talking about tweaking the codec over at Doom9:

http://forum.doom9.org/showthread.php?t=111275

Whoo! I can't tell you how glad I am for the codec to be in the hands of people who don't need to get a signoff from legal to release content. I know it's been a lot of me asserting stuff around here, but the proof is in the pudding.

One setting that Compression Optimization Type=1 doesnt' turn on that we use for HD DVD is integer chroma search. You do that via:

Motion Search Level=-2

There's a great thread talking about tweaking the codec over at Doom9:

http://forum.doom9.org/showthread.php?t=111275

Please tell me you don't have to go through that kind of "codec navel gazing" to get VC-1 HD-DVD discs encoded do you? :eek:

I though the computers were "smart" and figured out all of the hard stuff for you... ;)

b2b

Please tell me you don't have to go through that kind of "codec navel gazing" to get VC-1 HD-DVD discs encoded do you? :eek:

I though the computers were "smart" and figured out all of the hard stuff for you... ;)

What do you think that segment based re-encoding looks like :)?

What do you think that segment based re-encoding looks like :)?

Well I really don't know. I would think that some sort of automated grading system would compare source vs encode and be able to adjust encoding parameters as needed. Otherwise you will never have enough talent to be able to cover the sheer volume of content that needs to get encoded. I would guess that the real goal of a encoding suite is to leave as little to hand tweaking as possible, simply because of cost constraints. Otherwise it will all be outsourced to India and we all know what that means. ;)

b2b

I would think that some sort of automated grading system would compare source vs encode and be able to adjust encoding parameters as needed.

b2b

I don't think there's any way around manual tweaking. The finished content is always intended for humans, and it will be the human eye that judges. People do not see PSNR, they just see artifacts. If you could design software that is smart enough to see what we see and make the appropriate judgments, then look out, because that software is probably smart enough to re-task a few Stealth bombers and take us all out! :D

In the mean time, segment based re-encoding sound pretty good.

I have asked this question a few times and never seen a response. Can either benwaggoner or sspears tell us what display is used when professionally encoding HD films? What is the size, resolution etc? I have often wondered (especially on DVD - are you listening Miramax?) how a compressionist can sign off on a finished project with the results I sometimes see. I can't imagine you guys (or whoever is doing this next gen work) sitting in a darkened room with a ruby projector trying to encode video, so it must be a pro monitor. Can it resolve 1080 clearly and do they sit close enough to actually see it? Is there an eye exam requirement for this job? (just kidding on the last part)

Well I really don't know. I would think that some sort of automated grading system would compare source vs encode and be able to adjust encoding parameters as needed. Otherwise you will never have enough talent to be able to cover the sheer volume of content that needs to get encoded. I would guess that the real goal of a encoding suite is to leave as little to hand tweaking as possible, simply because of cost constraints. Otherwise it will all be outsourced to India and we all know what that means. ;)

Well, it's like DVD. The A list titles all get segment reencoding today. But the 99% of titles that get made (C-list titles, home movies) don't.

You can get a perfectly good-enough video with normal 2-pass encoding. But top-flight work takes segment reencoding today.

As codecs develop (and we've certainly seen this with ours in the last few months) the number of segments needed per movie go down.

I have asked this question a few times and never seen a response. Can either benwaggoner or sspears tell us what display is used when professionally encoding HD films? What is the size, resolution etc? I have often wondered (especially on DVD - are you listening Miramax?) how a compressionist can sign off on a finished project with the results I sometimes see. I can't imagine you guys (or whoever is doing this next gen work) sitting in a darkened room with a ruby projector trying to encode video, so it must be a pro monitor. Can it resolve 1080 clearly and do they sit close enough to actually see it? Is there an eye exam requirement for this job? (just kidding on the last part)


I sure hope they are still using CRT-based displays. Otherwise there would be zero precision in darker scenes.

I sure hope they are still using CRT-based displays. Otherwise there would be zero precision in darker scenes.

I have seen Stacey post that he has sony XBR960 in his office....

I have seen Stacey post that he has sony XBR960 in his office....

I have been wondering while all the HD-DVD titles that I have seen (11) have EE. For some it's very rare (Serenity has only a 2 minute scene with EE), but for others, it's endemic.

If they use a small CRT set to verify the work, that's understandable. Only a 2K FP system (like a Ruby or a Qualia) would allow perfection in that aspect.

I have seen Stacey post that he has sony XBR960 in his office....

Thats just to X-box with when things are slow (;

I have been wondering while all the HD-DVD titles that I have seen (11) have EE. For some it's very rare (Serenity has only a 2 minute scene with EE), but for others, it's endemic.

If they use a small CRT set to verify the work, that's understandable. Only a 2K FP system (like a Ruby or a Qualia) would allow perfection in that aspect.

Is it EE or just a compression artifact? Is it in the master?

Ben or stacey, do you know if any of the post-houses are putting any EE into their encodes?

Is it EE or just a compression artifact? Is it in the master?

Ben or stacey, do you know if any of the post-houses are putting any EE into their encodes?

It's probably not intentional. Ringing might a more accurate term, but EE has become a common term.

It might even be a telecine artifact. It would be interesting to have a definitive answer about the source of this problem.

Just to clarify, between us and post houses/studios, every title gets watched on every type of display known to man. We have professional CRTs, 1080p panels, 720 panels, multiple ruby and CRT projectors, DLP projectors, you name it, we have it :). And QC occurs primarily at the post houses, not Stacey's office :D. We do the double checking here but not the initial quality loop.

Amir

Just to clarify, between us and post houses/studios, every title gets watched on every type of display known to man. We have professional CRTs, 1080p panels, 720 panels, multiple ruby and CRT projectors, DLP projectors, you name it, we have it :). And QC occurs primarily at the post houses, not Stacey's office :D. We do the double checking here but not the initial quality loop.

Amir

So, at what stage is the EE/ringing introduced?

So, at what stage is the EE/ringing introduced?
I don't know that there is any :).

At some level though, one needs to understand that you are not looking at the original print. As soon as the program is transfered to D-5 master, mild JPEG-like compression is applied to it. That removes high-frequency information which as anyone knows from basic signal processing, is going to cause ringing around high frequency edges. Then there is any sharpening algorithm applied at telecine stage. And finally, anything other than lossless compression, is going to introduce some amount of filtering.

So there is no perfection here. I personally have not seen any ringing in normal viewing on the Ruby.

Amir

I don't know that there is any :).

At some level though, one needs to understand that you are not looking at the original print. As soon as the program is transfered to D-5 master, mild JPEG-like compression is applied to it. That removes high-frequency information which as anyone knows from basic signal processing, is going to cause ringing around high frequency edges. Then there is any sharpening algorithm applied at telecine stage. And finally, anything other than lossless compression, is going to introduce some amount of filtering.

So there is no perfection here. I personally have not seen any ringing in normal viewing on the Ruby.

Amir

Are there any interest in using 2k or 4k scans? That would eliminate the problem with D5 compression.

b2b

Are there any interest in using 2k or 4k scans? That would eliminate the problem with D5 compression.

b2b
Most definitely. Joe Kane was talking about a study at the Home Theater Cruise that showed 4K scans improve the picture quality even with DVD! No doubt, downsampling allows somoe noise removal.

And of course, not going to D-5 improves things yet again.

Needless to say, the cleaner the pipeline, the lower the data rate needed to compress it well. Compression artifacts/ringing at earlier stages make things a lot harder. We are seeing excellent quality now at just 10 Mbit/sec VC-1 on better mastered content (although they will probably release at higher data rates because "they can").

Amir

Looking at the imminent PS3 I would say that Sony is definitely going for quantity over quality.

The PS3 just ain't no ES series product. It is about as cheap as it can get. And Sony does that to sell huge volumes (at least that's what they hope).

So there you have your answer.


Umm, your point here is? the PS3 will decode Blu-ray titles via software compared to a standalone hd-dvd/blu-ray player.
you'd think this is a bad thing, but in fact it's quite the oposite, the Cell has so much raw power that decoding a 48 mbit h264 stream should be no problem whatsoever, with room to spare. hell, i wouldn't be at all surprised if it could decode multiple streams at the same time.

people seem to forget a few things about consoles, they are products which are sold at a huge loss when they first come out, and these companies make their money by selling software.
this is one of those few products where competition isn't as beneficial for consumers as in about any other product.

so in summary, comparing a PS3 to a blu-ray or HD-DVD player is like trying to compare a race car to a horse, the technology used in the PS3 _FAR_ exceeds what is needed to simply decode a video encoded in any of the current codecs.

so is a PS3 at 600 bucks cheap? well, not really. but when you realize what you're getting for those 600 it's actually a pretty damned good deal.

