Industry Insiders Q&A MASTER THREAD [separate thread for Xbox/Add On & PS3] - Page 88 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 
Thread Tools
post #2611 of 4841 Old 03-04-2007, 06:19 PM
AVS Forum Special Member
 
RBFilms's Avatar
 
Join Date: May 2006
Location: New York / California
Posts: 1,248
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 46 Post(s)
Liked: 29
I meant Dolby Digital Plus.....sorry.

I don't use Dolby Tru-HD ... it is basically MLP Audio. Dolby is not flawed...I just prefer DTS.

Rich

Quote:
Originally Posted by TrevorS View Post

You are here saying that lossy DTS and lossless Dolby TrueHD are absolutely not transparent to the source. Well, I would expect that for the DTS (though I always appreciated their 1.5Mbps sound tracks ), but I would like to know why that would be the case for TrueHD. Are you basically saying that MLP is a flawed "lossless" codec?

Anybody have any information as to what the real problem is here?


Richard J. Casey



Disney WOW - World of Wonder


Producers Guild of America, New Media Council
(BD Industry Insider)
RBFilms is offline  
Sponsored Links
Advertisement
 
post #2612 of 4841 Old 03-04-2007, 06:23 PM
AVS Forum Special Member
 
RBFilms's Avatar
 
Join Date: May 2006
Location: New York / California
Posts: 1,248
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 46 Post(s)
Liked: 29
My costs range from 20% to 50% more depending on the complexity of the menus...if that helps.

Quote:
Originally Posted by ILJG View Post

Without getting into numbers that he admitted could get him into hot water, there seems to be little ambiguity in this assertion.

Any counters?


Richard J. Casey



Disney WOW - World of Wonder


Producers Guild of America, New Media Council
(BD Industry Insider)
RBFilms is offline  
post #2613 of 4841 Old 03-04-2007, 06:58 PM
AVS Forum Special Member
 
skogan's Avatar
 
Join Date: Feb 2004
Location: Koror, Palau
Posts: 4,469
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by RBFilms View Post

My costs range from 20% to 50% more depending on the complexity of the menus...if that helps.

That was very helpful, thank you.
skogan is offline  
Sponsored Links
Advertisement
 
post #2614 of 4841 Old 03-04-2007, 11:26 PM
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,829
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1334 Post(s)
Liked: 749
Ron, appreciate you chiming in. It is great to have another compression expert to interact with on these topics . And sorry for the late reply. Had to go back to getting some business done while here.

Anyway, your objection is somewhat valid wrt to wording of the last statement I made, but not relative to the actual argument . So let me state that more clearly. There is no way to force AVC to use single pixel filtering as VC-1 uses. And that is factual and evident from the pseudo code. I can't quite tell from the wording of your statement if you are disputing this or not. I could read your statement as sometimes AVC can pixel filter just one pixel as opposed to it having a fixed one-pixel mode. Just in case you mean the latter, and for the benefit of others, please allow me to explain why this is not correct.

What the AVC spec allows, as with VC-1 or any other codec with a loop filter, conditions for not filtering content based on certain criteria. This includes things like whether this is an intra/inter block, or the block is a boundary block in a macroblock, or if this pixel lies on an edge (think of left side of your screen - you can't go past it to filter anything), and the quantization step size (how much you are compressing the video - the adaptive part of the filter). But when these special conditions are not met (say, a block encapsulating background grain in the somewhere away from the edge), the algorithm goes out and filters three pixels on each side of the block, even in a small 4x4 block! As a result, some/many pixels will be filtered multiple times!

You can verify the above from the pseudo code. The first part clearly lists that there are three pixels potentially being operated on (input sample values pi and qi (i = 0..2) which means pixels corresponding to i=0, 1 and 2, or three pixels). So we know this code is not talking about single filtering mode at all.

Then there are conditional if statement which examine the above factors and if they are not triggered, the filter goes to operate on the three pixels. If those conditions are met, the otherwise part kicks in and filters the one pixel which you highlighted (classic if then, else statement for you programmers ). Even when the otherwise part is triggered, it is only for that block. The next block can easily get the reverse treatment and get filtered all the way (which again can mean multiple filtering of the same pixels in a 4x4 block). And the block after that and so on.

Net, net, other than drilling into details of how this algorithm works (showing the complexity behind even seemingly simple things), the assertion I made remains valid. AVC will filter and reduce grain and fine texture as a result of this aggressive filtering as compared to VC-1. It will attack many blocks and pixels within them as it runs into difficulty coding them without artifacts. Not only is this evident from the code/spec, but from many encodings and shoot outs we have participated in. The loop filter without a doubt is a major factor in how AVC softens detail in difficult material.

Talking about real pictures, have you had a chance to encode any content with AVC for these HD formats? If so, have you seen an encoder with a single pixel loop filter? If so, who makes it and what title did you work on?

Thanks again for commenting. As Ken Watenabe said to Tom Cruise's character in The Last Samurai, I enjoy our conversations! (You can tell I am still in Japan, can't you? ).
amirm is offline  
post #2615 of 4841 Old 03-05-2007, 04:18 AM
Member
 
rjc999's Avatar
 
Join Date: Jun 2003
Posts: 29
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
If you're taking less samples, what about aliasing effects then? Surely, the values being filtered will be more vulnerable to aliasing effects, lowering the overall prediction efficiency.
rjc999 is offline  
post #2616 of 4841 Old 03-05-2007, 05:13 AM
AVS Forum Special Member
 
trbarry's Avatar
 
Join Date: Jan 2000
Location: Gainesville FL USA
Posts: 9,884
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by amirm View Post

Ron, appreciate you chiming in. It is great to have another compression expert to interact with on these topics . And sorry for the late reply. Had to go back to getting some business done while here.

Anyway, your objection is somewhat valid wrt to wording of the last statement I made, but not relative to the actual argument . So let me state that more clearly. There is no way to force AVC to use single pixel filtering as VC-1 uses. And that is factual and evident from the pseudo code. I can’t quite tell from the wording of your statement if you are disputing this or not. I could read your statement as “sometimes AVC can pixel filter just one pixel” as opposed to it having a fixed one-pixel mode. Just in case you mean the latter, and for the benefit of others, please allow me to explain why this is not correct.

What the AVC spec allows, as with VC-1 or any other codec with a loop filter, conditions for not filtering content based on certain criteria. This includes things like whether this is an intra/inter block, or the block is a boundary block in a macroblock, or if this pixel lies on an edge (think of left side of your screen – you can’t go past it to filter anything), and the quantization step size (how much you are compressing the video – the adaptive part of the filter). But when these special conditions are not met (say, a block encapsulating background grain in the somewhere away from the edge), the algorithm goes out and filters three pixels on each side of the block, even in a small 4x4 block! As a result, some/many pixels will be filtered multiple times!

You can verify the above from the pseudo code. The first part clearly lists that there are three pixels potentially being operated on (“input sample values pi and qi (i = 0..2)” which means pixels corresponding to i=0, 1 and 2, or three pixels). So we know this code is not talking about single filtering mode at all.

Then there are conditional “if” statement which examine the above factors and if they are not triggered, the filter goes to operate on the three pixels. If those conditions are met, the “otherwise” part kicks in and filters the one pixel which you highlighted (classic “if then, else” statement for you programmers ). Even when the “otherwise” part is triggered, it is only for that block. The next block can easily get the reverse treatment and get filtered all the way (which again can mean multiple filtering of the same pixels in a 4x4 block). And the block after that and so on.

Net, net, other than drilling into details of how this algorithm works (showing the complexity behind even seemingly simple things), the assertion I made remains valid. AVC will filter and reduce grain and fine texture as a result of this aggressive filtering as compared to VC-1. It will attack many blocks and pixels within them as it runs into difficulty coding them without artifacts. Not only is this evident from the code/spec, but from many encodings and shoot outs we have participated in. The loop filter without a doubt is a major factor in how AVC softens detail in difficult material.

Talking about real pictures, have you had a chance to encode any content with AVC for these HD formats? If so, have you seen an encoder with a single pixel loop filter? If so, who makes it and what title did you work on?

Thanks again for commenting. As Ken Watenabe said to Tom Cruise’s character in The Last Samurai, “I enjoy our conversations!” (You can tell I am still in Japan, can’t you? ).

Aren't both encoders smart enough to avoid the filtering completely when there is siffucient bit rate to avoid blocks anyway?

