x264 vs other H.264 encoders.

[Chengbin]

Let the encoder movie studios use for their Blu-ray releases called encoder x (I don't know what they use).

If x264 and encoder x get the same master print of a movie as the source and are asked to encode a 1080p version for Blu-ray. Would x264+avisynth be better than encoder x?

If x264 is better, what makes the movie studios use encoder x?

For x264. what kind of bitrate is need for a 1080p encode to reach transparency? How much difference is there with a 15mbps encode, using very slow x264 settings, and a 30mbps encode that's DXVA compatible?

DS, can you comment on how good the encodes are on a Blu-ray with the amount of bitrate they have? I know it varies, but most Blu-rays are around 25-30mbps.

[kemuri-_9]

x264 can't produce BD compliant streams (missing hrd and slices), so that's why companies use 'x' instead of x264.

[Chengbin]

What do hrd and slices do?

I'm reading a comparison of software H.264 encoders and what features they support on Wikipedia.

x264 doesn't support SI and SP Slices, FMO, ASO, RS, data partitioning, 4:0:0, 4:2:2, 4:4:4, 9-14 bit sample depth, separate color plane coding, and film grain modeling.

Are these features not supported because they're pointless or very difficult to code (or both)?

[LoRd_MuldeR]

What do hrd and slices do?

HRD is related to interlacing. Slices divide each frame into several parts, where each part can be decoded independently.

Using slices allows multi-threaded decoding without processing several frames in parallel. They also allow better error resilience. But they hurt compression efficiency!

The latter is the reason why x264 doesn't use slices anymore...

[Dark Shikari]

Would x264+avisynth be better than encoder x?Yes, regardless of what X is, because there is [in my experience] no encoder in the world that produces better general-purpose output than x264. There are encoders that might be better in specific situations, such as low-delay encoding, 1-pass CBR encoding, or with regards to features x264 doesn't support (4:2:2, or 10-bit). However, none of these are relevant to Blu-ray encoding.If x264 is better, what makes the movie studios use encoder x?Movie studios know absolutely nothing about encoding, as they have proven for a decade now with utterly incompetently encoded DVDs.For x264. what kind of bitrate is need for a 1080p encode to reach transparency?Depends on the quality and graininess of the source.DS, can you comment on how good the encodes are on a Blu-ray with the amount of bitrate they have? I know it varies, but most Blu-rays are around 25-30mbps.Blu-rays are generally crap. In my experience the majority either have way too low a video bitrate (in which case the 1080ps have hardly even 720p of detail) or a terrible source (in which case they're often no better than SD). The encoders they use are generally really terrible, which is why VC-1 is still used: the H.264 encoders are so bad that Microsoft's VC-1 actually gives decent results by comparison.

x264 would require significantly more bitrate for transparency if encoding from a lossless source than from a Blu-ray. Blu-rays generally seem to have very little high-frequency content, almost as if they were lowpassed. I wouldn't be surprised if they were--and even if they aren't, the quantizatoin matrices they use often do so for them anyways.HRD is related to interlacing.No it isn't.

[Manao]

HRD means hypothetical reference decoder. It's a way to check that a stream respect some mandatory constraints (VBV/CPB and DPB). HRD is initialized using picture timing SEI, which x264 (unpatched) doesn't write. As it happens, picture timing SEI, in addition to HRD initialization, also contains the picture structure information - which indicates pulldown / interlacing flags, thus LoRd_Mulder's mistake.

[LoRd_MuldeR]

HRD means hypothetical reference decoder. It's a way to check that a stream respect some mandatory constraints (VBV/CPB and DPB). HRD is initialized using picture timing SEI, which x264 (unpatched) doesn't write. As it happens, picture timing SEI, in addition to HRD initialization, also contains the picture structure information - which indicates pulldown / interlacing flags, thus LoRd_Mulder's mistake.

Oh, I see. With Trahald's "hrd_pulldown" patch, does x264 produce a HRD compliant stream then?