Umm, your point here is? the PS3 will decode Blu-ray titles via software compared to a standalone hd-dvd/blu-ray player.
you'd think this is a bad thing, but in fact it's quite the oposite, the Cell has so much raw power that decoding a 48 mbit h264 stream should be no problem whatsoever, with room to spare. hell, i wouldn't be at all surprised if it could decode multiple streams at the same time.

people seem to forget a few things about consoles, they are products which are sold at a huge loss when they first come out, and these companies make their money by selling software.
this is one of those few products where competition isn't as beneficial for consumers as in about any other product.

so in summary, comparing a PS3 to a blu-ray or HD-DVD player is like trying to compare a race car to a horse, the technology used in the PS3 _FAR_ exceeds what is needed to simply decode a video encoded in any of the current codecs.

so is a PS3 at 600 bucks cheap? well, not really. but when you realize what you're getting for those 600 it's actually a pretty damned good deal.

So what?

Just buy it and be happy :)

These last two posts seem out of context on this thread. Just to put things in perspective, I haven't played a video game since Asteroids.

Ron

Thanks for the comments Amir, but it still doesn't answer my question. I just wonder what the person actually doing the encode is looking at, and if it is of sufficient size/resolution. I know it depends on which studio etc. but surely someone must be able to answer.
Also, when you said what your (MS) people ("we", "our") were watching on, is that just to help with the VC-1 development? What other reason would MS be involved with pre-release HD DVD product? How are the warner VC-1 BD encodes coming along? Are they able to use the same encodes for both formats?

I just wonder what the person actually doing the encode is looking at, and if it is of sufficient size/resolution. I know it depends on which studio etc. but surely someone must be able to answer.
The resolution really doesn't change as part of compression (with VC-1 anyway). What they are looking for are compression artifacts. You know, what you see in HD broadcast when the camera pans. See more below.

Also, when you said what your (MS) people ("we", "our") were watching on, is that just to help with the VC-1 development?
The post houses (on behalf of studios) send us "check discs." These are near final discs that are getting ready for final replication. We play these on a range of displays to see if we see any issues with them beyond the critical eyes of the experts at the post houses. So "we" here are codec engineers, and others involved in this process.

What other reason would MS be involved with pre-release HD DVD product? How are the warner VC-1 BD encodes coming along? Are they able to use the same encodes for both formats?
Our interest here is to make sure our codec and HD DVD look the best. As a proponent of HD DVD, we know this is super important for the format.

On BD and VC-1, unfortunately, I can not comment at all. I am sorry about that.

Amir

Interesting Amir... you haven't had the opportunity to see VC1 on BD? Would the WB guys be able to give you a sneak-peak "unofficially"?

Interesting Amir... you haven't had the opportunity to see VC1 on BD? Would the WB guys be able to give you a sneak-peak "unofficially"?
That is not what I said :). I simply said that I can not comment about anything on this topic.

Amir

The post houses (on behalf of studios) send us "check discs." These are near final discs that are getting ready for final replication. We play these on a range of displays to see if we see any issues with them beyond the critical eyes of the experts at the post houses. So "we" here are codec engineers, and others involved in this process.

And despite this wide range of displays, you told us that you don't see any ringing on the released HD-DVDs.

Everybody I know personally sees ringing on titles like Swordfish. I'm quite surprised that compressionist experts can be so insensitive such an ugly artifact.

Thanks for the prompt replies Amir, I value your insight into these matters, whether I like the news or not.

I hate NDA's :p

Everybody I know personally sees ringing on titles like Swordfish. I'm quite surprised that compressionist experts can be so insensitive such an ugly artifact.
It is very hard for me to tell what you are seeing there and whether it is due to the display or the source. Here is the latest review on HD DVD from Home Theater Magazine with specific comments in this regard:

http://www.hometheatermag.com/discplayers/0506toshiba/

“The biggest improvement (to me) is the freedom from MPEG artifacts such as edge peakyness (white lines around objects) and mosquito noise (a wispy video noise). The removal of MPEG artifacts should be apparent on any screen size (as long as you are sitting close enough). Every disc I sampled was “clean;” the only “noise” one would see is the actual film grain, just like in a movie theater. The amount of grain would vary greatly with each film. Serenity looked spectacular, with a very fine grain structure along with vivid colors. The cleanliness of the presentation is something I have never seen before in a home setting. If you think over-the-air HD is real high definition, you ain’t seen nothing yet.”

Fact is that review after review, says these titles are perfect or near perfect in picture quality. Take a look at Widescreen Review magazine's review of HD DVD titles. We both know that they complain very quickly about ringing yet there is not a complaint in sight, with most of the titles getting a score of 5 out of 5 on picture quality. And Gary Reber told me that this is a new scale for HD titles.

Amir

It is very hard for me to tell what you are seeing there and whether it is due to the display or the source. Here is the latest review on HD DVD from Home Theater Magazine with specific comments in this regard:


You know, these reviewers are either clueless, or completely biased.

WSR is infamous for their glorious review of the first Highlander DVD, which is one of the worst ever DVDs. Their lame excuse is that it was good for the standard of the times, which is completely false. I would bet that they did not have the time to review the disk before the issue deadline, and just wrote anything about it.

They did it again with D-Theater, glorifying "Meet Joe Black", which was sourced from very bad material and look barely better than DVD. MJB is the worst D-Theater tape, and WSR ranks it highly.

They are doing it again, writing that Apollo 13 is better (or close) to Serenity, while there is a world of difference between the two.

They don't mention once the visible edge enhancement on most HD-DVDs. I see it on my Qualia, and friends see it on their Rubies, Qualias and CRTs.

EE (or ringing) manifests by a halo around objects against a white background. Just look at scenes with a bright background all over Swordfish. Or at the scene in the cemetery near the end of Serenity. Serenity is very good on that aspect, there is EE only for 2 minutes, but since it's a very specific scene, it's easy to spot.

Edit: to clarify my point of view about HD-DVD, let me say that I love the format. I love the lack of blocking artifacts we see so often on broadcast HD, I love the prospect of lossless audio, and I love the prospect of day and date releases with DVD. But EE (or ringing) is an artifact that must also go away, and so far, it has not. Denying it's here won't help.

Ringing is different than edge enhancement, though ringing appears on high frequency edges when compression gets to be too high.

Check your sharpness settings.

robena, I have not seen any ee or ringing whatsoever. I will take a closer look at the particular scenes you referenced with a few of my HD DVD systems.

-Robert

Ringing is different than edge enhancement, though ringing appears on high frequency edges when compression gets to be too high.

Yes, but they look similar. I use both words here, for the lack of knowing the real culprit.

Check your sharpness settings.

These artifacts are visible when sharpness is set to 0. They are on the discs.

“The biggest improvement (to me) is the freedom from MPEG artifacts such as edge peakyness (white lines around objects) and mosquito noise (a wispy video noise).

White lines around edges are NOT an MPEG artifact (mosquito noise is). If white lines appear after MPEG compression and are not there before it's because inside the MPEG encoder sharpening is applied which is not part of the actual MPEG encoding. DCT transforms and frequency coefficient manipulations can cause ringing but not white/black outlines.

Hi,

I have made the same experience Robert describes with Swordfish.
Of course we do not know what the master for the HD-DVD looked like but one still would hope that EE is soon to be a thing of the past. It is the most important artefact that still lessens our enjoyment of many movies and now that compression artefacts are much less of an issue than before I hope this is the next thing to go.

For those that are not so annoyed by edge enhancement/ringing here is a guide by my good friend Bjoern Roy:

Bjoern on EE (http://www.videophile.info/Guide_EE/Page_01.htm)

I would describe the EE visible on Swordfish as being of the D type on Bjoern's page (sixth picture from the top). As he says it is a kind of enhancement that results in a picture that many still think of as pleasing. The nearer to the screen you are however the more annoying it gets to see those artefacts and I would think that most viewers that are at 1.5 screen widths away or even less will not be amused.