- Tom

Why don't we power our electric cars from greener, cheaper Liquid Fluoride Thorium Reactors?

Tom Barry - Find my video filters at www.trbarry.com
trbarry is offline  
post #2617 of 4841 Old 03-05-2007, 11:01 AM
AVS Forum Special Member
 
benwaggoner's Avatar
 
Join Date: Sep 2003
Location: Portland, OR
Posts: 2,630
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 17
Quote:
Originally Posted by trbarry View Post

Aren't both encoders smart enough to avoid the filtering completely when there is siffucient bit rate to avoid blocks anyway?

Sure, but that's not a given at HD optical bitrates.
benwaggoner is offline  
post #2618 of 4841 Old 03-05-2007, 11:47 AM
Rio
Advanced Member
 
Rio's Avatar
 
Join Date: Feb 2003
Posts: 915
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by benwaggoner View Post

Sure, but that's not a given at HD optical bitrates.

Ben, are you referring to "inloop filter" specifically, or something another side effect in general caused by compression itself, like high QP and high numbers used in quant matrix?
Rio is offline  
post #2619 of 4841 Old 03-05-2007, 12:00 PM
AVS Forum Special Member
 
Phloyd's Avatar
 
Join Date: Dec 2001
Location: Santa Clara, CA
Posts: 2,846
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by benwaggoner View Post

Quote:


Aren't both encoders smart enough to avoid the filtering completely when there is siffucient bit rate to avoid blocks anyway?

Sure, but that's not a given at HD optical bitrates.

Hi Ben,

So are you saying that 'optical disc rates' are not high enough to guaranty optimal compression?

And if so, having more bitrate is actually beneficial to the encode?
Phloyd is offline  
post #2620 of 4841 Old 03-05-2007, 12:08 PM
AVS Forum Special Member
 
UxiSXRD's Avatar
 
Join Date: Feb 2006
Location: Southern California
Posts: 5,588
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 11
What bitrate would make that a given?

-- "No matter where you go, there you are."

SXRD/Denon/Polk/PS3/360/WiiU
Blu-ray bliss
SXRD Owner's Thread
PSN: Uxi

UxiSXRD is offline  
post #2621 of 4841 Old 03-05-2007, 12:41 PM
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,829
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1334 Post(s)
Liked: 749
Quote:
Originally Posted by rjc999 View Post

If you're taking less samples, what about aliasing effects then? Surely, the values being filtered will be more vulnerable to aliasing effects, lowering the overall prediction efficiency.

Good observation . Fortunately, it is not a concern here because we are not talking about the filter taps (samples) in the loop filter. Rather, how many samples are modified/filtered. The actual filter length is fairly comparable in the two codecs.

Even if you were right, surely you don't want to filter a pixel two or more times, in the name of anti-aliasing, do you? Because that is exactly what happens with AVC if you filter a 4x4 block, with 3 pixels from the sides.

If you are talking about the motion compensation, and not the loop filter, then I already answered that. Higher order filters ring, and that is not good. Codec design is not just science. It is also art. You have to balance things based on real experience with real content. And here, we had the upper hand, with a lot more HD experience, allowing us to play with different filters and see which one worked better. It is not like a higher order filter was an unknown science to us or difficult to do . It was the result of our testing with high-res content that led us to our choice there.

BTW, a bicubic 4-tap filter is quite nice by video standards. When I managed engineering of broadcast equipment, we couldn't afford the hardware cost of bi-cubic so the common choice was bilinear filtering!
amirm is offline  
post #2622 of 4841 Old 03-05-2007, 01:09 PM
AVS Forum Special Member
 
benwaggoner's Avatar
 
Join Date: Sep 2003
Location: Portland, OR
Posts: 2,630
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 17
Quote:
Originally Posted by Phloyd View Post

So are you saying that 'optical disc rates' are not high enough to guaranty optimal compression?

Not without triggering the loop filter, no.

Quote:


And if so, having more bitrate is actually beneficial to the encode?

For AVC, yes. VC-1 hits the detail bar at a lower bitrate.
benwaggoner is offline  
post #2623 of 4841 Old 03-05-2007, 01:19 PM
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,829
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1334 Post(s)
Liked: 749
Quote:
Originally Posted by Phloyd View Post

Hi Ben,

So are you saying that 'optical disc rates' are not high enough to guaranty optimal compression?

