This thread is in no way intended to "bash VC-1" as a codec per se but to make the community aware of one of the single
most annoying and severe picture artifacts we still have to bear with our next generation formats - POSTERIZATION.

This is not a format war topic either as both sides suffer from it equally.
And this artifact is in now way limited to VC-1 as all three available codecs are prone to it.

Why did I choose to state explicitly VC-1 in the topic title then?
Well for once VC-1 is quite well represented here @ avsforum - I sure would love to hear some ("marketing-free") insights from Amir and Ben about this.

Other than that posterization just happens to be a really big issue especially with some of the latest VC-1 releases.
So even with the "very latest", applied generation of VC-1/PEP and considering that we are talking about (most probably highly optimized) high-profile
releases there is strong indication that there is in fact an inherent problem with the VC-1/PEP workflow regarding difficult to avoid posterization.

To put it very simple these are potential causes for the posterization we see all too often (in no particular order):

- it is already present in the source/master

- "flawed" 4:4:4 10/12bit -> 4:2:0 8bit conversion

- unsufficient hand-tuning by the compressionist for problematic sequences. eg. proper dithering can help to hide any obvious banding/posterization artifacts

- unsufficient bitrate/bandwidth for particular sequences. Problem is that the
encoder likely is automatically applying very low bitrates due to the "simplistic nature" of some of these scenes - this "should" be manually corrected in the second pass.
Best case scenario is that the encoder is able to indentify such sequences automatically.

I invite Amir and Ben to complement and/or correct this simplistic list.

To demonstrate the problem I will post the following screenshots of recent VC-1 releases.
My thanks goes to members Xylon and House for all their efforts in providing these.
These screenshots are for demonstration purposes only - results may vary slightly depending on the playback configuration
BUT I personally verified these caps with my XA2 and the PS3 - both times the results were just as bad as shown here :

Happy Feet , VC-1, HD-DVD/Blu-Ray, unaltered, provided by member Xylon

http://www.imageviper.com/displayimage/81879/0/Happyfeethddvd3.png
Direct digital transfer so no master to blaim here ;)


Planet Earth, VC-1, HD-DVD/Blu-Ray, provided by member House

http://img262.imageshack.us/img262/6651/pebandax0.th.png (http://img262.imageshack.us/my.php?image=pebandax0.png)


VERY interesting to say the least is the comparison with the BBC h.264/AVC broadcast:

http://img176.imageshack.us/img176/7052/bbcbandhm2.th.png (http://img176.imageshack.us/my.php?image=bbcbandhm2.png)

So again this leads to the conclusion that the source/master is not the issue here (how could it ever be with such horrible posterization as shown here)
but the VC-1 encoding. Everybody is free to draw any conclusions from the comparison with the BBC capture.

These are just single samples of the problem at hand. eg. the recent Planet Earth release shows severe posterization (among other artifacts) not just occasionally
but a several times per 45min episode - some of these shots last for many seconds (like the one shown above) - starting with the initial BBC logo...

As another example I haven't really seen any high-def release with underwater sequences without a hint of posterization
(-> eg. VC-1 releases U571, Superman Returns, Happy Feet,... AVC/h.264 -> Ice Age 2...).

The important question is: What can be done to avoid it in the future?

I sincerely hope that Amir, Ben and other insiders admit this issue and provide insights of their strategies to avoid it.

- unsufficient bitrate/bandwidth for particular sequences. Problem is that the
encoder likely is automatically applying very low bitrates due to the "simplistic nature" of some of these scenes - this "should" be manually corrected in the second pass.
Best case scenario is that the encoder is able to indentify such sequences automatically.


Is exactly what's going on - any of the three codecs (MPEG2, AVC, VC1) will disply the similar levels of posturization when the above scenarion happens.

Basically it comes down to quality control and following best practices at the point of encoding.

Yes, but this can be manually corrected. I have done some video encoding, nothing major, but enough to know some things. If I use a variable bitrate, similar things happen, but if I use a 2 pass encoder, forcing the encoder to use a higher average bitrate, I end up with a only slightly larger file and a much better encode. By slightly, I mean I was getting about 300MB on the single pas, and around 330 with the two pass. Now, I realise that it gets a lot bigger with HD video, but it could get more advanced then I do by forcing the encoder to use higher bit-rates at certain points of the video. I personally have not done this, as I don't need to or know how, but I am pretty sure it can be done.

The things that happen here is thnat the encode see the scene as a BLACK scene 1 color, it doesn't see the whole color. In case of HF, i'am pretty sure the encoder saw a all black scene and putted 5mbits (witch is the bitrate for that scene) just like the Credit at the end and the computer didn't see the subtle gradiant of color.

Same with the sun, the color gradiant is so subtle and that computer must have seen only 1 color and dropped to 4-5 mbits (i don't know the bitrate scene i don't have planet earth) and voila Banding!

Those have to be hand corrected , i did a multiple of encode using x264 and i always needed to tweak those scene, especially if there's no grain. When there's grain, it not a problem because the encoder see it and boost the Bitrate and that compensate for it (altought in might need adjustements), you rarely see Posterization is a FILM filmed with Filmed, mostly only happen on Digital filmed stuff (with no grain) and CGI animation.

NIN show excel in that matter, it got TONS of gradiant and to my knowledge it never band..

Those have to be hand corrected , i did a multiple of encode using x264 and i always needed to tweak those scene, especially if there's no grain. When there's grain, it not a problem because the encoder see it and boost the Bitrate and that compensate for it (altought in might need adjustements), you rarely see Posterization is a FILM filmed with Filmed, mostly only happen on Digital filmed stuff (with no grain) and CGI animation.

NIN show excel in that matter, it got TONS of gradiant and to my knowledge it never band..

Now that's fascinating to me. It's another reason why people shouldn't be so eager to eradicate a films original grain from the process.

I found a lot of this on the SD of THE HUNGER, and remember reading somewhere that the materials on this film (though only 20 years old) basically had to be colorized due to fading. I wonder if the heavy digital manipulation left it more susceptible to this banding?

I know the BD of AMERICAN PSYCHO has been criticized for banding, and I think that's MPEG-2?

This is very bothersome for me too, I notice it all the time on my Toshiba CRT rear projection with HDDVDs.

I'd go as far as saying that this sort of posterization is the worst flaw of HD DVD and Blu-ray, as they become obvious when the rest of the image properties are so perfect.

If you have a Lumagen scaler with the 11 point gamma correction you can increase or decrease both banding and posterization - to increase the visibility of either, one would increase Luma at low IRE levels and do the opposite at to decrease both artifacts. I completely agree that this is a problem, especially for some displays, and could be rectified by employing TLC to these scenes. It's not a problem that is exclusive to VC-1 as has been pointed out. My opinion is that the problem lies with those applying VC-1 not being aware - in other words someone is asleep at the wheel or possibly the display being used isn't sceptible to these problems due the type of monitor and calibration.

Here are some examples I can think of - underwater scene in second Borne movie where the jeep goes over the bridge; Superman Returns has banding in the space pan down to upper atmosphere and posterization in the underwater scenes and of course Happy Feet underwater scenes (banding if I remember correctly). I suspect that with some displays this will hardly be noticable or may not be noticable at all and that's why some here will just shake their heads and wonder what we're talking about. If a friend says he doesn't see the problem you can simply watch the scenes pointed out here on his display and chances are you won't see the problem either. Let him watch on your display and see his reaction when those scenes are viewed.

Cheers,

Grant

you rarely see Posterization is a FILM filmed with Filmed, mostly only happen on Digital filmed stuff (with no grain) and CGI animation.

Hmm, how about "Open Season"? It's pure CG and looks PERFECT!

I hope it's not a "flaw" of VC1. I can buy "sleeping" at the wheel. But Microsoft keep telling us they have THE BEST tweak for VC1. If this is the BEST they can get, I wonder what if someone is in a coma at the wheel! :confused:

Btw, according to Microsoft, more bitrate won't help because that's already as good as it gets. Their aim is to REDUCE bitrate with more tweak as time goes on!

regards,

Li On

Hmm, how about "Open Season"? It's pure CG and looks PERFECT!

I hope it's not a "flaw" of VC1. I can buy "sleeping" at the wheel. But Microsoft keep telling us they have THE BEST tweak for VC1. If this is the BEST they can get, I wonder what if someone is in a coma at the wheel! :confused:
Li On
According to this post (http://www.avsforum.com/avs-vb/showthread.php?p=10405151&&#post10405151) on Blu-ray thread, there is posterization/banding on Open Season.
For those that say there's no banding in Open Season, I have proof now. I was at Best Buy, and literally watched the BD version on the 50" 1080p Panasonic plasma for just 2 minutes before seeing it glaring at me, mocking my very existence.

So, again, this is banding in OPEN SEASON:

18:01, and 18:43, in the sky. Banding is obvious, but...

The worst offender: 18 minutes, 43 seconds. Awful banding in the lower right of the screen, and pretty bad everywhere else on the screen.

This really just goes to show that some people just do NOT see banding, even when it's staring right at them.


Post on Banding on Open Season (http://www.avsforum.com/avs-vb/showthread.php?p=10405151&&#post10405151)

Interesting. Many people on AVS have been saying that the Blu-ray and HD-DVD versions of this are "much better" than the original BBC and Discovery HD broadcasts.

What bitrate was used in the BBC broadcast?

Interesting. Many people on AVS have been saying that the Blu-ray and HD-DVD versions of this are "much better" than the original BBC and Discovery HD broadcasts.

What bitrate was used in the BBC broadcast?

CBR 19.44Mbit/s h264.

According to this post (http://www.avsforum.com/avs-vb/showthread.php?p=10405151&&#post10405151) on Blu-ray thread, there is posterization/banding on Open Season.