Maybe we should really get to the root of the problem and demand that the people who are doing the D5 master or the actual transfer are required to have a good eye for these artefacts and also to do their work sitting very close to a big screen that is driven by a projector with 1920 x 1080 resolution, how about 7 ft away from a 7 ft wide screen as a minimum requirement ?

Oh well, one can always dream :)

And this is a heads up to Amir:

You seem to be really in this with a lot of enthusiasm and a desire to make things right, so please do make sure that the people who are responsible for the HD-DVD mastering are able to see what is wrong with Swordfish. If they don't the quality of HD-DVD releases will be hit and miss with regard to EE. The least thing you could do would be to make sure that for now the HD-DVD equals the D5-master you have to use, later on I would hope that the masters itself will be improved, too. And if nobody can see what is wrong get the DVDs out to enthusiasts that you will find in abundance on this forum, choose one that is very picky and has some time on his hands and let him help you with the quality control.

Oliver

Advice taken Oliver :).

Amir

That's great to hear coming from you Amir :) We now have the option for the first time in HT history to get it all right so I am very passionate about this. HD-DVD is the best format there is and it is worth the effort to make it even better.

Amir,

While I believe that the first batch of HD DVD titles represents a new era of high-fidelity 1080P images for home-viewing, I don't put my full faith into widescreen review or any other on-line/print review source. Having reviewed standard-definition DVDs for over 2 years now, I'm shocked how many times I find flaws in DVD mastering, when the same discs get praised as "perfect" by these other reviewers.

Again...not assuming that the HD DVDs aren't in fact superlative... but all reviews must be considered in context.


White lines around edges are NOT an MPEG artifact (mosquito noise is). If white lines appear after MPEG compression and are not there before it's because inside the MPEG encoder sharpening is applied which is not part of the actual MPEG encoding. DCT transforms and frequency coefficient manipulations can cause ringing but not white/black outlines.

Quite true. EXACTLY right. Mosquito-noise "crawlie" ringing is very different from the clean haloing seen when HF transitions are boosted... which is common practice prior to compression after HF information is first filtered to reduce entropy (why some assume it's an MPEG artifact... because it's often applied *during* the compression process in the encoder just prior to actual compression). I can imagine non-videophile compressists using the same general practice for VC1 unless someone managed to curb them of this tendency.

Here we go, Swordifsh EE, timestamp 46:40. My camera is very low quality, but that's enough to see clearly the problem:

http://xs101.xs.to/xs101/06222/EE_Swordfish.JPG (http://xs.to)

Robert,

thanks so much for that screenshot, that looks very much like what I see on my system with all enhancement options turned off.

Let's hope that some of the people who are responsible for this will see it too.

Oliver

Advice taken Oliver :).

Amir

Amir, I know you can't change the quality of the master tapes and a lot of them have issues and they will show up on HD-DVD and Blue Ray. But maybe you can point the issues out to people in charge so new masters don't have them anymore? The 2 biggest issues I see all the time are
- frequency boosting, leading to ringing and/or halos around high contrast edges. This is very unfilmlike, a misrepresentation of the original film element and quite obvious on high quality displays from a distance that allows one to see all 1080p (or whatever) detail. It's very widespread and the main reason DVDs look so videolike and are mostly such a disappointment when you seek out the look of film, or at least a look that does not scream out: digital, artifical, unnatural.
- digital noise/grain filtering messing with textures in motion, especially human skin, creating a spatiotemporal mess that looks very distracting and unnatural once you know what to look for. Whole films have been destroyed that way on the digital intermediate. These cases are hopeless. No solution for the HD disk. But if the master has no such filtering artifacts then PLEASE keep the film grain as on the master and DO NOT do any digital grain reduction unless you have the highest quality software and know exactly what you do. Top directors have signed off very shoddy work here and it's a heart break to see what they put up with for whatever reasons.

Hi, guys.
My thoughts about possibility max quality for HD-DVD depend from source.
For better result you need get negative film and scan to dpx or cineon and
then after restoration, color correction encode to wmv or h264.
When you have already compressed source on tape - you have average result.
Now when directors have cameras with ability shooting to digital RAW we can
get max quality. Just look at samples from SI1920-HDVR camera, which
capture from CMOS sensor straight to CineformRAW.
http://indiefilmlive.blogspot.com/

P.S. Remember about GiGo (garbage in - garbage out) :)
Any codec wmv or avc at 15-20Mbps with a good source
will be nice.

SI-1920HDVR SAMPLE FOOTAGE GALLERY
http://www.siliconimaging.com/DigitalCinema/gallery_footage.html

Robert, this is the same frame (hopefully :) ) from my HD cap.

http://img115.imageshack.us/img115/3018/swordfish0fc.jpg
a losslessly compressed version of that screencap can be downloaded here: http://rapidshare.de/files/21759715/swordfish.png.html

The file was captured from a properly IVTC'd stream, so it is a full 1080p frame.


I have made one observation:

When I use the overlay renderer, I get a result very similar to yours, but with a VMR9 renderer I get what's actually in the stream.

Now, when I take the screencap I posted into Photoshop and apply the smart sharpen filter to remove gaussian blur, I get results similar to what the overlay
renderer produces, and thus similar to what your Ruby shows.

See here:
http://img118.imageshack.us/img118/941/swordfishprocessed3fp.jpg
a losslessly compressed version of that screencap can be downloaded here: http://rapidshare.de/files/21763333/swordfish.processed.png.html

So, what could the reason for that be?

The following possibilities come to mind:

1 - the telecine transfer already used a sharpening filter
2 - the master D5 tape had sharpening applied to in post processing
3 - sharpening was applied when converting the master to VC-1
4 - the decoder in the Toshiba applies sharpening (it's a broadcom chip iirc, also to be used in some BD players). This would be similar to what I can reproduce with the overlay renderer.
5 - your Ruby, albeit set to sharpen=0 still does processing that applies a sharpening filter

I don't think you could rule out the Ruby by displaying my screencap on it, but it might nevertheless be worth a try (though I doubt it uses the same processing
for still frames).

Hope that helps,

Kai

Hi Kai,

Robert, this is the same frame (hopefully :) ) from my HD cap.

It is!

That's a good idea to compare the HD-DVD to the (horribly cropped) HBO version. I had completely forgotten that I had it, so after seing your post, I took a peek at it.

The EE (let's call it that way even if it might be technically something else) is indeed very similar to the one on the HD-DVD.

I see it with different viewers on my PC, I tried VideoRedo and Womble's Wizzard...



1 - the telecine transfer already used a sharpening filter
2 - the master D5 tape had sharpening applied to in post processing
3 - sharpening was applied when converting the master to VC-1
4 - the decoder in the Toshiba applies sharpening (it's a broadcom chip iirc, also to be used in some BD players). This would be similar to what I can reproduce with the overlay renderer.
5 - your Ruby, albeit set to sharpen=0 still does processing that applies a sharpening filter

I don't think you could rule out the Ruby by displaying my screencap on it, but it might nevertheless be worth a try (though I doubt it uses the same processing
for still frames).

... so this artifact is in the stream. I don't think my Qualia is the culprit, since I see that on my PC directly, and Olivers sees it on his FP CRT.

Most likely, this is due to the telecine transfer.

Still, this shows that reviewers are blind (these halos are all over Swordfish, every time there is a white background), and telecine operators and compressionists need to be educated if HD-DVD wants to hold true to its best possible image promise.

I'll be watching the first BR transfers with an hawk eye. It will interesting to see if Sony has been paying more attention to EE or not.


Hope that helps,
It did, thanks.

Edit: The sharpening you applied makes the EE much more visible, but it does not create it I think. It's visible on the original frame, enough to be annoying on a large screen as Oliver pointed out.

I'll be watching the first BR transfers with an hawk eye. It will interesting to see if Sony has been paying more attention to EE or not.
.

If the existing Sony HD transfers as seen on TV are any indication you will see plenty of EE. Maybe some CGI titles will have none if the MPEG encoder now has none.

Robert,

The "EE" might also be some sort of a halation created from either backlighting the objects, or from applying shine or glare filters in post production.

Also, I wouldn't rule out the Ruby's processing too fast. I can get a "halo"-free presentation of this clip on the PC using a CRT display and CRT PJ, but I can also use the overlay renderer that produce these halos.

That means:
VMR9: no distracting halos
Overlay: distracting halos (picture looks artificially sharpened, so I assume the renderer applies sharpening, as the decoder is the same as used for VMR9)






Kai

Great work guys. This is what AVS is all about. What a refreshing read after months of BD/HD DVD crap!


Amir, I know you can't change the quality of the master tapes and a lot of them have issues and they will show up on HD-DVD and Blue Ray. But maybe you can point the issues out to people in charge so new masters don't have them anymore? The 2 biggest issues I see all the time are
- frequency boosting, leading to ringing and/or halos around high contrast edges. This is very unfilmlike, a misrepresentation of the original film element and quite obvious on high quality displays from a distance that allows one to see all 1080p (or whatever) detail. It's very widespread and the main reason DVDs look so videolike and are mostly such a disappointment when you seek out the look of film, or at least a look that does not scream out: digital, artifical, unnatural.
- digital noise/grain filtering messing with textures in motion, especially human skin, creating a spatiotemporal mess that looks very distracting and unnatural once you know what to look for. Whole films have been destroyed that way on the digital intermediate. These cases are hopeless. No solution for the HD disk. But if the master has no such filtering artifacts then PLEASE keep the film grain as on the master and DO NOT do any digital grain reduction unless you have the highest quality software and know exactly what you do. Top directors have signed off very shoddy work here and it's a heart break to see what they put up with for whatever reasons.

Exactly.

BTW, Kai and others,

it's very possible that the ringing is in the original master. I know when Lawrence of Arabia had it's digital transfer to 1080P several years ago, the ringing was introduced during the telecine process!!! There was no way to get rid of it for the Super-bit DVD or HD demos. Sony would have to do a new film-digital transfer if they wanted to produce a proper, true-fidelity capture.

I think that many technicians actually *like* the look of sharpened video...and just do it as a matter of course.

That's our real problem!!!!!

Great work guys. This is what AVS is all about.



It is a good read. Why Serenity only has 2 minutes of ringing?

It may just be my browser or LCD display here at the office but the above 1080p pictures also appear to me to have a mild chroma shift.

- Tom

It is a good read. Why Serenity only has 2 minutes of ringing?

Because they did an excellent job at it, with just a minor slip-up!

I think that many technicians actually *like* the look of sharpened video...and just do it as a matter of course.

That's our real problem!!!!!

Thats really sad. I wish they do such things on SD DVD Full Screen ;)

This is a similar problem to music where a lot of people make it "hot" so that it would sound "good" on a boombox / car.

Kai,

I'd say that the HD-DVD looks worse than your screenshot.
I cannot compare it right now as I do not have my HD-DVD player at the moment but I am pretty sure what I saw looked more like the screenshot posted by Robert.

FWIW: I use the HDLeeza as a scaler or in this case linedoubler and everybody who knows the unit will tell you that it does not add any ringing when EE(sic) is turned to its minimum position, nor does my 10PG :)

to all:

It is so cool to see us all trying to get a better picture for the new formats, be it BlueRay or HD-DVD. I am pretty tired of this format war bullsh** and would mainly like the new formats to be the best they can be.

The masters will sometimes be crappy which is to be expected for a lot of Columbia Scope transfers and surely others too, this is something we cannot change. BUT: I would hope that in the processing chain that follows the master no more artefacts will be introduced due to so called detail enhancements or sharpeners, noise reduction or other things some mastermind may come up with.

Oliver

Broadly speaking about Microsoft's perspective on the EE and other video quality issues, I should point out how many passioniate home theater owners have day jobs on the teams responsible for to HD DVD and VC-1 in general*. We care enormously about getting an accurate image to the screen, and getting out of the way of that anyway we can. E.g., ISF certification efforts for Media Center.

Also, the cleaner, and more accurate the source, the better the encoding. EE actually makes the codec have to work harder. We're as anxious for high quality sources as you are. I'm hoping that, as digital intermediate sources begin to become available for HD DVD authoring, we'll see yet another improvement in quality.

* I should point out that I'm transitionning from the Professional Content Group (responsible for HD DVD authoring and playback) to the Core Media Processing Technologies (codec and encoding tools). Shouldn't have any profond effect on my AVS participation.

... We care enormously about getting an accurate image to the screen, and getting out of the way of that anyway we can. E.g., ISF certification efforts for Media Center....

Hi Ben,

I trust excerpting your email didn't take it out of context, apologies if I missed the point.

Do you have an opinion on the value of Thomson's Film Grain Technology and whether it improves or harms video quality on real world living room TVs?

I have not made a rigorous study, but it seems to me that most consumer LCD TVs have some (IMHO poor) pixel processing (sometimes described as 'coring'). It appears that they suppress texture below a certain threshold and amplify it above that threshold. The idea seems to be that low levels are classified as noise. On some film content scenes, the grain is such that it lies across this threshold, and the overall result can be distracting and unnatural.

My understanding FGT is not only mandatory, but also that user control (on/off) will not be permitted. (i.e. it is optional in the content, but not in the player). Can you correct or confirm this impression?

Assuming that FGT is, in fact, a requirement, how is it overlaid in a PC? GPU?

Thanks

Gern

Do you have an opinion on the value of Thomson's Film Grain Technology and whether it improves or harms video quality on real world living room TVs?

As far as I'm concerned FGT is anathema to image quality because you need a grain free or at least heavily grain reduced master for the technology to make sense. And this master if coming from a grainy original will have grain reduction artifacts with very high probability. So you end up with grain reduction artifacts plus potential compression artifacts plus synthetic grain on top. I'm really opposed to such nonsense. If the master is grainy and that's how the film was made use a high bit rate till the grain looks correct. If the Codec can't handle grain use another that can.

I agree with your sentiments, as I think to most videophiles.

but: "If the Codec can't handle grain use another that can."

The question isn't the codec so much as the *bitrate* it's afforded. That's why that FGT was concevied to begin with... a way of getting "grain" using very low bit-rates.

None of this should even apply to the bitrates used for HD media like HD DVD and BD.

Fact is that review after review, says these titles are perfect or near perfect in picture quality. Take a look at Widescreen Review magazine's review of HD DVD titles. We both know that they complain very quickly about ringing yet there is not a complaint in sight, with most of the titles getting a score of 5 out of 5 on picture quality. And Gary Reber told me that this is a new scale for HD titles.

Amir


Just to dispute, MOST of the titles did NOT get 5/5... 6 titles were reviewed, 2 of them got a 5/5 for picture quality. 2 were 4.5, 1 was 4, and 1 was 'NR'...

I guess what would be interesting would be is how the scale compares to SD DVD... i.e. does a 4 on HD DVD/BD beat a 5 on SD DVD?

Just to dispute, MOST of the titles did NOT get 5/5... 6 titles were reviewed, 2 of them got a 5/5 for picture quality. 2 were 4.5, 1 was 4, and 1 was 'NR'...

I think source is the limiting factor here instead of the encoding/transfer.



I guess what would be interesting would be is how the scale compares to SD DVD... i.e. does a 4 on HD DVD/BD beat a 5 on SD DVD?

Interesting?

I trust excerpting your email didn't take it out of context, apologies if I missed the point.
Not at all.
Do you have an opinion on the value of Thomson's Film Grain Technology and whether it improves or harms video quality on real world living room TVs?
I haven't done much testing with the implementation in HD DVD (it isn't in BD). People smarter than I am tell me that there are issues with that particular implementation. VC-1 does fine with film grain, and just encoding what you want the texture to look like seems like a much more straighforward solution to me.

I've done more than my fair share of grain removal work, and I suspect that'd be the hardest part of the process. Synthesizing grain is pretty straightforward - usefully getting it out of the movie in the first place can be another matter entirely :).

You might imagine any studios using AVC for HD DVD's might want to keep things as similar as possible when encoding for BD. But if that was the case then they likely would not use FGT because it isn't also supported in BD.

- Tom

quick note.

The BD demo I saw over the weekend (25gig recordable BD played on sony prototype player at Sony Style store in DC) looked phenomenal. *I* saw no artifacts WHATSOEVER.