And if so, having more bitrate is actually beneficial to the encode?

Ben may have his own answer to this. But I don't believe he was not talking about what is "optimal" and what is not. Rather, he was answering Tom's question as to whether a loop filter would ever kick in. And the answer is that it does. This is evidenced by blocking artifacts in some BD titles using MPEG-2 despite using the high peak rate. Lack of loop filter makes them visible and hence the reason it is part of our codec and that of VC-1.

Now whether the artifacts in case of MPEG-2 or the effect of loop filter, make the format less optimal is a subjective thing for people to go and argue to death, as we have done here at length . Fact is that the performance curve of advanced codes is even more non-linear than MPEG-2. They approach the quality of the source much faster (i.e. at earlier data rates) than MPEG-2 does, and as you increase the data rate, the incremental improvements can be very small. Would you be able to still see them? Only you can answer that. But if you are happy with MPEG-2 at BD rates, then you should be just as happy with VC-1 at HD DVD rates because of its increased efficiency and distortion mitigation. If this is not believed, then I can do another article on differences between VC-1 and MPEG-2.
amirm is offline  
post #2624 of 4841 Old 03-05-2007, 01:20 PM
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,829
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1334 Post(s)
Liked: 749
Oops. Didn't missed your post Ben while I was answering the same question . Sorry about that.
amirm is offline  
post #2625 of 4841 Old 03-05-2007, 01:34 PM
AVS Forum Special Member
 
benwaggoner's Avatar
 
Join Date: Sep 2003
Location: Portland, OR
Posts: 2,630
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 17
Quote:
Originally Posted by amirm View Post

Oops. Didn't missed your post Ben while I was answering the same question . Sorry about that.

Don't let that dissuade you from writing that MPEG-2/VC-1 comparison, though!
benwaggoner is offline  
post #2626 of 4841 Old 03-05-2007, 02:18 PM
AVS Forum Special Member
 
nilsp's Avatar
 
Join Date: Feb 2003
Location: Oslo, Norway
Posts: 1,230
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Amir,

in light of your recent discussion on codecs, one question:

In your opinion, would you say that neither MPEG2 and AVC cannot reach visual transparency using the bitrates avialable to it on Blu-ray?

NilsP

Life is better in HD.
nilsp is offline  
post #2627 of 4841 Old 03-05-2007, 04:41 PM
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,829
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1334 Post(s)
Liked: 749
Quote:
Originally Posted by nilsp View Post

Amir,

in light of your recent discussion on codecs, one question:

In your opinion, would you say that neither MPEG2 and AVC cannot reach visual transparency using the bitrates avialable to it on Blu-ray?

It would be unfair to make a general characterization that way wrt to the experience viewers have. Clearly BD has some really good looking titles now, gaining praise from reviewers and consumers alike. So for these titles, and using the eyes of enthusiasts, I would say BD has achieved visual transparency.

If you are asking me what I personally think when looking at the overall portfolio of titles in BD, then I am left wondering how many of them could be improved with VC-1. It is just our nature to analyze the picture, thinking like a human codec (), spotting things that we know would stress MPEG-2 or AVC when it comes to visual transparency. Having said this, I have to be frank and say that it is unclear whether people could see the visual improvements from VC-1 over these other codecs in every title. What we see should not always be confused with what others see.
amirm is offline  
post #2628 of 4841 Old 03-05-2007, 08:35 PM
Senior Member
 
kitzi's Avatar
 
Join Date: Nov 2005
Posts: 212
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
xboxboi is quoting Amir insinuating that Universal is dropping SD DVD and going combo only for new releases and HD DVD only for catalog? What's the time frame for the change over?
kitzi is offline  
post #2629 of 4841 Old 03-05-2007, 10:07 PM
AVS Forum Special Member
 
Phloyd's Avatar
 
Join Date: Dec 2001
Location: Santa Clara, CA
Posts: 2,846
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by amirm View Post

Rather, he was answering Tom's question as to whether a loop filter would ever kick in.

Ben or Amir (or both ),

If a loop filter is needed for a certain segment to mask blocking, is the image still 'transparent to the master', or does the act of filtering, while less distruptive to the viewer than blocking would be, cause the image to be no longer transparent to the master?