Just checked. No banding whatsoever on "Open Season". The poster had no idea of the topic IMO.

regards,

Li On

I'm watching "Sky Captain and the World of Tomorrow". So far there were 2 very bad banding scenes:
- chapter 7, 36min48-51sec, the plane just dove into water, MPEG2 at around 20mbps
- chapter 7, 37min12sec, underwater scene with the plane, MPEG2 at 17.4mbps

Besides that the blu-ray looks great!

regards,

Li On

CBR 19.44Mbit/s h264.

Thanks.

Based on this, it would seem that if the BBC version didn't have posterization, there is no reason the HD-DVD or Blu-ray versions should either. Obviously.

Thanks.

Based on this, it would seem that if the BBC version didn't have posterization, there is no reason the HD-DVD or Blu-ray versions should either. Obviously.
But there is posterization in the BBC version, just not to the same degree. This is a very simple picture (content wise) and the fact that 19Mbps AVC encode still has posterization tells me that there is something more than just compression causing this problem.

Sounds like the encoding box that did the BBC version deserves a promotion.

Thanks.

Based on this, it would seem that if the BBC version didn't have posterization, there is no reason the HD-DVD or Blu-ray versions should either. Obviously.

This posting (http://www.avsforum.com/avs-vb/showthread.php?p=10392399&&#post10392399) from this thread (http://www.avsforum.com/avs-vb/showthread.php?t=839376) suggest the bitrate for the shots The Lion posted are considerably lower. You would have to question the accuracy of the bit rate meter on the PS3 but I would doubt the error would be this great.

I see banding in both images The Lion provided for what its worth.

ted

Just goes to show what happens when proper dithering is not done. Sad really, but its not the codec's fault, thats just what was fed to it.

In my opinion the BBC version has a bit too much banding also.

As suggested, that kind of banding is normally due to source processing issues. We have a great technology as part of PEP/CineVision PSE that prevents banding from being introduced as part of the conversion to 4:2:0 8-bit, but if there's banding in the source, than there's going to be banding in the output.

Also, comparing a broadcast to a disc title isn't always that informative, since they could have been different telecines from all we know.

Encoding at too low a bitrate can also cause banding, but as was suggested above, that's exactly what segment reencoding is for. Without access to the source, I can't say what the issue is in particular, but there's nothing intrinsic to VC-1 that could cause that.

Also, comparing a broadcast to a disc title isn't always that informative, since they could have been different telecines from all we know.Would a telecine even be involved, since the broadcast/disc comparison was of Planet Earth, which was shot in HD?

As suggested, that kind of banding is normally due to source processing issues. We have a great technology as part of PEP/CineVision PSE that prevents banding from being introduced as part of the conversion to 4:2:0 8-bit, but if there's banding in the source, than there's going to be banding in the output.

Also, comparing a broadcast to a disc title isn't always that informative, since they could have been different telecines from all we know.

Encoding at too low a bitrate can also cause banding, but as was suggested above, that's exactly what segment reencoding is for. Without access to the source, I can't say what the issue is in particular, but there's nothing intrinsic to VC-1 that could cause that.

Ben,

thank you for your comments.

I have a pretty hard time to believe that the posterization galore that is the Planet Earth VC-1 release is anywhere near to being representative let alone transparent to any given master/source Warner/BBC could have come up with. Do you really suggest that posterization this severe and common throughout the runtime of Planet Earth is due to a "flawed source"? What about all the other compression artifacts - eg. as Fettastic so eloquently put it the flock of birds (beginning of the first episode) looks like it is being attacked by a swarm of killer bees... Lots of jelly-fish to be seen as well btw :p

Talking about Happy Feet - there is certainly no problem with the source as this is a direct digital transfer. You are stating that you provide "great technology as part of PEP/CineVision PSE that prevents banding from being introduced as part of the conversion to 4:2:0 8-bit" - great, so this is no conversion problem as well.
That leaves us with an encoding problem -> theory: encoding at too low a bitrate. This certainly is no issue with VC-1 per se as I clearly stated in my initial post:
" there is strong indication that there is in fact an inherent problem with the VC-1/PEP workflow regarding difficult to avoid posterization."

Therein lies the problem. Given the (rather questionable) trend of ever decreasing applied average bandwidth/bitrates for common VC-1 encodings (an approach and "accomplishment" that your team in general and Amir in particular seem to be quite fond and proud of) don't you agree that MS's PEP application "should" be able to indentify sequences that are prone to severe posterization and therefor AUTOMATICALLY apply sufficient bitrates during the first pass. This would make the quality of the encoding regarding posterization somewhat independent from the attentiveness of the compressionist.

Sure, you can always argue that there is no "problem" with the VC-1/PEP workflow that causes posterization - it is just the responsibility of the compressionst to hand-tune these sequences. BUT in the end we as customers have to agonize over many releases with severe posterization artifacts.

I think high-profile releases like Planet Earth, Happy Feet and Superman Returns clearly demonstrate that relying on the proper actions of any given compressionist in order to avoid posterization is certainly not an effective approach - at least not with our 12-15MBit/s avg <5Mbit/s minimum encodings.

Hi folks,

Allow me to show my ignorance but could someone tell me the difference between "posterization" and "banding"? They seem, to me, to be similar things.

Just trying to educate myself...
-t

I'll open up my P.E. over the weekend and take a look, but my best guess is that for those segments, quantization is set too high which results in lower bitrates but shouldn't really be used as the P.E. picture was so simple, it would have encoded well even with with lower quantization (lower is better).

It is possible that this and the MI3 example was merely operator (compressionist) error in dialing in too low a value for these frames. I think the only way we can prevent things like this is to try to keep a cap on how high quantization can go, even if the bitrate cap is 28Mbps, that should be high enough for most materials to work well --- and the P.E. frame was so simple there is no reason for that kind of posterization. (Note that the AVC version also had slight banding, but it became much more pronounced on the VC-1 version) Someone at MSFT should just take a VC-1 analyzer and look at that P.E. segment frame by frame and see if the theory of quantization set too high is correct.

Just so we have the information on record, has anyone profiled the data specs on the BR/HD discs for this title? It's 4 discs for either format, correct? Are these sl or dl, though? What is the file size per disc? Is it about 2.5-3 hr of footage per disc with a single DD track?

90% of the banding issues I have seen are display related. What's apparent on my Plasma as banding, is totally missing on the 9 inch CRT projector.

I have yet to see a Plasma or LCD that doesnt have banding when the contrast ratio is limited by the scene in question.

Until a poster can assure us somehow it isnt his display, this is a non issue.

This seems to have become worse since HDMI and DVI became the standard transmission method. When component or RGBHV are used. Evidence of banding is largely eliminated. To me this points to contrast being limited by available 8 bit DVI or 10 bit HDMI transmission methods.

Just so we have the information on record, has anyone profiled the data specs on the BR/HD discs for this title? It's 4 discs for either format, correct? Are these sl or dl, though? What is the file size per disc? Is it about 2.5-3 hr of footage per disc with a single DD track?

HD-DVD;

Disc 1: Size - 23,792,599,040, Length - 2:32:17
Disc 2: Size - 23,658,328,064, Length - 2:32:44
Disc 3: Size - 24,005,175,296, Length - 2:33:11
Disc 4: Size - 15,973,357,56, Length - 1:42:13

Just the single audio track, DD5.1 448Kbit/s.

This thread is in no way intended to "bash VC-1" as a codec Ummm.... sure it isn't.

90% of the banding issues I have seen are display related. What's apparent on my Plasma as banding, is totally missing on the 9 inch CRT projector.

I have yet to see a Plasma or LCD that doesnt have banding when the contrast ratio is limited by the scene in question.

Until a poster can assure us somehow it isnt his display, this is a non issue.

Agreed. If someone talks about noticeable banding, and blames the source, i'd definetly expect direct screenshots from the source before i accepted that argument over the display being at fault.

HD-DVD;

Disc 1: Size - 23,792,599,040, Length - 2:32:17
Disc 2: Size - 23,658,328,064, Length - 2:32:44
Disc 3: Size - 24,005,175,296, Length - 2:33:11
Disc 4: Size - 15,973,357,56, Length - 1:42:13

Just the single audio track, DD5.1 448Kbit/s.
Too bad, with so much space left on HD DVD 30, they wouldn't include some extra or better audio :mad: I guess they are limited to 25 GB SL BD.

They limited themselves, since there is no requirement to use one encode for both formats, nor is there a requirement to only use sl bluray discs.

This thread is in no way intended to "bash VC-1" as a codec per se but to make the community aware of one of the single
most annoying and severe picture artifacts we still have to bear with our next generation formats - POSTERIZATION.
*Chortle* When's the last time you attacked anything BD?

I thought VC-1 was a video codec used by both formats.

I thought VC-1 was a video codec used by both formats.
Only when it is used by both formats.

Isn't that what we're talking about? Happy Feet, Planet Earth. Both vc-1. Both lousy when it comes to gradations.

Only when it is used by both formats.

Well, with "Happy Feet" and "Planet Earth" it *is*.

Only the HD DVD folks get their panties in a wad when someone dares criticize VC-1 because it has a monopoly on those encodings.

Just goes to show what happens when proper dithering is not done. Sad really, but its not the codec's fault, thats just what was fed to it.

In my opinion the BBC version has a bit too much banding also.


This is a shame that the "Next Generation" formats display these kind of flaws, because their are people who delight in pooping on any kind of change (from SD), and this sorta thing gives them fuel. If it's prevalent enough, the sad thing is that they would be justified in their objections.

I can see that certain source materials are only as good as they are, but it's pretty lame if technicians know this is a problem and don't keep an eye out. It makes you think that the most basic measure of quality control; watching your own product, just isn't part of the process.

It seems abundantly clear from those here who are in the know that it's not code specific, but human error, so I gotta agree the thread title is a little salacious.

FWIW, BD folks seldom criticize mpeg -2 or see any sort of problems at all. Amazing, but true, and criticizing mpeg-2 gets their panies in a wad :)

All the 50 GB disks are being cranked out for PoTC since even night at the Museum I believe is 25 GB SL, and this is Fox - Perfect Hi Def or whatever their slogan is.....not a peep about from the BD crowd, is there? Mediocre transfer, but do we hear anything? noooooooo....it's Happy Feet and Planet Earth, both titles blow Night at the Museum to smithereens when it comes to PQ. But, hey....why bring that up ;)

Doesn't Night at the Museum have it's own thread full of praise and complaints? Take it over there. In this thread we're discussing banding in certain "sacred" vc-1 titles. ;)

FWIW, BD folks seldom criticize mpeg -2 or see any sort of problems at all. Amazing, but true, and criticizing mpeg-2 gets their panies in a wad :)

For God's sake, grow up.

:rolleyes:

Rob, firstly, I was responding to PeterTHX. Apparently that was grown up enough for you (so it was tongue in cheek retort). Second, don't advice me on personal matter. Stick to the forum topic. OK? Not too hard, is it? Don't pull your mother hen BS on me!

For God's sake, grow up.

:rolleyes:


I wouldn't count on it based on his posting history.

I am not sure you even know what that means ;)

For God's sake, grow up.

:rolleyes:

Childish or not that shoe fits and walks all over these boards. Plazman is pretty even handed IMO, and upfront about his bias. I've heard him praise many BDs.

The frame-grab fever makes me think people of both camps are only succeeding in proving that there needs to be more qaulity control in general if your going to tell people HDM is an upgrade worth their attention.

Isn't that what we're talking about? Happy Feet, Planet Earth. Both vc-1. Both lousy when it comes to gradations.
I'm just asking when he's crapped on anything "BD". He's very "clever"

What does this thread have to do with BRD or HD-DVD? I thought we're talking about VC-1 and it's "quirks".

You are "clever" too. How many BD only titles are VC-1? He's just taking a shot at HD, as usual.

Rob, firstly, I was responding to PeterTHX. Apparently that was grown up enough for you (so it was tongue in cheek retort). Second, don't advice me on personal matter. Stick to the forum topic. OK? Not too hard, is it? Don't pull your mother hen BS on me!

Ok, when I make a mistake I will admit it. I made a mistake here, since I did not see what you were responding to. Since it wasn't a post directly above yours, you probably could/should quote it. As such, I took your comments out of context.

In the "other topic" about banding occuring in PE, specific scenes have been cited for further investigation. They include the sun scene at 5 Mb/s, as well as other scenes at middle and even high bitrates (20 Mb/s). I wonder if the banding in these other scenes are as pronounced as the sun scene or less pronounced due to the greater bitrates or is the degree of banding entirely unrelated to the bitrate (it just happens when it happens?). Similarly, are we concentrating on the sun scene simply because it is the only one being posted, or is it simply where the banding is most readily visible (compared to the other scenes)?

The popular angle here is that the low bitrate is a primary suspect in the incidence of banding in the sun scene. However, this might not be the case if similar degree of banding is also occuring at mid and high bitrates. It is also possible that the bitrate was coincidentally low in that sun scene, but the banding is coming from an entirely different factor. It behooves us to investigate this further, if bitrate is possibly not the key factor behind it. Bitrate is an "easy target" if it happens to be low right when an artifact happens to occur, but it may not be telling the whole story, either.

The popular angle here is that the low bitrate is a primary suspect in the incidence of banding in the sun scene. However, this might not be the case if similar degree of banding is also occuring at mid and high bitrates. It is also possible that the bitrate was coincidentally low in that sun scene, but the banding is coming from an entirely different factor. It behooves us to investigate this further, if bitrate is possibly not the key factor behind it. Bitrate is an "easy target" if it happens to be low right when an artifact happens to occur, but it may not be telling the whole story, either.

What I find the head-scratcher is that this is digital to digital, on SD I used to blame this on too much hocus-pocus. For example sharpening, edge enhancement, or playing with the tonal range to compensate for blown-out lite tones or boosting murky grays? Even though the original footage is HD video, they coulda messed with it.

I do see the banding in the broadcast sample as well as the disk playback, any boost in contrast would accentuate it, but who knows were it occurs?

Don't pull your mother hen BS on me!Good ol Rob... the "Unofficial" AVS forum moderater.

From what I've gathered it seems like banding occurs at a below 10mbps bitrate and when there's a frame with subtle gradations, usually of the blue variety. Hmm.

Has there been any confirmed banding on AVC and mpeg-2 material? Maybe vc-1 just isn't cut out for these insanely low bitrates. Why continue to push it so low when there are obvious artifacts like this? Wouldn't the compressionist see this on his/her monitor?

I'm sure it wouldn't be hard to find on mpeg-2 material, and even avc material. I don't think it is really a surprise that mpeg-2 would do this, when bit-starved. The fix is relatively obvious- feed it more bitrate until it doesn't do it anymore. ;) There's nothing in this behavior to be seen as a codec flaw. It obeys a classic philosophy that the amount of data needs to scale with the information in the scene.

If vc-1 does it, that has a slightly different ramification, since it has been engrained into our heads that it can accomplish all these incredible feats with less bitrate. The bar has been set high for it, so if it misses on something as basic as smooth gradients, controversy ensues. It suggests that maybe vc-1 isn't the omnipotent codec it is touted as, maybe it does have limitations, or maybe it does have realistic requirements that are not so far removed from the codecs that preceded it.

It just may be the reality that vc-1 will band for the exact same reasons mpeg-2 will band. The latter can simply be fed more bitrate, but the former finds itself in a strange position between conceding that more bitrate would benefit it, as well, or continue to tow the line "if you have all the bits you need, adding more won't matter"...and subsequentially, banding is simply an inevitable artifact of the codec (since more bits won't have any effect).

I think you got it right, Hanky. In cases where a gradation scene presents itself Mr. VC-1 needs to be fed. If Mr. VC-1 isn't fed enough we get the kind of stuff you'd expect from a bad quicktime movie. This is just unacceptable.

I think we should stop treating Mr. VC-1 like an anorexic. You can't win friends with low conservative bitrates that are just efficient enough.

Would a telecine even be involved, since the broadcast/disc comparison was of Planet Earth, which was shot in HD?
Right, of course.

That said, there's lots of paths between the final edit in the Avid and the source used for encoding the titles.

I think you got it right, Hanky. In cases where a gradation scene presents itself Mr. VC-1 needs to be fed. If Mr. VC-1 isn't fed enough we get the kind of stuff you'd expect from a bad quicktime movie. This is just unacceptable.

I think we should stop treating Mr. VC-1 like an anorexic. You can't win friends with low conservative bitrates that are just efficient enough.
It's not an issue of more bits. There are encoding modes that can help reduce bandwidth with gradients while improving quality as well, like adaptive deadzone.

Once again, I point out that the v11 version of the VC-1 codec is freely available for anyone who'd like to experiment with combinations of sources, settings, and data rates.

Childish or not that shoe fits and walks all over these boards. Plazman is pretty even handed IMO, and upfront about his bias. I've heard him praise many BDs.

The frame-grab fever makes me think people of both camps are only succeeding in proving that there needs to be more qaulity control in general if your going to tell people HDM is an upgrade worth their attention.

Just wondering -- how much do you think HD video disc buyers are willing to pay to make sure that every frame (24 per second) is HD "perfect"?

It's not an issue of more bits. There are encoding modes that can help reduce bandwidth with gradients while improving quality as well, like adaptive deadzone.

Once again, I point out that the v11 version of the VC-1 codec is freely available for anyone who'd like to experiment with combinations of sources, settings, and data rates.

You'd think the compressionists that worked on Happy Feet and Planet Earth would know this. :D

From what I've gathered it seems like banding occurs at a below 10mbps bitrate and when there's a frame with subtle gradations, usually of the blue variety. Hmm.

Has there been any confirmed banding on AVC and mpeg-2 material? Maybe vc-1 just isn't cut out for these insanely low bitrates. Why continue to push it so low when there are obvious artifacts like this? Wouldn't the compressionist see this on his/her monitor?

It was pointed out in the very first post of this thread that the problem is common to each of the three Codecs. Banding being equal to Posterization. VC-1 just happens to be the "convenient" focus of discussion -- interesting how that works :).

In the examples provided by the OP, there is actually very little information present, but that which IS present is subtle. Throwing bits at it isn't the solution.

Good ol Rob... the "Unofficial" AVS forum moderater.


Spell check is your friend. It's m-o-d-e-r-a-t-o-r.

Can you say "moderator"? I know you can.

It was pointed out in the very first post of this thread that the problem is common to each of the three Codecs. Banding being equal to Posterization. VC-1 just happens to be the "convenient" focus of discussion -- interesting how that works :).

In the examples provided by the OP, there is actually very little information present, but that which IS present is subtle. Throwing bits at it isn't the solution.

Yep. But how do we know that throwing a higher bitrate at it wouldn't help? Do we know what the respective bitrates were for that scene on the BBC broadcast vs the VC-1 HD disc versions?

I have a pretty hard time to believe that the posterization galore that is the Planet Earth VC-1 release is anywhere near to being representative let alone transparent to any given master/source Warner/BBC could have come up with. Do you really suggest that posterization this severe and common throughout the runtime of Planet Earth is due to a "flawed source"? What about all the other compression artifacts - eg. as Fettastic so eloquently put it the flock of birds (beginning of the first episode) looks like it is being attacked by a swarm of killer bees... Lots of jelly-fish to be seen as well btw :p
I haven't seen the source or discs, nor was that a title I heard anything about during authoring, so I'm speaking only from general principals here. The great news is that there's way more VC-1 titles out there than our team can touch these days.

Talking about Happy Feet - there is certainly no problem with the source as this is a direct digital transfer. You are stating that you provide "great technology as part of PEP/CineVision PSE that prevents banding from being introduced as part of the conversion to 4:2:0 8-bit" - great, so this is no conversion problem as well.
That leaves us with an encoding problem -> theory: encoding at too low a bitrate. This certainly is no issue with VC-1 per se as I clearly stated in my initial post:
" there is strong indication that there is in fact an inherent problem with the VC-1/PEP workflow regarding difficult to avoid posterization."

Therein lies the problem. Given the (rather questionable) trend of ever decreasing applied average bandwidth/bitrates for common VC-1 encodings (an approach and "accomplishment" that your team in general and Amir in particular seem to be quite fond and proud of) don't you agree that MS's PEP application "should" be able to indentify sequences that are prone to severe posterization and therefor AUTOMATICALLY apply sufficient bitrates during the first pass. This would make the quality of the encoding regarding posterization somewhat independent from the attentiveness of the compressionist.
Are you stating with knowledge that Happy Feet was doing as a direct encode from the DI? Lots of CGI titles got mastered from D5 tape, FWIW.

Anyway, no, there's no intrinsic problem with VC-1 with banding. Remember it's architecturally close to a superset of MPEG-2, with lots of new tools that help reduce banding and posterization.

Sure, you can always argue that there is no "problem" with the VC-1/PEP workflow that causes posterization - it is just the responsibility of the compressionst to hand-tune these sequences. BUT in the end we as customers have to agonize over many releases with severe posterization artifacts.

I think high-profile releases like Planet Earth, Happy Feet and Superman Returns clearly demonstrate that relying on the proper actions of any given compressionist in order to avoid posterization is certainly not an effective approach - at least not with our 12-15MBit/s avg <5Mbit/s minimum encodings.
The vast majority of all posterization seen in final titles was present in the source. That's not really something a compressionist can do much about - after a certain point, you're talking remastering or even rotoscoping.

CGI and digital productions are much more prone to banding in source than film is, since there isn't nearly as much noise to mask banding. And rendering techniques can also wind up going through an 8-bit path more often than you might expect, further constrained by later filtering.

If you want a "fire-and-forget" way to reduce it, we could just throw on a low-pass filter, similar to the effect you get from the H.264 loop filter :). But we're all about preserving the source as accurately as possible, which can include warts and all if there be warts.

It was pointed out in the very first post of this thread that the problem is common to each of the three Codecs. Banding being equal to Posterization. VC-1 just happens to be the "convenient" focus of discussion -- interesting how that works :).

If it's common with the other codecs then what titles does it occur in? Ice Age 2 comes up but then gets lost in the shuffle.

In the examples provided by the OP, there is actually very little information present, but that which IS present is subtle. Throwing bits at it isn't the solution.

That shot from Planet Earth is anything but subtle IMO. It's halfway between being an artifact and being an artistic effect.

Interesting analogy, there- except, if it is a charactericature of anorexia you want to use, shouldn't it be Ms. VC-1? :p (...or maybe that is being sexist?) Similarly, Ms. VC-1 may be the young, hot codec on the block, but she still puts her pants on one leg at a time like any other codec. :D

I am intrigued, however, by the "blue colorspace contribution" theory, as well. I think it was explained earlier that due to the particular data contribution for the blue channel (chosen presumably because of human's typically poor vision in the blue range), blue-dominated scenes are literally left with a mere 16 possible levels to cover a full gradiation?! It's a wonder any kind of blue scene has a chance of being rendered properly, if that is the case. That would be like 4-bit color (albeit, just for the blue channel). Dithering the blue channel definitely seems like a mandatory measure (and naturally, that will have a bitrate cost, on its own, I suspect).

In the big picture view (no pun intended), the incidence of banding is likely attributable to the culmulative effect of all of these things, rather than one thing. Depending on the conditions, one thing may be more responsible than the other, however, the banding cannot be completely eliminated unless all factors are addressed together.

I hope to see more screen captures of some different scene scenarios suffering from banding, though.

Spell check is your friend. It's m-o-d-e-r-a-t-o-r.

Can you say "moderator"? I know you can.Good one... the old "You spelled a word wrong argument". I've seen you use it before when you are wrong about things.

Good one... the old "You spelled a word wrong argument". I've seen you use it before when you are wrong about things.

I may be wrong about things, but at least I know how to spell.

And it wasn't an "argument". It was a fact.

For the benefit of this topic, here is a quote from the other topic, where additional screen capture examples from PE at various bitrates, are linked. You have to change the number in the link (as instructed in the post) to see the 19 pix. I'm not saying I agree that banding is present in these pix. They are simply additional samples to be evaluated for potential banding scenarios.

I've taken pictures of all those points you've mentioned - hope I got them in the spots you're talking about.

http://theintarweb.info/pebanding/band1.png

Replace 1 with a number of your choice all the way up to 19 (numbered in the order you listed them, so band1.png is the cave, band19.png is the highspeed sunshot).

Just to comment on them, I think most of them are so minor they're not worth fussing over (and some of them I can't even see anything [in motion either]). The underwater or some of the sun shots for example are masked by noise; some of the other sun bits are just optics rather than the encoding. That sandstorm cloud there's some vertical striping (not sure I'd call it banding though), and the second sandstorm I'm a bit puzzled by the big green slotch. The cacti ones I'd say basically break down into some blocking rather than banding. The only one I can say I'd notice during playback and be annoyed is the last shot (maybe the sand dune one too, but that one you really need to see in motion). Nope of them are as horrific as what the OP listed.

Additional caption to the pix can be found here:
http://www.avsforum.com/avs-vb/showthread.php?t=839376&page=6&pp=30

More specifically:
Some time codes for Disk 2 for confirmation

02:42-43 Bitrate: ~10 Look at light from Cave during Fade in. (faint but there)
02:58-59 Bitrate: ~ 8 Look at Light halo around stalagmite. (faint but there)
31:23-29 Bitrate: ~ 17 - Light filtering above stingray.
51:20-25 Bitrate: ~ 10 (4.1 was lowest!) - Look at sand dune while panning.
53:55-59 Bitrate: ~ 14 Sun panning across sky before fade out.
54:12-15 Bitrate: ~ 20 Sun panning across sky again.
58:00-02 Bitrate: ~ 7 Look for it when sandstorm moves closer.
58:06-08 Bitrate: ~ 14 Look at leading edge of 'grey' in the sandstorm
1:03:28-34 Bitrate: ~ 20 Sun Rising
1:04:00-02 Bitrate: ~ 13 Sun rising, look under bottom branches of the tree.
1:04:42-45 Bitrate: ~ 8 Sun Rising, severe.
1:06:17-20 Bitrate: ~ 13 Sun Flare
1:09:32-35 Bitrate: ~ 5 Sun Flare
1:13:45-47 Bitrate: ~ 10 Fog and Sun
1:14:15-22 Bitrate: ~ 15 High Speed Camera pointing at sun with clouds. Faint outlines on edges of flare. Continues while looking up at the cacti
1:14:32-37 Bitrate: ~ 20 In Clouds after lightning strike. Severe on Right hand side of right cactus.
1:15:04-07 Bitrate: ~ 10 Sun flare edges.
1:15:53-56 Bitrate: ~ 20 Left hand Horizon. Evident with blue to dark blue transition during high speed shot.
1:17:20-25 Bitrate: ~ 5 Highspeed shot of sunrise, VERY evident.


I ran out of time at this point, but feel free to look for yourself.

I am intrigued, however, by the "blue colorspace contribution" theory, as well. I think it was explained earlier that due to the particular data contribution for the blue channel (chosen presumably because of human's typically poor vision in the blue range), blue-dominated scenes are literally left with a mere 16 possible levels to cover a full gradiation?! It's a wonder any kind of blue scene has a chance of being rendered properly, if that is the case. That would be like 4-bit color (albeit, just for the blue channel). Dithering the blue channel definitely seems like a mandatory measure (and naturally, that will have a bitrate cost, on its own, I suspect).
Eh? Of R, G, and B, B is least represented in Y' and most in the subsampled Cb. But it's got a full range of values, just at a lower resolution.

Also, this is true of all codecs.

The particular post I refer to escapes me, but it had something to do with that R+G+B formula with the different numerical weights for each color (or something like that).

Has there been any confirmed banding on AVC and mpeg-2 material?I believe that "The Wild" was AVC and had quite a bit of banding. It can be difficult to avoid with animations because of the cleanliness (adding grain can hide it). As some other people have noticed, avoiding banding or posterization in close to solid blues seems to be about the toughest. I believe it has something to do with how sensitive our eyes are to subtle changes in blue fields.

--Darin

Has there been any confirmed banding on AVC and mpeg-2 material?

Someone please read my post (http://www.avsforum.com/avs-vb/showthread.php?p=10409688&&#post10409688). Thanks!

regards,

Li On

Just wondering -- how much do you think HD video disc buyers are willing to pay to make sure that every frame (24 per second) is HD "perfect"?


That's a good question, but like most I'll avoid a direct answer with more questions. ;)

It depends on weather or not it's just someone falling asleep at the wheel. Is the current pricing system only feasible if compression is done in some automated way that doesn't involve someone watching what it's doing from beginning to end?

The thing is that just such issues may be caught all the time, but since our finished product has been corrected, it's hard to tell if it's a fluke or not.

What seems to be happening here is a ladder of

A VC-1 encoded disk shows banding > It could be because of too low a bit-rate > HD DVD uses VC-1 the most > HD DVD needs VC-1 because it only has 30GB> proof that the 50 GB of BD 50 with MPEG 2 is the way to go.

Is there anyone here with the credentials to back that up, or is it just one of a dozen possibilities?

IMO, all I can say about banding is if either format (Blu-ray or HD DVD) expects to truly get off the ground, then they better realize that it is quality that will pull this thing off.

I don't need the cost or aggrevation if they can't get the best techs out there to do their best work and clean up these discs.

I (for one) am not in this game to support some jerks corporate control or greed! What I mean is I (for one) expect something for my money...

IMO, all I can say about banding is if either format (Blu-ray or HD DVD) expects to truly get off the ground, then they better realize that it is quality that will pull this thing off.

I don't need the cost or aggrevation if they can't get the best techs out there to do their best work and clean up these discs.

I (for one) am not in this game to support some jerks corporate control or greed! What I mean is I (for one) expect something for my money...
I agree here with you DigVid. There have been some truly disgusting releases from both sides that these companies are making big margins on that look identical to their SD brethren.

Too bad, with so much space left on HD DVD 30, they wouldn't include some extra or better audio :mad: I guess they are limited to 25 GB SL BD.

So do you understand how we feel when WB continues to put out discs which are limited by the HD DVD spec?

Yep. But how do we know that throwing a higher bitrate at it wouldn't help? Do we know what the respective bitrates were for that scene on the BBC broadcast vs the VC-1 HD disc versions?

BBC b/c HD comes in at CBR 18Mbps - 20Mbps AVC depending on the type of program, I suspect that this is at the top of the tree.

FWIW, BD folks seldom criticize mpeg -2 or see any sort of problems at all. Amazing, but true, and criticizing mpeg-2 gets their panies in a wad :)

I think you will find that many here and on the blu-ray.com forum criticise MPEG-2, I for one don't care what codec something uses, as long as it gives us the best PQ available, from what I have seen AVC is the best right now, but it has problems with workflow as it is slow to use compared to VC-1. MPEG-2 also has its merits, if you have the need to pump out lots of titles and have an unlimited supply of BD-50s then it will perform very well, as seen in BHD or KoH.

It's not an issue of more bits. There are encoding modes that can help reduce bandwidth with gradients while improving quality as well, like adaptive deadzone.

Once again, I point out that the v11 version of the VC-1 codec is freely available for anyone who'd like to experiment with combinations of sources, settings, and data rates.

There are certainly other solutions, some maybe better. But I still believe the currently observed banding would go away at sufficiently high bit rates. The exception would be material that would cause visible banding at 8 bit color depth even with uncompressed video but I'm not sure that even exists.

- Tom

So do you understand how we feel when WB continues to put out discs which are limited by the HD DVD spec?

Actually, based on the file sizes, these were limited by the SL BR spec.

Actually, based on the file sizes, these were limited by the SL BR spec.

I was speaking in more general terms, i.e. when WB don't bother to optimise movies to the BD50 spec, or don't put PCM audio on a movie.

If it's common with the other codecs then what titles does it occur in? Ice Age 2 comes up but then gets lost in the shuffle.

The opening shot of a sunrise in Training Day has quite a bit of banding on the MPEG2 Blu-ray. There is still a bit of banding on the VC-1 HD DVD, but much less so.

Yep. But how do we know that throwing a higher bitrate at it wouldn't help? Do we know what the respective bitrates were for that scene on the BBC broadcast vs the VC-1 HD disc versions?
Rob,

Banding is present on the BBC broadcast, but to a lesser degree. BBC broadcast is reportedly 18-20Mbps CBR AVC. I think we can all agree that visually, this is a very simple shot. I seriously doubt that banding is due to low bandwidth given that there is very low complexity in the picture, yet banding is occuring even with high bitrate CBR.

So do you understand how we feel when WB continues to put out discs which are limited by the HD DVD spec?
It's a double edged sword which can cut both way. Since most BR releases are on BD25 now, it hurts HD DVD more, I guess. Anyway, I don't think it's that a big deal either way to me, since lack of extra/better audio is not a deal breaker. YMMV.

What's puzzling is disc 4, which uses ~15GB IIRC, there is room for another 10GB for both HDM, WB can certainly put some extra there.

Gee...I must have riled you all up there Lawrence of Arabia to get you so steamed you would search my post history and take one post out of context and repost it here.

Sorry, but despite your contentions to the contrary, nothing was taken out of context. I quoted your entire post on the subject. You made it very clear that it was, in fact, his DISPLAY that was causing the banding. You did not even leave open the possibility that it was the source. I also posted a link to the entire thread (which is discussing the same issues we are discussing here) so your argument that your comments were "taken out of context" is simply not true. Period.


Personally Rob, I think it is pretty sad. You pull out the two lamest tricks, and for what... not to add constructive input to this thread, but to attack me.

This is about as rich as it gets! Let's recap, shall we? Take a look at your first two posts in this thread:

Originally Posted by TheLion
This thread is in no way intended to "bash VC-1" as a codec

Ummm.... sure it isn't.

Hmm. If I didn't know any better, it would appear to me that you were ATTACKING the OP's personal motives/intent in creating this thread and discussing these issues despite the fact that he specifically indicated that this issue happens with all codecs!!!

And yes, your post "Ummm.... sure it isn't" really adds "constructive input to this thread" (quoting your words) doesn't it Guru? :rolleyes:

Then, your next post in this thread was this:

Good ol Rob... the "Unofficial" AVS forum moderater.

Hmm. Again, if I didn't know any better, it would appear that you were PERSONALLY ATTACKING me, despite the fact that I had never said a word to you in this thread. What's even more telling is the fact that your quote above was in response to something Plazman said, and was posted after I indicated to him that I was wrong and had taken his comment out of context!

So, why was it that you still felt the need to come in and personally attack me?

Oh, wait, I forgot. This is all my fault, I am the one who attacked you and your spelling ability.

You and I wouldn't even be having this exchange if you didn't feel compelled to come in and take a personal dig at me.

Your attempt to make yourself look like the victim in all this is beyond absurd.

Well, I can't spell, and I said something that could look bad when taken out of context from the original thread and without reading that specific threads progression. Fun times on AVS. :rolleyes:

Fun times indeed. :rolleyes:

Rob,

Banding is present on the BBC broadcast, but to a lesser degree. BBC broadcast is reportedly 18-20Mbps CBR AVC. I think we can all agree that visually, this is a very simple shot. I seriously doubt that banding is due to low bandwidth given that there is very low complexity in the picture, yet banding is occuring even with high bitrate CBR.

Thanks. Yes, I can still definitely see banding on the BBC broadcast, but to a much lesser degree. So, if the banding isn't due to low bitrates (and your comments do make sense in this regard), what would account for the difference in the amount of banding between the BBC broadcast and the HD-DVD/Blu-ray versions?

http://www.avsforum.com/avs-vb/showthread.php?p=10417318&&#post10417318
David,

Specifically looking at Happy Feet, I've worked with some clips from the movie on a disc for LG Electronics. Coming right from the D5 source, the pure blue sky had "banding". From my observation, it looks like they were trying to reflect the curvature of the earth in the sky. If that's right, it was an artist choice and therefore is properly represented. If not, it was introduced before any compressionist could have done anything about it. In either case, the encodes match the master; which is what we all want.

Sorry VC-1/HD DVD haters, you'll have to think up better FUD next time...

So do you understand how we feel when WB continues to put out discs which are limited by the HD DVD spec?

There are some nice BD 25 disks are there not, or do we just despise them in theory?

There are some nice BD 25 disks are there not, or do we just despise them in theory?

No, I despise the 30mbit bandwidth limit imposed by the HD DVD spec.

No, I despise the 30mbit bandwidth limit imposed by the HD DVD spec.Most likely that was an engineering choice that dictated the use of advanced video codecs (VC-1 or AVC) and audio codecs (LPCM is not a good fit, whereas TrueHD or dts-HD MA still fit). A choice of software techniques over hardware.

No argument that with BD50, then older MPEG2 for video and LPCM for audio becomes workable. It seems that Blu-ray specs were oriented towards older video codec and simple audio, but with complex, hard-to-replicate using stamping, hardware.

I have one question that I would like answered with this issue since it has so many peoples attention.

If source is going to be blamed here (again for PE), then why on the menu screen for disc 2 there is posterization above the blue horizon line? Surely you won't have me believe that this is a source issue? And yes, the posterization will move between the background changing.

I'm so sick of 'it's the source' arguments without a second glance. Somehow, I keep hearing this SAME argument over and over again with VC-1. Someone needs to admit the problem and fix it already. I want explinations, not excuses.

Most likely that was an engineering choice that dictated the use of advanced video codecs (VC-1 or AVC) and audio codecs (LPCM is not a good fit, whereas TrueHD or dts-HD MA still fit). A choice of software techniques over hardware.They could have spun the discs at 1.5x and still used advanced codecs. And the audio codecs help with space more than bandwidth. I have a feeling that if they had to do it over again they would use the 1.5x spin rate. I doubt they realized how much it would limit them for lossless audio and I don't know if they even really considered the seamless branching cases, where their 30Mbps limit looks like a lot compared to how much they are limited right before possible jumps (even though that limit is an average bitrate limit for a short period before that). I don't think it would have even take that much for them to have spun a little faster and had specs of 30Mbps for the video only and 40+Mbps for everything, but changing that now would be a bigger deal than if they had chosen that path originally. We've been told that 1.5x minimum spin rate is (or was at the time) being considered for TL51 discs, but I'm fairly skeptical that we will see a change to 1.5x, at least for this go around.

From one thing Amir posted, it looks to me like at least one of the studios wants to start using seamless branching to save money, or at least for something that won't really help the end users. If they do, then the people doing the encoding are going to have to start dealing with a lot more limitation than they had to in the first year. Higher bandwidth could definitely help with that.

--Darin

No you don't. You want someone to agree with you. You have obviously made your mind up what want to believe and now all you want is validation.

Robert,

First, thank you for skipping my question all together. You just proved my point with people passing by without looking at the problem.

Second, I want reasons which will make results, not excuses. So far I've heard nothing but excuses and spin to defend codecs and formats when a less than optimal transfer is released. With this type of attitude, NOTHING IS FIXED and all of us lose in the end.

Third, the politics, postering, and politicising of these types of issues makes me want to vomit. I deal with this kind of attitude everyday at work. At work all I hear everyday is that it's NEVER X departments fault when something is messed up, it's always another departments fault. Gee, doesn't this sounds familiar. "It's not the formats fault, it's the codec!" "It's not the codec fault, it's the source!".

I deal with this at work, I do NOT want to deal with it watching HDTV.

Now, if wanted this 'validation' you claim I need, how about I invite over my entire neighborhood, point it out, and ask if they see the same thing. Want to take bets who will and who wont? Perhaps I'll bring my player to the entire neighboorhood and test tons of TV's. Hell, lets do a study on those with 20/20 vision or those with 20/15 like myself.

So Robert, in case my intention wasn't clear. I don't need glorious validation by you or anyone else. I see it, therefore it exists. You in your wisdom do not need to convince me other wise, nor do I need someone else to tell me 'I see it too!'.

Now, what I need is the highest possible quality that High Def disk offers. I have no axe to grind with codecs. In fact I could care less what codec what is used on any movie, weither it be VC-1, MPEG2, or AVC. Hell throw in MKV for computer codecs if you wanted to, it really doesn't matter what codec is used to me. I don't gleefully rub my hands together with a smile on my face when I see AVC over VC-1 used, or vice versa.

Instead, I pop in a disc and look at end result. And right now, the end result is lacking. Weither it be compressionist fault, source fault (you'll have a hard time convincing me the title screen is a problem of the source), codec fault, Q&A fault, or encoder fault. Whatever is causing the problem, it needs to be fixed. And yes I'm not blind, but this seems to happen more often with VC-1 encodes by Warner.

But what really irks me is the fact that each and everytime this type of issue is brought up, weither it be AVC or VC-1 that it's always the format/sources fault or the issue is politicised and polarized to no end. I say bullocks to that. We are now a year into one formats life, and rapidly approaching the year of another. I have no need, want, nor desire for political garbage regarding codecs or formats. These types of issues might have been acceptible since the technology was new, but the time for beta testing encodes is over.

Stop the spin, admin the damn problem, and fix it. It's that simple, and everyone benefits.

As the old saying goes: If you are not part of the solution, you ARE part of the problem.

If it's common with the other codecs then what titles does it occur in? Ice Age 2 comes up but then gets lost in the shuffle.

As I specifically said, the OP mentioned the problem to occur with the other CODECs as well -- I didn't create that reference myself. I know I've personally read posts of people pointing out posterization problems with the other CODECs so I've no reason to disbelieve him, but if your position is that the other CODECs NEVER exhibit this posterization, then take it up with him, not me.

That shot from Planet Earth is anything but subtle IMO. It's halfway between being an artifact and being an artistic effect.

When I say subtle, I'm not referring to it as subtle relative to human vision, but rather subtle with respect to a computer scanning it. It's a slow gradual change. That's probably the main reason throwing bits at it wouldn't help -- as Ben said.

There's a continuing reaction that the solution to any flaw that shows up when displaying a VC-1 encoding, is to throw bits at it. However, as long as the people who designed the CODEC and regularly use the CODEC disagree, I don't see the point in promulgating that as a solution -- surely, they should know. If they say the problem is one of needing operator attention, then that would seem a very safe bet. (Reminds me of an automatic transmission -- if the transmission is slipping and the fluid level is above the minimum level mark, is the fix to add more fluid?)

Course, there's always the other possible explanation of it being a digital display related issue. We really don't know what the actual data stream looks like.

Superman Returns also had a fair bit of banding, again, especially during the underwater scenes. Is it the source, is it the processing, is it the downconverting to 4:2:0 component video, or the lossy video encoding (in this case, VC-1), or a little of all of these?

Why, oh why didn't they mandate at least 10 bit, 4:4:4 component mastering this time around (besides thinking ahead and allowing true 1080p/30 & 1080p/60 as part of the specs.)? At least Blu-ray had the extra room and bitrate ceiling for it.

Now we're stuck with this mess again.

That's a good question, but like most I'll avoid a direct answer with more questions. ;)

It depends on weather or not it's just someone falling asleep at the wheel. Is the current pricing system only feasible if compression is done in some automated way that doesn't involve someone watching what it's doing from beginning to end?

The thing is that just such issues may be caught all the time, but since our finished product has been corrected, it's hard to tell if it's a fluke or not.

What seems to be happening here is a ladder of

A VC-1 encoded disk shows banding > It could be because of too low a bit-rate > HD DVD uses VC-1 the most > HD DVD needs VC-1 because it only has 30GB> proof that the 50 GB of BD 50 with MPEG 2 is the way to go.

Is there anyone here with the credentials to back that up, or is it just one of a dozen possibilities?

You left out the possibility of this posterization example actually resulting from the digital display's own interpretation of the signal.

We as viewers have no metrics by which we can deduce whether bitrate is "high" enough.

From what I've read, we know that increasing bitrate to a level that fully enables the quantization filter loop feedback is as much bitrate as can be usefully utilized without manual guidance. That suggests if we are seeing problems in quazi or greater complexity scenes (unlike the one in question), then increasing the bitrate may well help.

So from that perspective alone, I can easily believe bitrate is not the solution.

Yes, there's the issue of whether the post house is willing to do more than leave the CODEC on automatic. That translates as a tradeoff in convenience on one hand and quality of the result on the other hand. So where do the studios land on this question?

From what I've read, Sony is looking for 100% automatic. That's easy for MPEG-2, just set it high and see what comes out. The down side is in the detail/edge quality plus, of course, the bit consumption. MPEG-2 is not capable of providing as fine detail and as smooth edges as either AVC or VC-1. That doesn't mean it can't produce an enjoyable picture, but it can't adhere as closely to the master as either of the other two CODECs. So OK, you have the convenience of just letting it run, but you pay in terms of transparancy to the master.

In the case of AVC, you can also let it run, but it's probably going to have much the same blindnesses on automatic that VC-1 has. AVC is simliar to VC-1 in that it also pretty much bitrate max's at the point where its quantization filter feedback is fully enabled. Beyond that, it has greater limitations than VC-1 regarding detail and edge smoothing (tends to preserve detail less). So you have some of the same disadvantages as VC-1 (throwing bitrate at it isn't a global fix) and a worse situation for achieving transparency to the original master (detail tends to suffer due to less quantization specificity both in the frame analysis and in edge smoothing).

So it seems to me that VC-1 is at least potentially the optimal solution, except there remains that question -- how much operator intervention is needed Vs how much operator intervention will actually occur.

Got me! (But I'm satisfied that if one cares about getting the best possible HD image, VC-1 is currently the most capable of the three choices.)

PS. Ben -- If I'm out-of-line with anything I said above, please provide the appropriate corrections -- thanks :)!

You may consider Ben Wagonner and Amir to be liars, but I'm strongly inclined to believe them regarding the way VC-1 works. Throwing bits at the issue has been proposed over and over again by persons such as yourself. The responses from people who actually work with VC-1 both externally and internally have been consistent.

Do we know for certain the digital display itself is not contributing to the problem?

Trevor,

I think that the PE release is rather special. For the first time we have the first example of different codecs for the same source material in high def. We have the MPEG2 Discovery broadcast, the H264 BBC Broadcast, and the VC-1 HD-DVD/Blu-Ray encode. We also know the bitrates of these broadcasts and a general idea of the bitrate during VC-1 scenes. It's not a perfect compairson (say 3 different encodes for disc) but it's very close. For the first time, we can actually look at how each of these handled the source material. Granted, each of these depend on the method of delivery (FIOS vs D* or such for domestic us) but there are copies on DVR's out there that do have the best possible broadcast material for both codecs.

Also, we can be fairly certain that the digital displays are not the cause of this issue. Various screenshots from the disc and broadcast show the differences. Unless our monitors are only working on half the screen, we know it's not a calibration issue.

With that said, I'm not saying each did not have their flaws. Also that's not saying the H264 BBC Broadcast did not have the same posterization issues, they were only less. But in the scenes that have issues, the BBC version looked better and handled the posterization at higher bitrate than the VC-1 did with lower bitrate.

So the natural reaction (and possibly correct one) is that the 19 or so bits CBR compaired to the 10-5 VBR on the VC-1 during the worst scenes actually DID help in the reduction of the posterization issue. This leads me to believe that more bits might have been the answer. Take that how you will, but VC-1 obviously took the source different than the H264 version. Weither this was bit rate is yet to be determined. This also puts a serious hole in the argument that VC-1 is transparent to it's master. Select scenes off these broadcasts and disc's show that this not the case.

Spin by various people involved in each of these codecs can't change the fact of what our eyes are seeing. I seriously doubt it's an intrinsic problem with VC-1 per say, but something is not right.

So the natural reaction (and possibly correct one) is that the 19 or so bits CBR compaired to the 10-5 VBR on the VC-1 during the worst scenes actually DID help in the reduction of the posterization issue. This leads me to believe that more bits might have been the answer. Take that how you will, but VC-1 obviously took the source different than the H264 version. Weither this was bit rate is yet to be determined. This also puts a serious hole in the argument that VC-1 is transparent to it's master.

I understand what you are saying, but there is one comment that I question.

As just described in my above post, I do believe VC-1 to be "capable" of being closer to the master, and I suspect that most of the time it is indeed closer to the master. However, there are times when operator intervention is required in order to achieve that "closer to the master". If that operator intervention doesn't take place, or if the problem is in the master, then the very strengths of VC-1 can effectively turn against it.

If this posterization is not in the master, then intervention is needed in order to more accurately render the gradient.

If the posterization IS in the master, then what the CODEC is seeing is essentially a step function. Of the three CODECs, VC-1 is best able to represent a step function -- and in this situation, will be most strongly objected to as a result.

I hope what I'm saying makes sense to you -- if the master is at fault, then VC-1 is actually providing a higher fidelity image.

If the master is NOT at fault, then additional attention was required during encoding.

PS. Again, Ben -- if I'm out of whack on this -- please step in.

So do you understand how we feel when WB continues to put out discs which are limited by the HD DVD spec?

FWIW -- I've read on the insider's thread that studios have actually performed separate encodings for both (allowing higher bitrate for BR) and concluded there was no observable difference in the image. As a result they standardized on just the one encoding.

And yes, I understand your point :)

What other things are left for the operator to "adjust" for this posterization scenario, if not bitrate?

Regarding PE, a lot of that material was shot in 8 bit, using DVCPRO HD and HDCAM. This will certainly account for a lot of the banding seen in those samples. As soon as you start to shift the 8 bit levels in the color grade session you introduce banding.
Both are also heavily compressed (For a source format) using 6:1 DCT and 3:1 DCT, and both use subsampling to 1440 horizontally.

Also If those stills are grabbed using a software player you'll have a levels transform and color space conversion both from and to a 8 bit image, this will also introduce banding.

I agree here with you DigVid. There have been some truly disgusting releases from both sides that these companies are making big margins on that look identical to their SD brethren.

Identical?

I know I'm dissapointed with "The Fugitive", I think my LD provides a better representation of the opening sequence, though the HD DVD looks just fine to me for the last third (and OK for the second third). Still, that was one of several vertically filtered 1080i masters (for reducing 1080i jaggies) that were inverse telecined for HD DVD. Are those the ones you are referring to?

What other things are left for the operator to "adjust" for this posterization scenario, if not bitrate?

From what Ben and Amir have said, there are a number of control options available. The correct one or combination depends on the particular situation. It seems to be not a question of adding bits, so much as intelligently redirecting bits. That is why free training and consultation is provided to assist operators in getting "up-to-speed".

The ability to redirect resources to handle specific issues is apparently one of the strengths of the CODEC tools. It's desired to have the CODEC AI be capable of handling as much as practical, but it doesn't obviate the need for a knowledgable operator to catch the situations that the AI misses. As usual, human senses remain more capable than artificial senses in recognizing anomalies.

"redirecting" and "adding" bits might as well be the same thing, for the scope of this discussion. Some stage in the algorithm is essentially not getting enough bits, so you give it more. Whether or not those bits came from somewhere else, or you are just adding it globally is really beside the point.

"redirecting" and "adding" bits might as well be the same thing, for the scope of this discussion. Some stage in the algorithm is essentially not getting enough bits, so you give it more. Whether or not those bits came from somewhere else, or you are just adding it globally is really beside the point.

Actually, it's entirely the point. Adding them globally won't get the job done. Reallocating them specifically will!

You really can't talk global bitrate as a solution, the CODEC and tools simply don't work that way.

Actually, it seems to me the most important issue is whether or not the encoding is at fault to being with. That is -- is this problem already present in the master?

If so, then the encoding is not at fault (likewise, the CODEC is not at fault).

Actually, it's entirely the point. Adding them globally won't get the job done. Reallocating them specifically will!

W/o knowing exactly what the problem is, this is just as much conjecture as suggesting that globally boosting bitrate will solve the problem. The common theme remains that more bits were needed.

Actually, it's entirely the point. Adding them globally won't get the job done.Not in every case. I am confident that there are cases where it would. Just think about it. As the bitrates are lowered, there are different levels where more problems come in that need to be caught and fixed. With high enough bitrates, less problems should occur. How many less depends largely on the implementation. I think at least one version of AVC has a mathematically transparent mode even. With VC-1, I asked whether given a high enough bitrate VC-1 still would have had a problem in that case where the post house doing "Babel" for Paramount decided to use AVC for it instead of VC-1, and I don't recall ever getting an answer. I remember reading something about blockiness occuring with VC-1 and Amir's position was that they should have stuck with VC-1 and spent more time on it. Obviously, that wasn't their position as it got AVC, just like "Payback" did on HD DVD. Sounded to me like if they weren't paying enough attention and VC-1 had been used for VC-1, blockiness could have gotten through, but they fortunately (IMO) noticed it in time to make a decision.

--Darin

W/o knowing exactly what the problem is, this is just as much conjecture as suggesting that globally boosting bitrate will solve the problem. The common theme remains that more bits were needed.

The problem is that the viewpoint of those raising the bitrate issue (including yourself) has been specifically a need to increase the global bitrate to eliminate this particular issue. That is a completely diffirent philosophy from that of reallocating bitrate that is already present.

Further -- there is actually no clear indication that more bits needed to be added anyway. If the master is already posterized, then the encode is already correct.

We simply don't know!

Yes, there are three different images using multiple CODECs and it would appear the VC-1 image provides the most obvious posterization. However, if the master is posterized, that merely indicated VC-1 is doing it's job!

It comes back to the very same old problem. In the absence of access to the master by someone, there is no intelligent comparison from which judgement can be made.

Further -- there is actually no clear indication that more bits needed to be added anyway. If the master is already posterized, then the encode is already correct.

We simply don't know!Then you really shouldn't say:
Adding them globally won't get the job done.It might and it might not. It depends on the specific situation. I doubt a truly trustworthy codec expert would claim that there aren't situations where a bitrate higher than most common on HD DVD would have kept a problem from occurring in the first place, meaning there wouldn't be a possibility of the problem slipping through the cracks (since it wouldn't exist). Fortunately, in the past some of these studios have been willing to give the people doing the encodings plenty of time to find and fix problems. How much they will continue to spend the money that results from that position as much in the future, we'll have to see.

--Darin

What is the ultimate argument created by this thread? That no matter the CODEC, we need to crank the bitrates through the roof?

Every CODEC has it's flaws, and simply throwing more bits into it will not always eliminate those flaws.

If VC-1 was really the issue, then there should be ample evidence of banding of posterization in just about every VC-1 title, but there is not. You want like comparisons? Look at Happy Feet, then go look at Ant Bully (another VC-1 CGI direct digital) and see if it shows the same type of issues.

It seems that one encoder or group of encoders having more skills and a sharper set of eyes would be more feasible than saying VC-1 can not achieve the desired results at lower bitrates, period.

And displays will factor into the equation. If you have spent any time at all doing imaging work across multiple platforms and displays you would know that you can spot flaws on some displays that are invisible on others.

This can be traced back to the encoding monitors if they are not up to task... a CRT will hide many issues that a fixed pixel LCD will reveal, etc.

With ‘Planet Earth’ broadcast, we have no idea what the signal chain consists of before it reaches any given set.

There are just way too many possible unknowns here to place absolute blame on VC-1 as a CODEC, especially when it has delivered great results on many encodes with none of the banding or posterization issues singled out in this thread.

This honestly seems like more of the same "we need more space" agenda pushing that is being played out by one side here on a daily basis.

This is ridiculous, if everytime something looks awry in a vc-1 encoding, the argument is going to be made that vc-1 is actually doing it "right" compared to different results from different encodings and the problem lies in the master. It's a circular argument.

Not in every case. I am confident that there are cases where it would. Just think about it. As the bitrates are lowered, there are different levels where more problems come in that need to be caught and fixed. With high enough bitrates, less problems should occur. How many less depends largely on the implementation. I think at least one version of AVC has a mathematically transparent mode even. With VC-1, I asked whether given a high enough bitrate VC-1 still would have had a problem in that case where the post house doing "Babel" for Paramount decided to use AVC for it instead of VC-1, and I don't recall ever getting an answer. I remember reading something about blockiness occuring with VC-1 and Amir's position was that they should have stuck with VC-1 and spent more time on it. Obviously, that wasn't their position as it got AVC, just like "Payback" did on HD DVD.

--Darin

I was actually referring to the specific case under discussion, as per the OP.

As I understand, at minimum, the bitrate has to enable the quantization feedback filter. If the set values of average and max don't permit that, then I would expect block edges to become apparent.

To be sure, if the set rates do not permit correct encoding, it's not going to happen. The advantage AVC has over VC-1 is that it will give a "smoother" image. If Babel was recorded with relatively grainy film stock or with shooting conditions that accentuate that, AVC will look "nicer" and will not need as much attention to encode (less is being expected of it.) I wouldn't be surprised if that kind of consideration is in effect when AVC is selected over VC-1.

This is ridiculous, if everytime something looks awry in a vc-1 encoding, the argument is going to be made that vc-1 is actually doing it "right" compared to different results from different encodings and the problem lies in the master. It's a circular argument.

That argument hasn't been made (at least not by me).

You just appear to object to that as being one of the two possibilities. After all, there are only two possibility's.

In case you overlooked it -- here is the content of post #103 above:

Regarding PE, a lot of that material was shot in 8 bit, using DVCPRO HD and HDCAM. This will certainly account for a lot of the banding seen in those samples. As soon as you start to shift the 8 bit levels in the color grade session you introduce banding.
Both are also heavily compressed (For a source format) using 6:1 DCT and 3:1 DCT, and both use subsampling to 1440 horizontally.

Also If those stills are grabbed using a software player you'll have a levels transform and color space conversion both from and to a 8 bit image, this will also introduce banding.

This is ridiculous, if everytime something looks awry in a vc-1 encoding, the argument is going to be made that vc-1 is actually doing it "right" compared to different results from different encodings and the problem lies in the master. It's a circular argument.

What's more ridiculous is singling out a few titles as an absolute example of a CODEC's strengths and weaknesses, while ignoring 95% of other titles that said CODEC was used on which did not show any of those problems.

There is nothing circular to that argument, that's just turning a blind eye to push an agenda.

I sense your intent to be argumentative, rather than discussing what has been presented in the topic.

From what I've read, Sony is looking for 100% automatic. That's easy for MPEG-2, just set it high and see what comes out. The down side is in the detail/edge quality plus, of course, the bit consumption. MPEG-2 is not capable of providing as fine detail and as smooth edges as either AVC or VC-1. That doesn't mean it can't produce an enjoyable picture, but it can't adhere as closely to the master as either of the other two CODECs. So OK, you have the convenience of just letting it run, but you pay in terms of transparancy to the master.Where did you read that? It sounds very unlikely, AVC is more suitable for automatic solution because it doesn't need higher bitrate to look nice.

W/o knowing exactly what the problem is, this is just as much conjecture as suggesting that globally boosting bitrate will solve the problem. The common theme remains that more bits were needed.
That's a common speculation, but in the specific cases of gradiants and VC-1, more bits is rarely the required solution. It's in using other encoding modes, like Differential Quantization and Adaptive Dead Zone correctly. It's quite common that applying the right settings to content like gradients both fixes any quality issues and also REDUCES the bitrate.

Now, for MPEG-2 segment reencoding, throwing more bits at it really was the main thing that was possible (different quant tables are used as well). But our VC-1 implementation is much, much deeper than any of the MPEG-2 tools ever were.

Heck we have a Region of Intrest mode, where you can tweak encoding settings down to indivdual macroblocks.

As just described in my above post, I do believe VC-1 to be "capable" of being closer to the master, and I suspect that most of the time it is indeed closer to the master. However, there are times when operator intervention is required in order to achieve that "closer to the master". If that operator intervention doesn't take place, or if the problem is in the master, then the very strengths of VC-1 can effectively turn against it.

If this posterization is not in the master, then intervention is needed in order to more accurately render the gradient.

If the posterization IS in the master, then what the CODEC is seeing is essentially a step function. Of the three CODECs, VC-1 is best able to represent a step function -- and in this situation, will be most strongly objected to as a result.

I hope what I'm saying makes sense to you -- if the master is at fault, then VC-1 is actually providing a higher fidelity image.

If the master is NOT at fault, then additional attention was required during encoding.

PS. Again, Ben -- if I'm out of whack on this -- please step in.
That sounds right to me.

Robert,

First, thank you for skipping my question all together. You just proved my point with people passing by without looking at the problem.

Second, I want reasons which will make results, not excuses. So far I've heard nothing but excuses and spin to defend codecs and formats when a less than optimal transfer is released. With this type of attitude, NOTHING IS FIXED and all of us lose in the end.
Isn't diagnosing the problem an important step in finding a solution?

The codec and format is never to blame for a transfer issue!

Anyway, there's been tremendous efforts behind the scenes in studios to improve masters, and lots of titles have been cancelled because the masters aren't up to snuff yet. There's one high profile recent title coming down the pike soon that's already on its fourth remastering attempt by a well-regarded restoration house, and there are still master issues detectible.

I understand what you are saying, but there is one comment that I question.

As just described in my above post, I do believe VC-1 to be "capable" of being closer to the master, and I suspect that most of the time it is indeed closer to the master. However, there are times when operator intervention is required in order to achieve that "closer to the master". If that operator intervention doesn't take place, or if the problem is in the master, then the very strengths of VC-1 can effectively turn against it.

If this posterization is not in the master, then intervention is needed in order to more accurately render the gradient.

If the posterization IS in the master, then what the CODEC is seeing is essentially a step function. Of the three CODECs, VC-1 is best able to represent a step function -- and in this situation, will be most strongly objected to as a result.

I hope what I'm saying makes sense to you -- if the master is at fault, then VC-1 is actually providing a higher fidelity image.

If the master is NOT at fault, then additional attention was required during encoding.

PS. Again, Ben -- if I'm out of whack on this -- please step in.


Trevor,

Yes, I understand what your saying completely.

I don't think anyone is arguing the fact that any codecs on paper, can theoretically be nearly transparent to the master. However real life implementation does not always agree with the textbook so to speak.

If VC-1/AVC was transparent to the master, then we can stop all development of any new video codecs because we already have reached perfection. We both know this is not true. In the years to come there will be VC2-3 and MPEG5-6 or completely different codecs that will have the same claims. This is why I do not buy into the spin or advertisements of any company or entity claiming such. If there's flaws in an encode, be it MPEG2 (I can point out one example of it on KOH), or AVC I'm going to say something about it.

Now on to the next point. So as far as posterization in the master, lets assume there was posterization in it for the sake of argument. In the pictures of PE regarding the sunshots we see the posterization in both VC-1 and H264. Now lets assume that the source was the same and had the same posterization when encoded. The issue at this point is that VC-1 actually magnified the problem in the source and was not as transparent to the master as h264. The h264 encode shows better color gradation into the bands surrounding the sun and is not as noticable, whereas VC-1 shows larger and thicker posterization bands and less gradation.

This can be attributed to three things:

1.) The master's for the h264 and VC-1 encodes were completely different, with the VC-1 getting the inferior master.
2.) If the master was identical, the H264 in this instance handled the inherent flaw of the master better than VC-1.
3.) Neither codec had issues and handled the encodes properly. The compressionist cared more for the H264 broadcast, than Warners VC-1 compressionist.

#1 is quite easy to answer. It's simply something that we have to live with.
#2 There is an issue with VC-1 encoding. It needs to be admited to, corrected, and everyone benefits. It does not need to be spun into damage control.
#3 Warner needs more training with the compressionist, or a better QA staff. The end result is the same as #2, everyone benefits.

Also, explinations of being more transparent to a posterizied master can not be attributed to my original question in this topic. Why if the source is inheritantly flawed, does the Title screen show the exact same issue? The title screen was not part of the master, and should not be effected by any flaws it had. Yet, the issue is there too.

I'm just glad that this is the first time that there cannot be excuses by physical proof everyone can see.

Again, anyway this topic spins the problem remains. It still needs to be fixed, either with a better encoder or more competient staff. We are now a year into one format and nearly a year into another. These types of problems should have been eliminated and worked out a long time ago.

But alas, spin and damage control prevent this from being fixed. It needs to stop so it can benefit us all.

I forgot, but did anyone check to see if the dvd version shows the same artifacting in the infamous sun scene?

Isn't diagnosing the problem an important step in finding a solution?

The codec and format is never to blame for a transfer issue!

Anyway, there's been tremendous efforts behind the scenes in studios to improve masters, and lots of titles have been cancelled because the masters aren't up to snuff yet. There's one high profile recent title coming down the pike soon that's already on its fourth remastering attempt by a well-regarded restoration house, and there are still master issues detectible.

Ben,

I completely agree that diagnosing the problem is a very important step in resolving a problem. That is what makes me so disgruntled about the whole thing. It seems that everyone is pointing fingers towards different root causes of the problem and it's never diagnosed. We call that the work place salute where I come from, where it's always 'It wasn't my department!' by crossing our arms and pointing our fingers out from each other in a 180 degree mark.

The question here is what caused the breakdown of quality. Was it the way VC-1 handled the source? Was it the compressionist? Was it the Q&A staff? Was it the source alone?

Something happened with PE that one of those things failed. It can be isolated down to certain root causes such as those above, but as long as no one looks into it by pointing fingers then it will never be fixed. Someone or something dropped the ball or a combination of all of the above.

Personally I lean towards two things. First being compressionist error, and the other being that VC-1 didn't like the posterization of the source and magnified the problem. The great thing about diganosing it to either of those two potential problems is that they can be fixed and avoided in the future. That's all I ask. A fix with no finger pointing that works and makes everyone benefit.

Also, I'm glad that restoration of masters for high profile films are being done, and rightfully so. I love the idea of having pristine transfers for future generations to enjoy. I've seen some amazing work done on films I would have written off as being to old or damaged. But with a high profile new release such as PE, no restoration should have had to taken place.

I hope that makes sense.

Where did you read that? It sounds very unlikely, AVC is more suitable for automatic solution because it doesn't need higher bitrate to look nice.

Given the feedback filter in AVC plus the smaller quantization block option, I agree it would be an improvement over MPEG-2 (plus the greater efficiency). I imagine Sony would just continue to set a relatively high bitrate for automatic. However, given it's model is similar (though precedent) to VC-1, I'm expecting it could also hit circumstances where intervention is needed for an optimal encode. If Sony insists on 100% automatic, it has to result in "artifacts". A trade-off situation -- so what else is new :).

Only thing is, that trade-off could be the same issue that VC-1 might be under fire from in this thread. (Presuming it's not actually a mastering issue.)

This can be attributed to three things:

1.) The master's for the h264 and VC-1 encodes were completely different, with the VC-1 getting the inferior master.
2.) If the master was identical, the H264 in this instance handled the inherent flaw of the master better than VC-1.
3.) Neither codec had issues and handled the encodes properly. The compressionist cared more for the H264 broadcast, than Warners VC-1 compressionist.

#1 is quite easy to answer. It's simply something that we have to live with.
#2 There is an issue with VC-1 encoding. It needs to be admited to, corrected, and everyone benefits. It does not need to be spun into damage control.
#3 Warner needs more training with the compressionist, or a better QA staff. The end result is the same as #2, everyone benefits.

Bear in mind I haven't seen anything of this title other than what's been discussed here, and so I'm speaking from general principals.

But for #2, I absolutely don't want our encodings being in the business of fixing problems with masters, since that's something nearly impossible to usefully program into a codec. There's a rich industry out there of products that do this, and source issues really should be addressed upstream of compression. There's a place for tools that do both (basically including preprocessing and codec in a single GUI), but I don't think that's appropriate for A-list titles as an "always on" feature.

As for PEP/PSE, we even make the 8-bit 4:2:0 conversion its own step with its own QA, in order to make it clear what source the codec itself gets.`

If Sony insists on 100% automatic ...Where did this come from? I could say, "If Universal insists on 100% automatic ...", but I wouldn't do that (other than to make this point). