Ask questions to Compression Engineer insider "drmpeg"

[iceman]

This thread is dedicated to asking questions of industry insider "drmpeg" (Compression Engineer insider) who has graciously taken the time to participate here. drmpeg and all our insiders do this out of their free time and to try to keep us informed to their best abilities, and therefore are to be treated with respect and courtesy.

Before asking a question, please skim at least the last weeks worth of posts in order to make sure that the question hasn't already been asked or answered. Using the search function is also always a good idea. Please conduct your inquiries in a professional manner and avoid asking "chicken little" questions or asking when unannounced titles will come out.

drmpeg - Compression Engineer - Track posts

[irfoton]

drmpeg,
I asked this in the "Judgement Day" thread but I'm sure this was overcome by the amount of posts.

As HD takes hold and begins to eclipse DVD, how will compressionists be trained? There appears to be more choices than ever for HD. Which codec is the most forgiving? Which provides the best results in the least amount of effort? Which provides the best results when you have all the money and time in the world?

Thanks

[scott1256ca]

Have you done any, or are aware of any side by side comparisons of VC-1 vs. AVC at BD bitrates? If so, can you elaborate on the pros and cons of each?
Are there any hard "this one is better at these bitrates" types of statements you can make?

Thanks

[scott1256ca]

As a followup to my post, how close are either of these codecs to "set it and forget it" when it comes to encoding a 2 hr. 1.85 movie?
Any expectations on when this will be doable with uniformly excellent results, or is that still quite a ways off? Which of the 2 codecs seems to have more room for future growth in this area?

Thanks

[DaViD Boulet]

we see a lot of HD DVD/Amir-speak suggesting that the bit-rates on BD above 20mbps are not needed for video transparency.

Can you comment? Even if transparency can be acheived by careful compresson, are there cost-savings with making things easier by opening up the bit-rate during encoding? Do AVC and VC-1 behave similarly at the various bit-rate levels? Are there advantages with one over the other in different applications?

[PaulGo]

Along the same theme - Now that Warner is going Blu-ray exclusive do you see them upping the bitrate of the VC-1 codec (will it do and good?) or do yousee them going to AVC. Would their be a difference in picture quality (for the better) if they switched?

[drmpeg]

Originally Posted by DaViD Boulet

we see a lot of HD DVD/Amir-speak suggesting that the bit-rates on BD above 20mbps are not needed for video transparency.

Can you comment? Even if transparency can be achieved by careful compression, are there cost-savings with making things easier by opening up the bit-rate during encoding? Do AVC and VC-1 behave similarly at the various bit-rate levels? Are there advantages with one over the other in different applications?

Depends on where your coming from. From a broadcast perspective, bitrates over 20 Mbps are astronomical. Everybody is trying to develop encoders that will go as low as possible and anything over 10 Mbps is considered high.

But blue laser is a different animal. IMHO, there's no reason not to use all the available bits on the medium (unless the movie is just too short or not complex enough). Higher bitrates are always better than lower bitrates. You may not be able to perceive the difference, but why worry about that if you have the bit budget?

Although I work at an H.264 company and know more about it, there's no doubt that VC-1 is an excellent codec. Given some of the high bitrate encodes we've seen lately (like Ratatouille), it's seems like a waste of energy trying to declare a winner. The real winner is the consumer. You're getting some truly "transparent" encodes (aside from 8-bit and 4:2:0 issues) with both codecs, and you should be ecstatic that the blue laser formats and advanced codecs make this possible.

Ron

[drmpeg]

Originally Posted by scott1256ca

As a followup to my post, how close are either of these codecs to "set it and forget it" when it comes to encoding a 2 hr. 1.85 movie?
Any expectations on when this will be doable with uniformly excellent results, or is that still quite a ways off? Which of the 2 codecs seems to have more room for future growth in this area?

Thanks

Creating a good multi-pass encoder is more about gathering the most useful statistics during the first pass and then applying a heuristic algorithm to those statistics that can arrive at a constant quality encode (for a particular average and peak bitrate). It's fairly codec independent (it's more of an entropy issue), although the algorithm does have to have a good idea of the codecs quality versus bitrate curve.

There's no doubt that higher average and especially peak bitrate make things easier. Both VC-1 and H.264 benefit from the higher rates available on Blu-ray.

As for future growth, there's no doubt that H.264 offers more areas to explore. MPEG-2 saw incremental gains every year over the last 12 years, and it's expected that H.264 will be very much the same.

Ron

[drmpeg]

Originally Posted by scott1256ca

Have you done any, or are aware of any side by side comparisons of VC-1 vs. AVC at BD bitrates? If so, can you elaborate on the pros and cons of each?
Are there any hard "this one is better at these bitrates" types of statements you can make?

Thanks

Well, the studios do this all the time, but the results never get published. As for doing a public "shootout", there's many pitfalls to the concept. First, there's very little copyright free uncompressed content available. Certainly no full length movies. Even when there is some content available (like Elephants Dream), everyone complains it's not a real movie.

Second, there's always the "you didn't use our latest and greatest encoder" excuse. Since there's always another latest and greatest on the way, it's very easy to trivialize the results of any shootout.