[Manao]

I haven't checked with an analyzer, but it should.

[LoRd_MuldeR]

I haven't checked with an analyzer, but it should.

So the lack of slice support is the only thing left that prevents x264 from producing BD compatible streams.

Can slice support be ported back to current x264 in a way that doesn't replace the frame-based multithreading?

[Inspector.Gadget]

Hang on a second. Both MeGUI and BD Rebuilder use x264 to re-encode Blu-ray video for later Blu-ray use. So wouldn't it be correct to say that x264 doesn't support all the features in the spec, rather than that it is unable to produce BD-compatible streams?

[Dark Shikari]

Hang on a second. Both MeGUI and BD Rebuilder use x264 to re-encode Blu-ray video for later Blu-ray use. So wouldn't it be correct to say that x264 doesn't support all the features in the spec, rather than that it is unable to produce BD-compatible streams?No, it can produce BD-compatible streams.

The streams technically don't fall within the spec (no slices), but they work on all the players anyways.

[Fr4nz]

x264 would require significantly more bitrate for transparency if encoding from a lossless source than from a Blu-ray. Blu-rays generally seem to have very little high-frequency content, almost as if they were lowpassed. I wouldn't be surprised if they were--and even if they aren't, the quantizatoin matrices they use often do so for them anyways.No it isn't.

Hi Dark, what is the format for lossless sources that movie studios use?

[Dark Shikari]

Hi Dark, what is the format for lossless sources that movie studios use?Depends. I suspect it's often just raw 10-bit 4:4:4, often stored in proprietary formats like v210 or similar.

[Chengbin]

In my experience the majority either have way too low a video bitrate (in which case the 1080ps have hardly even 720p of detail) or a terrible source (in which case they're often no better than SD). The encoders they use are generally really terrible, which is why VC-1 is still used: the H.264 encoders are so bad that Microsoft's VC-1 actually gives decent results by comparison.

x264 would require significantly more bitrate for transparency if encoding from a lossless source than from a Blu-ray. Blu-rays generally seem to have very little high-frequency content, almost as if they were lowpassed.

I've never seen any Blu-rays that are not significantly better than SD. Most Blu-rays I've seen are excellent 1080p encodes (maybe not for the bitrate, but it definitely looks very good). I'm just curious, what Blu-rays have you seen to make that judgment?

I thought the human eye isn't very sensitive to high frequency content, so encoders can compress HF content harsher and it won't look that bad.

Yes, that's what I mean, from a lossless source. So can you give me a range of bitrates for transparency? Is 40mbps enough?

[Fr4nz]

Depends. I suspect it's often just raw 10-bit 4:4:4, often stored in proprietary formats like v210 or similar.

Thank for your answer. Then another question: in your opinion, what is the (medium) bitrate needed to encode losslessly their sources with their proprietary formats?

[Dark Shikari]

I've never seen any Blu-rays that are not significantly better than SD.Close Encounters of the Third Kind is no better than SD, at least from the screenshots I saw.

Of course, also remember that most DVDs aren't that great either, and so they don't make very good examples of "perfect" SD video.Most Blu-rays I've seen are excellent 1080p encodes (maybe not for the bitrate, but it definitely looks very good). I'm just curious, what Blu-rays have you seen to make that judgment?The worst example I've seen is Cowboy Bebop: The Movie, but all of them in general seem significantly less sharp than lossless footage like parkrun. And I don't have to see them to know that an awful H.264 encoder at 20 megabits cannot make 1080p anywhere near perfect.
I thought the human eye isn't very sensitive to high frequency content, so encoders can compress HF content harsher and it won't look that bad.If define the "B" values as high frequency in the DCT as follows:

A A A A B B B B
A A A A B B B B
A A A A B B B B
A A A A B B B B
B B B B B B B B
B B B B B B B B
B B B B B B B B
B B B B B B B B

And you remove them entirely, the result is exactly equivalent to cutting the resolution of the video in half. Saying "high frequency data isn't important" is equivalent to saying "high resolution video isn't important."Thank for your answer. Then another question: in your opinion, what is the (medium) bitrate needed to encode losslessly their sources with their proprietary formats?Bitrate? Did you read where I said "raw 10-bit 4:4:4"?

[Fr4nz]

Bitrate? Did you read where I said "raw 10-bit 4:4:4"?

Please:

don't be so harsh, I didn't offend you and I was very kind when I made my question to you.

[Dark Shikari]

Please:

don't be so harsh, I didn't offend you and I was very kind when I made my question to you.I didn't mean to offend. I was just pointing you to the relevant information (rather than answering a question that had no real valid answer).

[Fr4nz]

I didn't mean to offend. I was just pointing you to the relevant information (rather than answering a question that had no real valid answer).

Okay :)

[benwaggoner]

Movie studios know absolutely nothing about encoding, as they have proven for a decade now with utterly incompetently encoded DVDs.
I don't think that's fair at all. I've worked with a number of studio compressionists, and the best among them are very good indeed, and have developed some great techniques.

An argument can be made that they're too slow in adoping new encoders; I've seen some oriignal Toshiba MPEG-2 encoders from around the DVD launch still in use. But they have formal QA processes, get mulitple eyes on the encode, etcetera.

That said, different studios and post houses vary in their techniques, and A-list titles get a whole lot more TLC than the long-tail stuff. But I've compared the source with the output for may studio HD titles, and they often verge on the transparent.

Depends on the quality and graininess of the source.Blu-rays are generally crap. In my experience the majority either have way too low a video bitrate (in which case the 1080ps have hardly even 720p of detail) or a terrible source (in which case they're often no better than SD).
Any particular studios/titles you're thinking of for reference?

But yes, source quality is always a problem. And there's the big "grain reduction" debate. In some circles, it is felt that having an HD "pop" results in better reviews from the home theater magazines, and so the masters can wind up denoised and then sharpened. In other cases, too much degrain is used to provide a cleaner image, even if sharpening isn't done.

The encoders they use are generally really terrible, which is why VC-1 is still used: the H.264 encoders are so bad that Microsoft's VC-1 actually gives decent results by comparison.
:).

I think a big part of that is in workflow, particularly the dithering algorithm and management and previewing of segment reencoding. We also tuned it for HD DVD (0.6 sec GOP, peak bitrates below 30 Mbps), making Blu-ray a relatively luxurious spec to work within.

But real-world quality issues are much less a problem of codec than of source and workflow.

Also, why do you think Blu-ray rates are too low? There are plenty with 30 Mbps ABR, with PBR approaching 40 Mbps. That's far more bits/pixel than H.264 gets in other domains. Blu-ray was designed for MPEG-2 bitrates; modern codecs were a bonus taht came later..

x264 would require significantly more bitrate for transparency if encoding from a lossless source than from a Blu-ray. Blu-rays generally seem to have very little high-frequency content, almost as if they were lowpassed. I wouldn't be surprised if they were--and even if they aren't, the quantizatoin matrices they use often do so for them anyways.
Likely aggressive degraining.

That said, it's easy to overestimate how much real "signal" visual detail there really is in a typical 35mm production, particularly anything shot before the last round of low-grain flim stocks or now all-digital workflows. Heck, some of the early digital cinema cameras actually captured at 1280x1080 anamorphic. Lack of detail often reflects the negative more than many realize.

One of the upsides of grain is that the texture adds as lot of perceptual sharpness, one of the reasons VP6 adds it in postprocessing. And so when it gets degrained, the underlying image is revealed as rather soft, hence sharpening that gets applied.

My opinion is that film is film, and people just just get used to it. The goal of compression is to make content "as much like itself as possible" and heavy preprocessing outside of creative intent isn't part of that.

[Dark Shikari]

Also, why do you think Blu-ray rates are too low? There are plenty with 30 Mbps ABR, with PBR approaching 40 Mbps. That's far more bits/pixel than H.264 gets in other domains. Blu-ray was designed for MPEG-2 bitrates; modern codecs were a bonus taht came later.I've seen lists of Blu-ray bitrates, and most seem to be in the mid-20s or even below 20 megabits average. Assuming that many of these encoders are at least 30-70% worse than x264, that surely can't be anywhere near transparent.That said, it's easy to overestimate how much real "signal" visual detail there really is in a typical 35mm production, particularly anything shot before the last round of low-grain flim stocks or now all-digital workflows. Heck, some of the early digital cinema cameras actually captured at 1280x1080 anamorphic. Lack of detail often reflects the negative more than many realize.

One of the upsides of grain is that the texture adds as lot of perceptual sharpness, one of the reasons VP6 adds it in postprocessing. And so when it gets degrained, the underlying image is revealed as rather soft, hence sharpening that gets applied.

My opinion is that film is film, and people just just get used to it. The goal of compression is to make content "as much like itself as possible" and heavy preprocessing outside of creative intent isn't part of that.That's part of the problem as well. They need to start using real HD digital cameras ;)

[Forteen88]

List with bitrate of Blu-ray movies:
http://www.blu-raystats.com/Stats/

Often pretty low bitrates, as DS says.

EDIT: Blackbeard - MPEG2 13 Mbps :P
300 - VC-1 16.8 Mbps. Pretty low bitrate for such a grainy movie.

[Dark Shikari]

List with bitrate of Blu-ray movies:
http://www.blu-raystats.com/Stats/

Often pretty low bitrates, as DS says.10.5: Apocalypse MPEG2 1080p 1.78:1 14.9 Mbps :scared:

[LoRd_MuldeR]

Why in the world do they use these low bitrates, if they have an entire 50 GB disc available? So many extras on the disc?

[ajp_anton]

50GB is more expensive than 25GB.
The question is, why don't they use the entire 25/50GB of the disc (including extras)?

[Chengbin]

50GB is more expensive than 25GB.
The question is, why don't they use the entire 25/50GB of the disc (including extras)?

+1

The Dark Knight takes up 37GB on Blu-ray, why can't they use up the full 50GB and give like 30mbps to the movie? Nevertheless the movie looked very good.

[benwaggoner]

I've seen lists of Blu-ray bitrates, and most seem to be in the mid-20s or even below 20 megabits average. Assuming that many of these encoders are at least 30-70% worse than x264, that surely can't be anywhere near transparent.
Why would we assume 30-70% lower? Not arguing, but that seems a pretty specific assumption.

A lot of Blu-ray titles wind up having In-loop deblocking turned off for most of the title in order to eliminate the softening effect. That seems a little...aggressive to me, but that certainly would have an impact.

But anyway, I suspect you'd find that actual discs's defects are found in the IYUV masters more often than not.

But even given 20 Mbps and 70%, and using power of 0.75 as a rule of thumb, that'd be still be 14 Mbps 1080p, equivalent to 7.6 Mbps ABR 720p with x264, which should be well more than ample.

That's part of the problem as well. They need to start using real HD digital cameras ;)
They are, but we're still going to be watching movies shot with all kinds of cameras for decades to come.

Even I find excellent compresibility poor consolation for a bad story or poor acting :).

[Chengbin]

Can a 15mbps H.264 encode match a 40mbps MPEG 2 encode?

[Dark Shikari]

Why would we assume 30-70% lower? Not arguing, but that seems a pretty specific assumption.We know for a fact that many H.264 encoders used in Blu-rays were significantly worse than VC-1. We also know that x264 wipes the floor with VC-1 by a very large margin, so the advantage of x264 over those H.264 encoders must be even greater.

I have seen Blu-rays that don't even use partitions or adaptive transform.

Part of the problem IMO is that Blu-ray requires 1080p, but many sources barely deserve 720p, so they're forced to encode a low bitrate 1080p stream rather than do the correct thing and provide only 720p.But even given 20 Mbps and 70%, and using power of 0.75 as a rule of thumb, that'd be still be 14 Mbps 1080p, equivalent to 7.6 Mbps ABR 720p with x264, which should be well more than ample.14 megabit 1080p with x264 is good, but not perfect: you certainly won't get perfect source transparency at that bitrate, especially for something really grainy like 300.Even I find excellent compresibility poor consolation for a bad story or poor acting :).That's why we compress movies rather than watch them, because they're such a waste of time these days ;)

[IgorC]

Something like this also happens to audio CD.
I think the main reason is piracy.
The best period of CDs was 1995-2002. Pirates weren't so active, there wasn't fast internet. So industy was interesting in quality of production to sell.
Now when CDs sells had fallen musician can't pay for expensive services of high qualified audio and electronic professionals.

There are still musicians who cares about quality but they must manage with limited budget and try to get some incomes from tours, marketing etc...

[CruNcher]

List with bitrate of Blu-ray movies:
http://www.blu-raystats.com/Stats/

Often pretty low bitrates, as DS says.

EDIT: Blackbeard - MPEG2 13 Mbps :P
300 - VC-1 16.8 Mbps. Pretty low bitrate for such a grainy movie.

Though not seen the master 300 looks good to me :) but it's no secret anymore that x264 can reduce this VC-1 to around 9 mbit without losing any of this quality (utilizing all psy features) which lets you wonder how would a x264 Blu-Ray looked like in comparison utilizing the full 16.8 mbps or even more :D
The problem is we can't imagine as we didn't saw the master ;) but we know that + 8 mbit almost double the bitrate would be hell a lot off additional details that VC-1 must have thrown away here :P (if those detail existed in the first place as we know the grain is artificial and how much more details the camera captured can be only compared on the master)

Though im quite sure if just 1 Studio would decide yeah come on lets test x264 and release 1 Film with it the Feedback from the extreme viewers like on AVS-Forum would be best Blu-Ray ever and Microsoft would shake in their Shoes if the other H.264 studios then change to x264 (though that's just a dream contracts,money and x264 being not commercial backed won't make this happen so soon but who knows maybe some boss eats one day the blue pill and crazy stuff happens) ;)

[Dark Shikari]

Sorry but does it look bad to you i mean 300 looks hell good :)Good?! 300 was covered in blocking in some dark areas.

You can hardly claim that you have enough bitrate when you have grainy blocks decimated to pure-DC blocks in the middle of a low-motion scene.

[benwaggoner]

We know for a fact that many H.264 encoders used in Blu-rays were significantly worse than VC-1. We also know that x264 wipes the floor with VC-1 by a very large margin, so the advantage of x264 over those H.264 encoders must be even greater.
...assuming that the same ratios apply at these higher bitrates and frame sizes, and that it's codec differences and not workflow or preprocessing. Lots of the advanages of H.264 in general matter less as bitrates go up, like CABAC and in-loop deblocking. Plus GOP lengths are short and the VBR/ABR ratio is a lot smaller than for file distribution.

No doubt x264 has a big advantage in "lowest bitrate without objectionable artifacts" but that's not the domain being discussed. I don't know that a good MPEG-2 encoder would have more than a 70% disadvantage at higher bitrates like this.

Part of the problem IMO is that Blu-ray requires 1080p, but many sources barely deserve 720p, so they're forced to encode a low bitrate 1080p stream rather than do the correct thing and provide only 720p.14 megabit 1080p with x264 is good, but not perfect: you certainly won't get perfect source transparency at that bitrate, especially for something really grainy like 300.
But 300 is a pretty extreme edge case; digital production, highly stylized, heavily sharpened and grainy, desatured and colorized. It's fun to do codec testing on it because it makes things so stark, but it's far from typical of general film/video content.

[honai]

Something like this also happens to audio CD.
I think the main reason is piracy.
The best period of CDs was 1995-2002. Pirates weren't so active, there wasn't fast internet. So industy was interesting in quality of production to sell.

No, it wasn't. During 1995 to 2002 the Loudness War was already waging in full force.

I guess in terms of picture quality we are now in the middle of the DNR war.

[Dark Shikari]

...assuming that the same ratios apply at these higher bitrates and frame sizes, and that it's codec differences and not workflow or preprocessing. Lots of the advanages of H.264 in general matter less as bitrates go up, like CABAC and in-loop deblocking. Plus GOP lengths are short and the VBR/ABR ratio is a lot smaller than for file distribution.

No doubt x264 has a big advantage in "lowest bitrate without objectionable artifacts" but that's not the domain being discussed. I don't know that a good MPEG-2 encoder would have more than a 70% disadvantage at higher bitrates like this.I'm not sure I would call 20 megabits that high, especially with such a short GOP length and the kind of restrictions that Blu-ray forces.

From what I've seen Blu-ray quantizers are generally around 15-25, which is well above the range at which CABAC and other such tools become meaningless.

[benwaggoner]

No, it wasn't. During 1995 to 2002 the Loudness War was already waging in full force.

I guess in terms of picture quality we are now in the middle of the DNR war.
The middle? I hope we're near the end; this started a decade ago with DVD. We can pick up on it more with HD, but DVDs got tons of denoising (and arguablly needed it given MPEG-2).

But at 1080p we're near the perceivable "real" detail for most flim sources. it's not like you can pick out small objects better in a 4K than a 1080p master for all but a very few digital productions, but the grain is actually resolvable specs instead of a scaled-down texture or noise like with 480p.

[shon3i]

Can a 15mbps H.264 encode match a 40mbps MPEG 2 encode?
No, H264 is rougly 30% better than MPEG2 in very low bitrates, so on high bitrates, H264 is still better but by invisible margin, i'd say at <25mbps MPEG2 vs H264 is totaly transparent each other, but can depend on implementation, at 30mbps i think there is no way to see any difference with any of H264, VC-1, MPEG2 implementation.

[honai]

The middle? I hope we're near the end; this started a decade ago with DVD. We can pick up on it more with HD, but DVDs got tons of denoising (and arguablly needed it given MPEG-2).

But at 1080p we're near the perceivable "real" detail for most flim sources. it's not like you can pick out small objects better in a 4K than a 1080p master for all but a very few digital productions, but the grain is actually resolvable specs instead of a scaled-down texture or noise like with 480p.

Not sure I understood you correctly, but I hope we can agree that almost all Universal titles on HD DVD/Blu-ray have excessive DNR applied that indeed removes detail, not film grain. And given the fact that most reviewers point this out but still Universal titles sell very well, I don't think anything is going to change about this practice in the foreseeable future.

Or do you have inside knowledge to the contrary?

[benwaggoner]

Why in the world do they use these low bitrates, if they have an entire 50 GB disc available? So many extras on the disc?
I think real-world bitrates are quite a bit higher than that on average. There's always edge cases for whatever reasons, but we should be looking at average discs if we're talking about BD in general.

Blu-ray also lacks a high fidelity manadatory audio codec, so a lot of discs get AC-3 640 Kbps + PCM, which eats up a lot of capacity.

[benwaggoner]

Not sure I understood you correctly, but I hope we can agree that almost all Universal titles on HD DVD/Blu-ray have excessive DNR applied that indeed removes detail, not film grain. And given the fact that most reviewers point this out but still Universal titles sell very well, I don't think anything is going to change about this practice in the foreseeable future.

Or do you have inside knowledge to the contrary?
As a matter of policy I can't really talk about the particulars of particular studios or titles I've worked on.

I do agree that overuse of DNR is a problem in the industry, and I fear we're encouraging viewers to adopt an inappropraite aesthetic expectation for what HD should look like.

[benwaggoner]

No, H264 is rougly 30% better than MPEG2 in very low bitrates, so on high bitrates, H264 is still better but by invisible margin, i'd say at <25mbps MPEG2 vs H264 is totaly transparent each other, but can depend on implementation, at 30mbps i think there is no way to see any difference with any of H264, VC-1, MPEG2 implementation.
30%? That's far, far too low. MPEG-4 pt 2 ASP was perhaps only 30% better than MPEG-2, but even VC-1 can outperform MPEG-2 by 2x or more for movie content at moderate bitrates.

[Dark Shikari]

No, H264 is rougly 30% better than MPEG2 in very low bitratesAre you nuts? Even in PSNR, I've seen x264 beat Xvid by over 100% at medium bitrates. And surely you wouldn't say MPEG-2 is better than Xvid.

Heck, MPEG-4 ASP's entropy coding advantage over MPEG-2 alone might even more than 30%.

[CruNcher]

hehe look @ simps his current 640x encode now do the same with hcenc and look at the quality difference i mean not being able to decipher any object anymore because of all the blocks is surely more then 30% :P

[Chengbin]

DS, with RDRC, do you think 30-35mbps is enough for transparency for most 1080p encodes?

Would you use RDRC if you're a compressionist for a studio?

Is 40mbps a hard limit for Blu-ray players? Can you go over 40mbps for peaks, or you must VBV it at 40mbps?

[Dark Shikari]

Would you use RDRC if you're a compressionist for a studio?No, it's not stable enough visually, especially in fades. Plus, the entire concept of RDRC is antithesis to Blu-ray encoding; in Blu-ray encoding you want good quality everywhere, while RDRC is built around sacrificing quality in the places that have the least temporal impact.

[Chengbin]

I thought it was a very bad idea to turn off in-loop deblocking. Why is some blu-rays have deblocking disabled? Is it another example of the studio's stupidity in video encoding?

[Dark Shikari]

I thought it was a very bad idea to turn off in-loop deblocking. Why is some blu-rays have deblocking disabled? Is it another example of the studio's stupidity in video encoding?At such low quantizers, it doesn't hurt too much. And it's a great way of creating fake sharpness if you don't have a good enough encoder to do it properly.

[benwaggoner]

I thought it was a very bad idea to turn off in-loop deblocking. Why is some blu-rays have deblocking disabled? Is it another example of the studio's stupidity in video encoding?
The feeling is that the loop filter increases softness too much. It'd be turned on scene-by-scene as needed vai segment reencoding.

This seems unlikely to me, but this is what I'm told. This was presented as a feature by Sony at NAB when describing thier updated encoder.

Now, it is definitely true that there have been H.264 encoders where this was an issue several years ago (like the first DVD Forum tests where H.264 came in behind MPEG-2 and VC-1), but since the strength is both adjustible and QP-adaptive, I don't see how turning it off entirely makes sense due to a limitation of the bitstream itself.

I'd say that in-loop debocking is the single best feature of H.264, followed by CABAC. The total of the rest of the efficency gains are probably less than those two put together.


...perhaps it also makes decoding in software eaiser? Although dropping CABAC would be a bigger help there with less efficiency impact at these bitrates.

[Chengbin]

...perhaps it also makes decoding in software eaiser? Although dropping CABAC would be a bigger help there with less efficiency impact at these bitrates.

Wouldn't every Blu-ray player must be able to decode 40mbps H.264? Why would the difficulty of decoding matter?

[benwaggoner]

Wouldn't every Blu-ray player must be able to decode 40mbps H.264? Why would the difficulty of decoding matter?
Well, yes they'd need to be able to decode it, but the spec doesn't say how many fps they need to actually get to the screen :).

But no, that doesn't make much sense to me anyway. And if they're struggling with 1080p24, 1080i30 would be completely out of reach.