I am enjoying this informative discussion
Phloyd is offline  
post #2630 of 4841 Old 03-05-2007, 11:36 PM
AVS Forum Special Member
 
benwaggoner's Avatar
 
Join Date: Sep 2003
Location: Portland, OR
Posts: 2,630
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 17
Quote:
Originally Posted by Phloyd View Post

If a loop filter is needed for a certain segment to mask blocking, is the image still 'transparent to the master', or does the act of filtering, while less distruptive to the viewer than blocking would be, cause the image to be no longer transparent to the master?

Yes, the output of a loop filter can be visually transparent the master, and certainly a lot closer than without using the loop filter. Part of why loop matters is that it can eliminate a small error which would then propogate into future frames, and so on down through the GOP.

The key is to get enough of an effect to be useful, but not so much as to overly soften.
benwaggoner is offline  
post #2631 of 4841 Old 03-06-2007, 12:20 AM
AVS Forum Special Member
 
dr1394's Avatar
 
Join Date: May 2002
Location: Mizar 5
Posts: 3,470
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 186 Post(s)
Liked: 116
Quote:
Originally Posted by trbarry View Post

Aren't both encoders smart enough to avoid the filtering completely when there is sufficient bit rate to avoid blocks anyway?

- Tom

For H.264, the loop filter is controlled by quantization level Qp. Here's the table for the decision on whether more that a single pixel per block edge is filtered.
Code:
0  1  2  3  4  5  6  7  8  9  10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  2  2  2  3  3  3  3  4  4  4

26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51
6  6  7  7  8  8  9  9  10 10 11 11 12 12 13 13 14 14 15 15 16 16 17 17 18 18
Pixels are labeled like so, with the block edge in the middle.

p3 p2 p1 p0 | q0 q1 q2 q3

The top value is Qp and the bottom value is the level that the amount of pixel difference between p1 and q1 must be less than for the filter to engage.

If the average Qp between the two blocks is 15 or less, the filter does not engage. For Qp 16, the pixel level difference between p1 and q1 must be less then 2. In other words, if the difference between p1 and q1 is larger than 2 (meaning there's some detail in the block), the filter is not used on p1 and q1 (and p2 and q2).

Amir likes to characterize the H.264 loop filter as a "blunt eraser". Looking at the numbers above, I would say that filtering of anything but the block edges is done with a "light touch".

Ron

HD MPEG-2 Test Patterns http://www.w6rz.net
dr1394 is offline  
post #2632 of 4841 Old 03-06-2007, 03:42 AM
AVS Forum Special Member
 
trbarry's Avatar
 
Join Date: Jan 2000
Location: Gainesville FL USA
Posts: 9,884
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Ron -

I don't even know if there is any industry consensus that filtering only 1 vs 3 pixels or doing it with only 4 vs 6 taps is really better anyway. It may be one of those religious issues. But I do know that on the popular open source X264 codec you can enter about 4 different levels for relatively how much loop filtering you want. I'm assuming the commercial codecs can do similarly.

To my questions ... there have indeed been some comments that some AVC encodes tend to look softer than we might expect. Do you feel this is really from the AVC loop filtering, some other reason (other filtering?), or just false?

Also, do you know if most AVC highdef disc encodes are using an 8x8 transform? It seems they would be since that was put into the profile for HD. If so, that would avoid the double loop filtering that has been discussed.

- Tom

Why don't we power our electric cars from greener, cheaper Liquid Fluoride Thorium Reactors?

Tom Barry - Find my video filters at www.trbarry.com
trbarry is offline  
post #2633 of 4841 Old 03-06-2007, 04:47 AM
AVS Forum Special Member
 
dr1394's Avatar
 
Join Date: May 2002
Location: Mizar 5
Posts: 3,470
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 186 Post(s)
Liked: 116
Quote:
Originally Posted by trbarry View Post

Ron -

I don't even know if there is any industry consensus that filtering only 1 vs 3 pixels or doing it with only 4 vs 6 taps is really better anyway. It may be one of those religious issues. But I do know that on the popular open source X264 codec you can enter about 4 different levels for relatively how much loop filtering you want. I'm assuming the commercial codecs can do similarly.

To my questions ... there have indeed been some comments that some AVC encodes tend to look softer than we might expect. Do you feel this is really from the AVC loop filtering, some other reason (other filtering?), or just false?

Also, do you know if most AVC highdef disc encodes are using an 8x8 transform? It seems they would be since that was put into the profile for HD. If so, that would avoid the double loop filtering that has been discussed.

- Tom

Loop filtering "strength" is controlled with an offset for the table (and another table for the block edge pixels) in my previous post. The offset can be up to plus or minus 12 (in steps of two) added to the Qp value. For example (with the posted table), if you set the offset to -12, then the filter won't engage until Qp = 28. This offset can be changed every picture.

I'm working on a trick to show exactly which pixels are being loop filtered on decoded images. Stay tuned.

I don't have much H.264 movie content. On the one clip I do have (Equilibrium), it is High Profile, and the encoder selects 8x8 quite a bit.

Ron

HD MPEG-2 Test Patterns http://www.w6rz.net
dr1394 is offline  
post #2634 of 4841 Old 03-06-2007, 04:57 AM
AVS Forum Club Gold
 
Kosty's Avatar
 
Join Date: Sep 2005
Location: North East USA
Posts: 14,628
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Quote:
Originally Posted by kitzi View Post

xboxboi is quoting Amir insinuating that Universal is dropping SD DVD and going combo only for new releases and HD DVD only for catalog? What's the time frame for the change over?

The direct question was already directly answered by Amirm.

Combos when used will be only used for new releases not previously released on DVD, possibility still of some dual HD DVD and DVD releases still for some of the new titles.

Previously released DVD catalog titles will be released HD DVD only.
Quote:


Originally Posted by brez
Amir,

Have you had any response from the studios on the combos? (on Combos)

Quote:
Originally Posted by Amirm View Post

Glad you asked .

I spoke with Uni's Ken Graffeo this week, who clarified Universal's combo disc strategy. Given that most of us already own DVD catalog titles of our favorite movies, Universal plans to only release new movies in combo disc, and to release catalog titles as stand-alone HD DVDs. That means they'll no longer have tier pricing for combo discs and new releases will be competitively priced with other titles in the market. This is good news for us, since that rationalizes the price points, which I know has been a subject of debate on the Forum.

So you see, the voices of majority do have some power here .

https://www.avsforum.com/avs-vb/showt...&&#post9886623

Quote:
Originally Posted by Amirm View Post

Guys, sorry about causing confusion here. My only job was to find out Uni's HD DVD combo/pricing policy, not DVD. I did not mean to imply anything about them turning off normal DVD production and switching to combos only. I can see how my wording could be construed the way it was though. So hopefully no flogging occurs for folks who were led astray .

https://www.avsforum.com/avs-vb/showt...&&#post9950328

.
"A lot of good arguments are spoiled by some fool who knows what he is talking about." - Miguel de Unamuno


follow me on Twitter
Kosty is offline  
post #2635 of 4841 Old 03-06-2007, 05:05 AM
AVS Forum Special Member
 
dr1394's Avatar
 
Join Date: May 2002
Location: Mizar 5
Posts: 3,470
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 186 Post(s)
Liked: 116
Quote:
Originally Posted by trbarry View Post

Ron -

I don't even know if there is any industry consensus that filtering only 1 vs 3 pixels or doing it with only 4 vs 6 taps is really better anyway.

- Tom

BTW, there is no 6-tap filter in the H.264 loop filter. Only 3, 4 and 5-tap.

Ron

HD MPEG-2 Test Patterns http://www.w6rz.net
dr1394 is offline  
post #2636 of 4841 Old 03-06-2007, 06:21 AM
AVS Forum Special Member
 
BenDover's Avatar
 
Join Date: Nov 2003
Location: NY
Posts: 5,347
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by dr1394 View Post

For H.264, the loop filter is controlled by quantization level Qp. Here's the table for the decision on whether more that a single pixel per block edge is filtered.
Code:
0  1  2  3  4  5  6  7  8  9  10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  2  2  2  3  3  3  3  4  4  4

26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51
6  6  7  7  8  8  9  9  10 10 11 11 12 12 13 13 14 14 15 15 16 16 17 17 18 18
Pixels are labeled like so, with the block edge in the middle.

p3 p2 p1 p0 | q0 q1 q2 q3

The top value is Qp and the bottom value is the level that the amount of pixel difference between p1 and q1 must be less than for the filter to engage.

If the average Qp between the two blocks is 15 or less, the filter does not engage. For Qp 16, the pixel level difference between p1 and q1 must be less then 2. In other words, if the difference between p1 and q1 is larger than 2 (meaning there's some detail in the block), the filter is not used on p1 and q1 (and p2 and q2).

Amir likes to characterize the H.264 loop filter as a "blunt eraser". Looking at the numbers above, I would say that filtering of anything but the block edges is done with a "light touch".

Ron

dr,

according to your explanation, there are instances when the filter is not triggered; how often would you say this occurs, satistically?

when the conditions for filtering ARE met, which pixels get filtered? all three on each side of the boundary? are there conditions that would entail a varied number of pixels to be filtered or is it all or nothing?

also, when all ARE filtered, how does the level of filtering vary wrt distance from the boundary?

thx
BenDover is offline  
post #2637 of 4841 Old 03-06-2007, 06:27 AM
AVS Forum Special Member
 
dr1394's Avatar
 
Join Date: May 2002
Location: Mizar 5
Posts: 3,470
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 186 Post(s)
Liked: 116
Quote:
Originally Posted by BenDover View Post

dr,

according to your explanation, there are instances when the filter is not triggered; how often would you say this occurs, statistically?

when the conditions for filtering ARE met, which pixels get filtered? all three on each side of the boundary? are there conditions that would entail a varied number of pixels to be filtered or is it all or nothing?

also, when all ARE filtered, how does the level of filtering vary wrt distance from the boundary?

thx

I hope to be able to answer these questions graphically with real decoded images by modifying the reference decoder. As I said before, stay tuned.

Ron

HD MPEG-2 Test Patterns http://www.w6rz.net
dr1394 is offline  
post #2638 of 4841 Old 03-06-2007, 07:26 AM
Senior Member
 
kitzi's Avatar
 
Join Date: Nov 2005
Posts: 212
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
[quote=Kosty]The direct question was already directly answered by Amirm.

Combos when used will be only used for new releases not previously released on DVD, possibility still of some dual HD DVD and DVD releases still for some of the new titles.

QUOTE]


Are you an insider? Possibility that there will still be DVD and HD DVD new releases...are you implying that they are dropping DVD for most of their new releases and going combo only? that's not how I read what Amir said at all...
kitzi is offline  
post #2639 of 4841 Old 03-06-2007, 07:50 AM
Oppo Beta Group
 
RichB's Avatar
 
Join Date: Nov 1999
Location: Massachusetts
Posts: 10,924
Mentioned: 28 Post(s)
Tagged: 0 Thread(s)
Quoted: 1756 Post(s)
Liked: 1258
With HD DVD we have Mandatory Managed Copy.

Really, How do I make one?

- Rich

Oppo UPD-205 x 2 | UPD-203 | Sonica DAC | Emotiva XMC-1 (v3) | Revel Salon2s, Voice2, Studio2s | Benchmark AHB2 x 2 | Velodyne HGS-15 | LG 65C7 | Lumagen 2020 | HDFury Vertex x 2
RichB is online now  
post #2640 of 4841 Old 03-06-2007, 08:23 AM
Oppo Beta Group
 
RichB's Avatar
 
Join Date: Nov 1999
Location: Massachusetts
Posts: 10,924
Mentioned: 28 Post(s)
Tagged: 0 Thread(s)
Quoted: 1756 Post(s)
Liked: 1258
Quote:
Originally Posted by JBlacklow View Post

Both formats have Managed Copy. Mandatory is up to the studios not the formats, and AFAIK Managed Copy hasn't been implemented yet.

Not to quibble, but I did not name it "Mandatory Managed Copy" that was done by the HD DVD / BD crowd. Since they named it, I was hoping for an insider to explain this. You would think this would be a good time when there is now an unmanaged copy option.

- Rich

Oppo UPD-205 x 2 | UPD-203 | Sonica DAC | Emotiva XMC-1 (v3) | Revel Salon2s, Voice2, Studio2s | Benchmark AHB2 x 2 | Velodyne HGS-15 | LG 65C7 | Lumagen 2020 | HDFury Vertex x 2
RichB is online now  
Sponsored Links
Advertisement
 
Closed Thread HDTV Software Media Discussion

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off