I've seen some of the statements made on AVS about AVC looking better at high bitrates and VC-1 looking better at low bitrates. At low bitrates, the encoder implementation will most likely overshadow any particular codec features. At high bitrates, H.264 may have an advantage since the loop filter can be entirely disabled (which seems to be the case for many current movies) and more high frequency detail can be coded by the use of flatter scaling matrices. However, my guess is that both codecs look remarkably alike at high bitrates.

Ron

[aygie]

Hi Ron

do you have any involvement with UK TV at all?

[phloyd]

Originally Posted by drmpeg

At high bitrates, H.264 may have an advantage since the loop filter can be entirely disabled (which seems to be the case for many current movies) and more high frequency detail can be coded by the use of flatter scaling matrices.

This is interesting since Ben W from *cough*soft has stated a number of times that at the bitrates used for HD Media the loop filter is needed. It appears that his statement is incorrect based on these real examples - something I am happy to see (not that he was wrong, but that the loop filter is not always neeed).

[drmpeg]

Originally Posted by phloyd

This is interesting since Ben W from *cough*soft has stated a number of times that at the bitrates used for HD Media the loop filter is needed. It appears that his statement is incorrect based on these real examples - something I am happy to see (not that he was wrong, but that the loop filter is not always needed).

I've created a version of the reference decoder that reports the number of filtered luma pixels, so that folks can check movies (to see if the loop filter is disabled) for themselves. Here's the post on AVSForum:

http://www.avsforum.com/avs-vb/showt...1#post12558851

Ron

[benwaggoner]

Originally Posted by phloyd

This is interesting since Ben W from *cough*soft has stated a number of times that at the bitrates used for HD Media the loop filter is needed. It appears that his statement is incorrect based on these real examples - something I am happy to see (not that he was wrong, but that the loop filter is not always neeed).

Did I? I don't recall saying that specifically about H.264 and BD. If you really want to max out the bitrate and accept that reduction in runtime/extras, I agree with drmpeg that turning it off would be feasiable, and probably useful.

This isn't something we've sweated with VC-1 since ours is more focused, and so kind of fades away at higher bitrates without needed to be explicitly turned off.

Also, the loop filter isn't something that needs to be on/off for a whole movie. It can be turned on or off for different sections, at least in our VC-1 encoder.

[drmpeg]

Originally Posted by irfoton

drmpeg,
I asked this in the "Judgement Day" thread but I'm sure this was overcome by the amount of posts.

As HD takes hold and begins to eclipse DVD, how will compressionists be trained? There appears to be more choices than ever for HD. Which codec is the most forgiving? Which provides the best results in the least amount of effort? Which provides the best results when you have all the money and time in the world?

Thanks

To be honest, I don't have much exposure to compressionists, since my background is almost entirely in real-time encoders for broadcast applications.

As for codecs, it's very difficult to compare VC-1 against H.264 in a pure apples to apples fashion. The problem is that you can only evaluate encoder implementations of each standard. There are many algorithms, techniques and tricks (also known as "secret sauce") that are outside of the codec standard that encoder developers can put in their product.

The best encoders are the result of a lot of time and effort. That's why Microsoft's VC-1 encoder does so well. MS was willing to spend the time and effort to make a great authoring encoder. Of course, since VC-1 was pretty much rejected in the broadcast space, MS more or less had to create a good authoring encoder since blue laser was the last large segment they could enter.

In the H.264 space, many of the very best compression folks are working on real-time broadcast encoders and blue laser authoring encoders are pretty much being ignored by these companies. There's just not enough volume to justify the R&D (one encoder can create a lot of titles). We're lucky that at least a few companies (like Panasonic) were willing to develop an H.264 authoring encoder. BTW, Panasonic was very active in the development of the H.264 standard. Almost all of the the High Profile features (like 8x8 transform and scaling matrices) were suggested by them.

Ron

[Penton-Man]

Originally Posted by drmpeg

To be honest, I don't have much exposure to compressionists, since my background is almost entirely in real-time encoders for broadcast applications.

Aren't you the Jerry Garcia look-a-like ?

[drmpeg]

Originally Posted by Penton-Man

Aren't you the Jerry Garcia look-a-like ?

Yes, Jerry Garcia, Tommy Chong and the guy on the cover of Zig-Zag papers.

Ron

[drmpeg]

Originally Posted by benes

This is the greatest thread ever.

I see a lot of talk about 4:4:4 content in the future. But I have another pie-in-the-sky question for you. Do you think it will ever be feasible to deliver content to the home with INTRAframe video compression? This is what is used for D-Cinema and it basically eliminates all motion compression artifacts.

I know blu-ray doesn't support this and it would take an incredible bitrate. But maybe in the future when we all have the 100Mbps fiber connections that amir is always talking about.

There's nothing magical about intra-only encoding. It can also have artifacts, but they will be spatial in nature rather than caused by motion (or more precisely, inadequate motion estimation). With either intra or inter encoding, artifacts are still a function of bitrate.

The big problem with intra-only is that artifacts can happen even during still scenes that have large spatial complexity. Because the viewer is more sensitive to this (there's no "temporal masking"), the peak bitrate has to be really high. I believe D-Cinema goes up to 250 Mbps (although it can be VBR).