Any "edge ringing" that Plazman may have seen IMO was entirely due to the DRC processing in the Sony set which was upscaling the incoming 1080i signal from the BD player (and it still looked crystal-clear and 100% natural to my eyes... astonishing).

I'll report more in the germane threads.

A little more info about the DC Sony Style demo:

It was playing on a Qualia 006 70" display via HDMI. The picture was indeed spectacular. The 006 only accepts 1080i input via HDMI, so the set was actually de-interlacing (I suspect that's what David meant when he said 'upscaling'). No visible picture aritfacts whatsoever, but there was a slight hiccup in stream occasionally.

Audio was Dolby Digital via coax.

No word on codec used or bitrates.

Yes, by "upscaling" I meant the DRC/bobbing deinterlacing of the set.

More on my thoughts of the demo:

http://www.avsforum.com/avs-vb/showthread.php?p=7775391&&#post7775391

Well, I might also wonder if it was the mastering / transfer process and not the codec itself.Could the title be better HDDVD transfer vs Wowow presentation.I know one thing you can get right up on the screen even in high contrast stuff and not see EE in Serenity. This was one thing I looked closely for ,edges looked like edges in film not video.

Art

Art,

I'll go with this explanation of yours. The Japanese are very particular about their broadcast content that the deals they make with studios to show content usually include some kind of quality clause.

Chris.

people seem to forget a few things about consoles, they are products which are sold at a huge loss when they first come out, and these companies make their money by selling software.
this is one of those few products where competition isn't as beneficial for consumers as in about any other product.


Haven't we already debunked the myth selling consoles at loss? It's actually very uncommon.

"One of the most annoying parts of the gaming industry is how everyone thinks they are an expert and that, just because everyone assumes something is true, it must be true. So now The Gord will strike down with the horrible truth upon the realm of lies!

The first proclamation on setting straight these false truths, the mythological idea that consoles have always been sold below cost."

Read more here: http://www.actsofgord.com/Proclamations/chapter02.html

Finally, in studio tests, even 20-22 Mbit/sec VBR encoded MPEG-2 can not match the quality of VC-1 at much lower data rates. If this was not the case, there would be no pressure to add VC-1 to BD as there is.

Amir

Amir, after all the arguing, I think you have been proven right. I just saw House of Flying Daggers playing on the Samsung BD player attached to a 42" 1080p LCD, and what I saw was awful! MPEG-2 macroblocks were clearly visible throughout the scene where Zhang Ziyi dances around hitting gongs with her scarf. In comparison, I had recently watched a good bit of Phantom of the Opera and Last Samurai on the Toshiba HD DVD player attached to the very same display, and I never saw any obvious blocking artifacts at all. Specs are one thing, but I go by what my eyes tell me: MPEG-2 at these bitrates does NOT get the job done.

EDIT: Sorry, I posted the above immediately upon returning to forum after being offline for about a week. After skimming the Blu-Ray Software forum, it appears that my post is redundant - I get the distinct feeling that I'm not the only one who thinks that MPEG-2 on BD-25 sucks. ;)

It is surprising that Sony would release any commercial title with signficant artifacts--MP2 encoding should be well understood by now and movies are relatively easy to compress since they are only 24 fps and inherently progressive.

Do you happen to know what the bit rate of the HD-DVD was, and whether
both were encoded VBR or CBR? And was one movie substantially longer than the other? (or was more bonus material on the disc?)

In any case, the most I could conclude from the viewing of those discs is that the
particular MP2 encoder used on BD was not well suited to the source
material and bit rate in question. I wouldn't go so far as to make assertions about the suitability of MP2 for HiDef movies.

Even if the bit rate were similar between the two, there can be a fair amount
of post-processing in a decoder's video output stages to filter and reduce compression artifacts. Consequently, unless the same player (and content) is used in both cases, the data provided is noteworthy but not conclusive.

In my experience, video compression tests are the (2nd to) last refuge of scoundrels, and any point can be demonstrated by choosing a suitably bad encoder "A" to compare against encoder "B," which has been tuned for the test material.

That said, it would be pathetic if VC-1 weren't significantly better than MP2, given that 13 years of industry and academic research and the work of JVT were available to heist in coming up with a new codec. The real quality benchmark will be H.264.


Amir, after all the arguing, I think you have been proven right. I just saw House of Flying Daggers playing on the Samsung BD player attached to a 42" 1080p LCD, and what I saw was awful! MPEG-2 macroblocks were clearly visible throughout the scene where Zhang Ziyi dances around hitting gongs with her scarf. In comparison, I had recently watched a good bit of Phantom of the Opera and Last Samurai on the Toshiba HD DVD player attached to the very same display, and I never saw any obvious blocking artifacts at all. Specs are one thing, but I go by what my eyes tell me: MPEG-2 at these bitrates does NOT get the job done.

EDIT: Sorry, I posted the above immediately upon returning to forum after being offline for about a week. After skimming the Blu-Ray Software forum, it appears that my post is redundant - I get the distinct feeling that I'm not the only one who thinks that MPEG-2 on BD-25 sucks. ;)

It is surprising that Sony would release any commercial title with signficant artifacts--MP2 encoding should be well understood by now and movies are relatively easy to compress since they are only 24 fps and inherently progressive.

HD is not that easy to compress.

Do you happen to know what the bit rate of the HD-DVD was, and whether
both were encoded VBR or CBR? And was one movie substantially longer than the other? (or was more bonus material on the disc?)

The basic situation is that the Sony discs are using MPEG-2 compression and lossless (uncompressed) audio on discs that are either slightly (25GB) or possibly significantly (20GB) smaller than HD DVD (30GB). Sony is using MPEG-2 at bitrates similar to HD DVD (possibly lower), but HD DVD is using VC-1 compression which is on the order of twice as efficient. There's nothing inherently wrong with MPEG-2, but it's just not cut out for the application Sony is using it for.

In any case, the most I could conclude from the viewing of those discs is that the
particular MP2 encoder used on BD was not well suited to the source
material and bit rate in question. I wouldn't go so far as to make assertions about the suitability of MP2 for HiDef movies.

Again, it's not that MPEG-2 is unfit for HD, nor is there evidence that the encoder is poor. The situation is quite simply, that MPEG-2 is simply not capable of the compression required for the application (25GB BD).

Even if the bit rate were similar between the two, there can be a fair amount
of post-processing in a decoder's video output stages to filter and reduce compression artifacts. Consequently, unless the same player (and content) is used in both cases, the data provided is noteworthy but not conclusive.

Remember, HD DVD is using VC-1, which is a significantly more advanced codec, capable of producing the same quality at much lower bitrates. Further, consider that HD DVD actually has more capacity at the moment, and HD DVD isn't using uncompressed audio. And you'll understand the situation Blu-ray is in.

Blu-ray was (or at least appears to be) designed to handle MPEG-2 movies, to do that, dual-layer 50GB discs were specified. It also appears that from a tools perspective they were banking on having 50GB of space from the get go. However, Sony/BDA appears to be unable to produce 50GB disc in retail quanities and as a result they're stuck with BD25 and MPEG-2, which is the worst possible combination of circumstances.

If you look back in the forum a few weeks/months, this shouldn't be a surprise. Many members were predicting that BD25 with MPEG-2 would not be good.

In my experience, video compression tests are the (2nd to) last refuge of scoundrels, and any point can be demonstrated by choosing a suitably bad encoder "A" to compare against encoder "B," which has been tuned for the test material.

Granted, but we have a rather consistent bar of quality on both sides, such that I think comments are pretty representative. Everything makes sense on the Blu-ray side, that being many/most releases are sub par, namely because they're using MPEG-2 at (reportedly) about 18Mbps. The two standouts that are good, Terminator and 50 First Dates are exceptions, Terminator being a shorter movie and having the more reasonable 22-24Mbps, and 50 First Dates being an "easy" to compress movie.

On the HD DVD side, there seems to be a constent high bar of quality, with flaws generally being attributed to the master (The Fugitive).

That said, it would be pathetic if VC-1 weren't significantly better than MP2, given that 13 years of industry and academic research and the work of JVT were available to heist in coming up with a new codec. The real quality benchmark will be H.264.

Not sure where you're going with that, the current state of HD DVD vs Blu-ray seems to be a perfect demo for VC-1's prowas, both formats are using essentially the same bitrates overall, and HD DVD (VC-1) has a consistently higher quality tham MPEG-2. And if the experts/insiders are to be believed, VC-1 is being given much more bitrate than necessary.

Further, I don't know what you mean by H.264 being the "real quality benchmark", sounds like more blind anti-MS talk to me.

So...

does anyone have any news about the progress of getting VC1 on BD???

The real quality benchmark will be H.264.I don't get this either; blatant MS bashing. VC-1 is achieving transparency to the master. There can't be any higher quality than that. The only thing to improve from here is efficiency.

I don't get this either; blatant MS bashing. VC-1 is achieving transparency to the master. There can't be any higher quality than that. The only thing to improve from here is efficiency.

What do you expect with this kind of comment

That said, it would be pathetic if VC-1 weren't significantly better than MP2, given that 13 years of industry and academic research and the work of JVT were available to heist in coming up with a new codec.

He is accusing MS of stealing. If I were the mod I'd ask him to either prove or retract the statement. That too a newbie ....

Here we go, Swordifsh EE, timestamp 46:40. My camera is very low quality, but that's enough to see clearly the problem:
http://xs101.xs.to/xs101/06222/EE_Swordfish.JPG (http://xs.to)

It looks like part of the distortion in your picture is due to your digital camera. Purple fring is a very common problem with digital camera, it's particular obvious here, but absent from the 1080p HD cap file.

Amir, after all the arguing, I think you have been proven right.
Thanks a lot for the kind words :). It has been a long journey (3+ years and hundreds of arguments later). But there results were worth it I think. I see people letting me even post in blu-ray forums now! To say nothing of the various member signatures with VC-1 in them. So humbling to see the recognition come so fast when the data is before people...

Thanks again.

It looks like part of the distortion in your picture is due to your digital camera. Purple fring is a very common problem with digital camera, it's particular obvious here, but absent from the 1080p HD cap file.

So, you're saying that I first reported ringing on Swordfish, then managed to find a camera able to fake that ringing to prove my point?

The other people seing the same ringing with a completely different setup must have a bionic eye of the same brand than my camera...

In my experience, video compression tests are the (2nd to) last refuge of scoundrels, and any point can be demonstrated by choosing a suitably bad encoder "A" to compare against encoder "B," which has been tuned for the test material.


I'm afraid you are missing the point. This is not a "video compression test" we are talking about, this is real life product offered for sale on the open market! It is way, way too late to be talking about tests. I agree with your point about MPEG-2 encoders - they are certainly mature and capable of providing excellent results. The problem is this: the macroblock artifacts I am seeing are almost certainly the result of not enough bits being allocated to the scene in question by the encoder. There are a number of different reasons why this might have happened, but it really doesn't matter why. Too little space on BD-25 (BD-20?), crappy encoding software, human error, bad luck - it really doesn't matter. What matters is that Sony has most likely risked hundreds of millions (maybe even billions) of dollars on developing the Blu-Ray format only to produce a final result that hardly looks better than macroblock-infested OTA HDTV!

Students of irony should pay close attention to this fiasco. If you were to take all the posts in this forum in the last year made by Blu-Ray supporters downplaying the significance of Sony's MPEG-2 strategy and lack of progress with BD-50 and all the posts made by those same Blu-Ray supporters who assured us that Blu-Ray's superiority would be evident when both formats had launched, and then you printed those posts on paper and pasted them end-to-end... the resulting ribbon of Blu-Ray hype would be long enough to reach the moon and back! The irony, of course, is that this result was totally predictable. There were plenty of omens to warn the wise along the way. Consider, for example, the absurd maligning on these very forums of Microsoft employees and other industry insiders who have generously devoted so much time to educating us about the HD DVD format and how it compares to Blu-Ray, and then ask yourself one simple question: Has Sony EVER, in all the months and years the format debate has raged on AVS Forum, deigned to send an official Sony representative down here to regularly discuss the Blu-Ray format with us lowly enthusiasts and get our feedback?? Microsoft sent us a VP, and Sony can't even spare a f---ing janitor to extol the virtues of Blu-Ray. No wonder Blu-Ray supporters are now reeling in shock and disappointment, pulling their hair out as every new review of the BD launch titles hits the web - their only sources of information have been press releases and rumors. The reason for this ridiculous outcome is utterly simple: a group of people who claimed that quality was their only priority have put their faith in Sony, a company which has not cared one bit about quality since the long gone glory days of the Trinitron TV. Even as I write this I can look over at my bookshelf and see the dust-covered MiniDisc player I purchased years ago when I still believed that Sony was a brand I could trust, a forlorn artifact of one company's greed and stupidity and total incompetence when it comes to launching and nurturing new formats.

To those who would still fork out $1000 for a BD player at this point, I can only say this: Make sure there is some space on your own bookshelf, you are going to need it.

lazyn00b

This would probably be a very good time for Microsoft to roll out some new VC1 (or even HD DVD) authoring tools with a bunch of fanfare and offer them to Fox & Disney. ;)

- Tom

lazyN00b,

I applaud your response to this thread. As one of your supposed "HD Insiders", my responses have not been to extoll HD's superiority over BD, but to emphasize the quality being achieved from the advanced codecs. I don't think Sony's BD technology will fail here, but I do feel Mr. Eklund and his encoding team was lazy in their research and went for what was handy. I've seen 18 ABR/35 PBR MPEG-2 material from Sony's own encoder that about matches a 13/24 VC-1 from MS' December05-built encoder (significant improvements have been made since). I've heard 24/48 TrueHD that beats the quality, but matches the bitrate, of 16/48 PCM. Sony's encoding people have made bad choices and we can only hope they learn from them as do the other studios in both camps (war metaphors...). Competition is only a good thing as has been shown from the US/Russian Space Race to today as it consistently produces better, more reliable, and usually higher quality products (Beta/VHS aside). However, I hope all of this comes from one or the other before the world says a collective "screw discs..." with their wallets. I hope for a merged format. It can still happen!!!

Sincerely,
Cjplay.

CJplay,

of course, WB's True HD tracks are only 16/44.1 too... though I agree that efficiency makes lossless packing preferred to flat LPCM any time whether BD or HD DVD (and if we can get 24-bit in the same space as 16-bit LPCM that's super!).

So... allow me to ask yet again...

does anyone know the progress report of getting VC1 on production BD? Are we closer? Problems? Moving forward? Time-frame???

Seems to be the most relevent point of discussion yet no one is bothering to talk about it.

Good rant. Except for this:

Sony, a company which has not cared one bit about quality since the long gone glory days of the Trinitron TV.
lazyn00b

...which is going overboard to say the least. Blu-Ray stumble aside, Sony has continued to push the envelope in terms of quality in it's upper-tier A/V products. The SXRD projectors and RPTVs were universally lauded, consistently rated "the best" or top of class
in their category, by experienced reviewers (and lots of happy customers). I can't watch the Sony Ruby projector and conclude with you that this is from a company that doesn't care "one bit" about quality.

It's fine to point out the bad, but keeping it real tends to add more weight, even to rants.

Cheers,

Rich is right. Sony still has its moments. The Ruby is certainly one of them....

Who has access to the 'Pro VC-1 Encoder'?
Only the post houses making HD DVD movies today.

Does Scenarist clearly state that they have the 'J6P VC-1 Encoder'?
I don't know what they state. I do know that it is not yet the above encoder.

So, you're saying that I first reported ringing on Swordfish, then managed to find a camera able to fake that ringing to prove my point?

Obviously I said no such thing. Purple fring happens on most, if not all digital cameras during high contrast scenes, since you mentioned you used a low quality camera, the purple fring in your picture definitely contains contribution from the camera/jpeg compression, to which degree I couldn't say. The real giveaway is the purple, which is not in Kai's sharpened picture.

It looks like you used a Nikon 5200, if so, it's purpe fringing is not too bad, but still, here is a review, look at the picture to the left:
http://www.dpreview.com/reviews/nikoncp5200/page5.asp

Even though many people take side in either HD or BD camp, I only take side of what's good for consumers. Take it easy, just because I pointed out some possible limitation in your picture, it does not mean I have some agenda here :-)

It looks like you used a Nikon 5200, if so, it's purpe fringing is not too bad, but still, here is a review, look at the picture to the left:
http://www.dpreview.com/reviews/nikoncp5200/page5.asp

Ok, I'm definitively impressed that you found out the make and model of my camera!

You should contact the James Randy foundation (http://www.holysmoke.org/sdhok/randi.htm), and claim the 1 million dollar price they have been offering for any scientifically validated paranormal feat!

Even though many people take side in either HD or BD camp, I only take side of what's good for consumers. Take it easy, just because I pointed out some possible limitation in your picture, it does not mean I have some agenda here :-)

I apologize, I was bit curt with you. I just should have said that the camera shows the ringing that exists, and that what was added by its own imperfections is negligible compared to what you can see directly in that context.

I have no agenda either. I love HD-DVD, and will love BR too.

Robert, your image contains so-called EXIF meta-data. It is embedded in the picture:

see here:

http://img240.imageshack.us/img240/5900/062220061017301sv.th.jpg (http://img240.imageshack.us/my.php?image=062220061017301sv.jpg)

Ok, I'm definitively impressed that you found out the make and model of my camera!

You should contact the James Randy foundation (http://www.holysmoke.org/sdhok/randi.htm), and claim the 1 million dollar price they have been offering for any scientifically validated paranormal feat!
...
Maybe I should make a claim like this :cool: and pray for them to accept it. $1 million bucks, sounds like I could use it somewhere :) If only Kai had not spoiled the plot :-(

Seriously, As Kai showed, your jpeg file contains camera model and some other parameters related to that picture. Many graphics programs can access that info.

Robert, your image contains so-called EXIF meta-data. It is embedded in the picture

Thanks, that's interesting to know.

That said, it would be pathetic if VC-1 weren't significantly better than MP2, given that 13 years of industry and academic research and the work of JVT were available to heist in coming up with a new codec. The real quality benchmark will be H.264.

I'll disregard the H.264 reference (irrelevant to this thread) as well as the "heist" malarky.

Your main point above is that, of course it should be expected that VC-1 > MPEG2. That's true, but it's clearly NOT what the BR supporters were saying before now, as evidenced by this very thread's existence. The talk from many BR supporters was that VC-1 might be technically superior to MPEG2 in theory, but that in reality MPEG2 was superior because it was more mature. That hypothesis has been thoroughly discredited.

I think it's time to give Microsoft and whoever else was involved in VC-1's development some credit here.

BTW, remember the anti-VC-1 FUD that was going around saying that various patent holders of tech used by VC-1 were going to stop VC-1 from getting standardized by STMPE? What happend with that, anyway? Was that just some more BS?

Yep, Microsoft do deserve credit for VC-1. It's been proven "in the field". Hats of to Microsoft. (Where would HD DVD have been without it?)

Is H.264 discussion out of scope for this thread just because of the title? Should we make another VC-1 vs. H.264 thread? Now that Panasonic has announced H.264 support in their Blu-ray tools, and probably both VC-1 and H.264 titles will be out in the near future, it's time for us to get down and dirty in some predictions. Soon it will be too late. :)

Yep, Microsoft do deserve credit for VC-1. It's been proven "in the field". Hats of to Microsoft. (Where would HD DVD have been without it?)

Is H.264 discussion out of scope for this thread just because of the title? Should we make another VC-1 vs. H.264 thread? Now that Panasonic has announced H.264 support in their Blu-ray tools, and probably both VC-1 and H.264 titles will be out in the near future, it's time for us to get down and dirty in some predictions. Soon it will be too late. :)

Why are the BD people reluctant to use VC1 which at least has a track record, a short one but none the less a record? Could it be they simply don't want to be labeled as a "me too" format?

Robert D,

Who are "BD People"?

The studios other than Sony are clamoring for VC-1 on BD. We consumers at AVS are waiting for it. Who isn't hoping to get VC1 on production BD ASAP???

Robert D,

Who are "BD People"?

The studios other than Sony are clamoring for VC-1 on BD. We consumers at AVS are waiting for it. Who isn't hoping to get VC1 on production BD ASAP???

I don't think that Robert was referring to studios, but to certain BR supporters who don't like VC1 for some reason (my guess is that it's because VC1 is made by Microsoft (of course, Microsoft also has patents in H.264)). Over the past two years, or so, certain BR supporters have belittled VC1. I remember them saying that Microsoft lied about its capabilities, that patent holders of tech in VC1 would block its standardization, that MPEG2 was better because it was more mature, and that H.264 was vastly superior.

Also, certain members of BDA have something against VC1, those members being Sony and Apple. Sony has publicly trashed VC1, saying it was inferior to MPEG2 (Sony now has scrambled egg on its face); and Apple, for some idiotic reason, tried to get VC1 removed from the BD spec.

Some might have belittled VC-1 early in the game, but no more, it has been proven to do the job quite nicely.

That said, as you say, some studios are less likely to use VC-1. Obviously it is then better to use MPEG4 than MPEG2. Or rather, that is the hope, which is why we need such titles so we can compare them to VC-1 titles. My guess that they won't differ much in picture quality.

Robert, I'm not against VC-1 by any means, as I thought I mentioned in my post above. But what is wrong with seeing what MPEG4 can do? It would seem a bit silly at this stage to settle for VC-1 as the #1 standard, before we've seen what MPEG4 can do... (Not counting Quicktime.)

Of course, since it is more time consuming to compress AND takes "more" hardware to decode, VC-1 might be preferred since it probably would play back smoothly on more PC's.)

Take a look at what the highly respected Joe Kane has to say about the choice of MPEG-2 over both VC-1 and h.264.

http://www.cinenow.com/us/news-2118.html

Starting around 7th minute:

"It is my personal believe, that Blu-Ray is all about greed" - Joe Kane.

The guy interviewing him seem not having a clue what he talks about.

Diogen.

I have seen now most of the HD-DVD titles, and several Blu Ray ones.

With VC-1, the grit that's all over most low or average bitrate MPEG2 encodings has completely disappeared. The only remaining artifact is EE or false contouring, which might be due to bad masters.

Sony on the other hand has managed to make MPE2 encodes that most of the time look terrible despite a bitrate that should have been enough to get an acceptable (but not great) result. Except with Underworld Evolution, all the BR titles I sampled looked awful, drab with SD like color dynamic, poor details and all over shimmering.

Considering how good Wowow and D-Theater encodes can look, Sony seems to have been using the worst MPEG2 encoder that can be found.

My opinion is that BR is just not ready, but that Sony had to rush it out anyway because HD-DVD was here.

BR is a shameful product at the moment, I hope it will catch up in a few months. BR has however one redeeming quality: without the competition that it generated, HD-DVD would be red laser based...

If it hasn't been mentioned already, WME Studio Edition beta has been released:
http://www.microsoft.com/windows/windowsmedia/forpros/encoderse/default.aspx

One big disappointment, though. From what I gather on microsoft.public.windowsmedia.encoder.studioedition, it does not accept MPEG2 transport streams as input...

If it hasn't been mentioned already, WME Studio Edition beta has been released:
http://www.microsoft.com/windows/windowsmedia/forpros/encoderse/default.aspx

One big disappointment, though. From what I gather on microsoft.public.windowsmedia.encoder.studioedition, it does not accept MPEG2 transport streams as input...
Yep, sorry. Beta 1 only supports raw/yuv or .avi sources.

If you transcode from .ts to a Huffyuv .AVI file, it'll work fine.

WME SE will take AVISynth scripts though ;) At least it lets you import them.

Yep, Microsoft do deserve credit for VC-1. It's been proven "in the field". Hats of to Microsoft. (Where would HD DVD have been without it?)

Is H.264 discussion out of scope for this thread just because of the title? Should we make another VC-1 vs. H.264 thread? Now that Panasonic has announced H.264 support in their Blu-ray tools, and probably both VC-1 and H.264 titles will be out in the near future, it's time for us to get down and dirty in some predictions. Soon it will be too late. :)

Based upon discussion threads in doom9 & previous test results of the wmv9 platform my money is on Panasonic. vc-1 is having a hard time against x264.

See here (http://forum.doom9.org/showthread.php?t=111275&page=7)

Based upon discussion threads in doom9 & previous test results of the wmv9 platform my money is on Panasonic. vc-1 is having a hard time against x264.

See here (http://forum.doom9.org/showthread.php?t=111275&page=7)
Not relevant to HD formats. These guys are testing with a different versions of the encoder than what the studios use (the studio have access to a lot more controls, and it's a more recent build with some important enhancements) and they're testing at much lower bits/pixel rates.

And they're not comparing it to the Panasonic encoder either.

Remember we compare implementations, not codec standards.

Not relevant to HD formats. These guys are testing with a different versions of the encoder than what the studios use (the studio have access to a lot more controls, and it's a more recent build with some important enhancements) and they're testing at much lower bits/pixel rates.

And they're not comparing it to the Panasonic encoder either.

Remember we compare implementations, not codec standards.

Well I have not seen any evidence yet in any real world demonstration that implies that an expert vc-1 implementation matches/exceeds an expert h.264 high profile implementation given the same testing parameters. DVD-Forum & BDA never got around to testing vc-1 v. high profile ;) . We all know that its the implementations & people that matter in the end. So time will tell. :cool:

Life is full of humorous ironies:
One the one hand we have Sony bragging about the superiority of MPEG2 for Blu-ray.
On the other hand we have Sony listing an MPEG Noise Reduction Circuit in their brand spanking new SXRD display:

Digital MPEG Noise Reduction Circuit which reduces MPEG artifacts such as "block noise" and "mosquito noise" caused by compression used in digital broadcasts used to maximize bandwidth.

http://www.sonystyle.com/is-bin/INTERSHOP.enfinity/eCS/Store/en/-/USD/SY_DisplayProductInformation-Start?ProductSKU=KDS60A2000&

Kick me if I'm stupid:)

Well I have not seen any evidence yet in any real world demonstration that implies that an expert vc-1 implementation matches/exceeds an expert h.264 high profile implementation given the same testing parameters. DVD-Forum & BDA never got around to testing vc-1 v. high profile ;) . We all know that its the implementations & people that matter in the end. So time will tell. :cool:
Time has already told :). We have three US studios using VC-1. They tested them all, and could use any that they liked. They picked VC-1. Remember that for HD DVD, we have always had the option of using ALL three codecs including AVC.

Toshiba has HD DVD titles in Japan using AVC HP. All of them underperform the US VC-1 titles with compression artifacts and general softness. They are better than MPEG-2 on BD mind you, but lose out to VC-1. And you don't have to take my word for it. For less than $50, you can import one and play it on your HD DVD player and see for youself.

And I hope we all agree that the studios have the proper means and capabilities that average internet user does not have to test with. And it is not like it is one studio. We have all three.

Sure someone could build a better AVC encoder. But we don't know that. Nor do you all know where we are going with our encoder.

You may be interested to know that there is ONE title in AVC coming from a US studio. But that was the first title they encoded and they quickly switched to VC-1 and are now very happy. I will not disclose the title but when you see it, you are not going to be dancing with joy over picture quality :).

You may be interested to know that there is ONE title in AVC coming from a US studio. But that was the first title they encoded and they quickly switched to VC-1 and are now very happy.

Are they using the Panasonic H.264 HP encoder for this ONE title? I certainly hope not.

Are they using the Panasonic H.264 HP encoder for this ONE title? I certainly hope not.
Nope. The Panasonic encoder will probably only work for BD. At least I have not heard of them offering it for HD DVD.

Amir,

Any reason (that you can say) Toshiba Entertainment (and any other Japanese studios) haven't adopted VC1 as of yet for their titles?

Toshiba has HD DVD titles in Japan using AVC HP. All of them underperform the US VC-1 titles with compression artifacts and general softness.True, but it is not hard for a company to make a bad video encoder.


Nope. The Panasonic encoder will probably only work for BD.Amir, just curious but have you recently seen any video encoded with the Panasonic encoder?

....coming from a US studio...... I will not disclose the title....

It has to be Paramount, or am I reading too much into this?

At least I have not heard of them offering it for HD DVD.

Maybe they find the market for HD DVD is very hard to penetrate. :)

(typing on the phone)

Ja, no good reason other than lack of focus. our codec engineers are in us so it is more difficult to support japan. also folks there like richard thought avc was perfect :). but things have changed and vc1 is getting deployed there....

It has to be Paramount, or am I reading too much into this?
I have to get better at giving obscure hints :D.

Maybe they find the market for HD DVD is very hard to penetrate. :)
Or even BD. Did they not say they have no studio customers yet for their AVC encoder?

You're killing us Amir.
I've got some titles on pre-order and I'm waffling on some others.
Now I've got to know which one is the "bad" title.

You're killing us Amir.
I've got some titles on pre-order and I'm waffling on some others.
Now I've got to know which one is the "bad" title.
Well, first know that my eyes may be different than yours. But are you a music fan???

Well, first know that my eyes may be different than yours. But are you a music fan???

D'OH!

Yeah...Amir.

Let's work on those "hints." You know you're not supposed to outright give it away, right? ;)

Just out of curiosity...

Will this film still look significantly better than its SD release? I mean...how bad could this codec be? I really want to buy this title but if it is no different than the SD release then I will be kinda mad!

Plus they are my favorite band so I gotta buy it :D

It takes years to develop a top performing encoder. Encoders go through countless revisions. Experience is learned with each step. I would never expect a encoder slapped together in a tight schedule to beat one authored by a specialist in the field. Regardless of codec specs. :rolleyes:

It takes years to develop a top performing encoder. Encoders go through countless revisions. Experience is learned with each step. I would never expect a encoder slapped together in a tight schedule to beat one authored by a specialist in the field. Regardless of codec specs. :rolleyes:


VC-1 was not created in an overnight session.

Do you work for PHL? I seem to remember your nick from a presentation some time ago.

Well, first know that my eyes may be different than yours. But are you a music fan???

My money is on U2: Rattle & Hum from Paramount... ;)

VC-1 was not created in an overnight session.

Do you work for PHL? I seem to remember your nick from a presentation some time ago.

I'm not talking about vc-1 in this instance.

Time has already told :). We have three US studios using VC-1. They tested them all, and could use any that they liked. They picked VC-1. Remember that for HD DVD, we have always had the option of using ALL three codecs including AVC.

Toshiba has HD DVD titles in Japan using AVC HP. All of them underperform the US VC-1 titles with compression artifacts and general softness. They are better than MPEG-2 on BD mind you, but lose out to VC-1. And you don't have to take my word for it. For less than $50, you can import one and play it on your HD DVD player and see for youself.

And I hope we all agree that the studios have the proper means and capabilities that average internet user does not have to test with. And it is not like it is one studio. We have all three.

Sure someone could build a better AVC encoder. But we don't know that. Nor do you all know where we are going with our encoder.

You may be interested to know that there is ONE title in AVC coming from a US studio. But that was the first title they encoded and they quickly switched to VC-1 and are now very happy. I will not disclose the title but when you see it, you are not going to be dancing with joy over picture quality :).

Has this have anything to do with the infamous Toshiba demos?

There is substantial PQ difference between Toshiba H.264 encoder and Panasonic one. It doesn't make sense talking inferior H.264 implementation as the limitation of H.264 itself.

That was what I was getting at. :)

BR has however one redeeming quality: without the competition that it generated, HD-DVD would be red laser based...


This is off topic but I thought I was under the impression that BD was a blue-laser and HD-DVD was red. However about a week ago I read some of the specs of HD-DVD and saw that it relied on Blue Laser Technology as well. Was this a later change in the spec that occured? When it was initially proposed was it red laser based?

Thanks

Tim

HD DVD was always blue laser. a long time ago WB and Toshiba were pushing for a red laser solution, but no one wanted it. And with the formation of the Blu ray disk founders (later renamed BDA) the DVD forum did not have a choice.

That being said both HD DVD and BD have a red laser variety that is HD DVD/BD logical structure on a DVD disk (called HD DVD-9 and BD-9)

Yep, sorry. Beta 1 only supports raw/yuv or .avi sources.

If you transcode from .ts to a Huffyuv .AVI file, it'll work fine.


I did this just to test - took up a three drive raid configuration :>

Can't wait until it supports mpeg transport stream conversion....... I will rid my life (and hard drives) of mpeg2.


Any chance we will ever see a VC-1 HD video camera? I love my HDV resolution, but don't like the compression or transport.

What affordable encoders (except Microsoft's PEP and Sonic CineVision) able to encode Cineform 4:2:2 10 bit AVI to VC1 elementar stream? Thank you.