As for high bitrates to the home, I'll have to say that "that's beyond the scope of this class". My gut feel about movie downloads is that too many folks believe it's the next big thing. Usually, the truly cool next big thing is less obvious.

Ron

[damonous]

Originally Posted by drmpeg

To be honest, I don't have much exposure to compressionists, since my background is almost entirely in real-time encoders for broadcast applications. ...

drmpeg (Ron), thanks for taking the time out for some great info on this Forum. I have a couple of long-burning questions you might be able to answer...


1) Since your background is broadcast application, what do you know of Cable; specifically TWC? I've read unsubstantiated info in the past that Cable providers sometimes re-encode MPEG streams to fit their bandwidth needs. I never used to think this, because compared to my friends' Dish/satellite HD, I've always felt Cable was superior.

But, of late, certain things on TWC in my area (Cleveland) has been looking VERY inconsistent, with some very unexpected 'smelly sights'. In particular, the Super Bowl looked atrocious with gobs of messy macroblocking. Ditto to the Grammys. Other times, I see some absolutely spectacular HD, like Prison Break on Fox (in 1080i even, according to my TV's indicator).

In total, though, all channels seem to be worse (more macroblocking) than in the past. Is it possible that TWC is re-encoding and selectively cranking-down the bits?


2) You posted a page or two back about banding, and I'm curious about what it the primary cause. Is it the bit depth for color, or bit depth for encode, or the codec itself?

Case in point: I recently rented Sunshine on BD. I've always been VERY fond of the quality of Fox/MGM BD's, owning about 15.

For the most part (like 99.99999% ), this was another stunning transfer with spiderweb-fine lines clearly depicting the spaceship's communication arrays, and excellent compression and greyscale control (especially for a film advertised on the sleeve as AVC at only 14mbps!).

Yet, there is one absolutely atrocious moment at the 3/4 mark of the movie, where the ship is moving in space against an eclipse of Mercury against the Sun. Suddenly, a frightening blooming mush of muddy 256-color-looking orange halo around the ship (where a proper haze/mist cloud should have been).

It's a BD50, --so what gives? ...Was the compressionist asleep at the wheel? Is it a limitation of AVC? Is it a limitation of available colorspace?

[drmpeg]

Originally Posted by damonous

drmpeg (Ron), thanks for taking the time out for some great info on this Forum. I have a couple of long-burning questions you might be able to answer...


1) Since your background is broadcast application, what do you know of Cable; specifically TWC? I've read unsubstantiated info in the past that Cable providers sometimes re-encode MPEG streams to fit their bandwidth needs. I never used to think this, because compared to my friends' Dish/satellite HD, I've always felt Cable was superior.

But, of late, certain things on TWC in my area (Cleveland) has been looking VERY inconsistent, with some very unexpected 'smelly sights'. In particular, the Super Bowl looked atrocious with gobs of messy macroblocking. Ditto to the Grammys. Other times, I see some absolutely spectacular HD, like Prison Break on Fox (in 1080i even, according to my TV's indicator).

In total, though, all channels seem to be worse (more macroblocking) than in the past. Is it possible that TWC is re-encoding and selectively cranking-down the bits?

HD encoders are too expensive for head-ends to re-encode all of their HD channels. Instead, they use rate-shaping boxes called "cherry pickers" that reduce the bitrate of the incoming streams by re-quantizing the macroblocks in the MPEG-2 video elementary stream.

The only way to really tell what your cable company is up to is to capture some cable QAM bitstreams and compare them to an OTA capture made at the same time. If you can capture both versions, let me know, and I can analyze them for you.

Originally Posted by damonous

2) You posted a page or two back about banding, and I'm curious about what it the primary cause. Is it the bit depth for color, or bit depth for encode, or the codec itself?

Case in point: I recently rented Sunshine on BD. I've always been VERY fond of the quality of Fox/MGM BD's, owning about 15.

For the most part (like 99.99999% ), this was another stunning transfer with spiderweb-fine lines clearly depicting the spaceship's communication arrays, and excellent compression and greyscale control (especially for a film advertised on the sleeve as AVC at only 14mbps!).

Yet, there is one absolutely atrocious moment at the 3/4 mark of the movie, where the ship is moving in space against an eclipse of Mercury against the Sun. Suddenly, a frightening blooming mush of muddy 256-color-looking orange halo around the ship (where a proper haze/mist cloud should have been).

It's a BD50, --so what gives? ...Was the compressionist asleep at the wheel? Is it a limitation of AVC? Is it a limitation of available colorspace?

It's mostly a problem with 8-bit video, but the encoder (of any codec, not just AVC) can make it worse. A look at the quantization levels for those frames would show if the encoder was making things worse.

Without seeing the 10-bit source, it's difficult to know what the compressionist was thinking. The compressionist may have taken a look at an 8-bit version of the source and decided the encode was faithful (enough) to that.

Ron

[kapitalisten]

Ron,

may one inquire about what kind of tools you use to visualize quantization levels, residuals and so forth - or is that a "secret of the trade"?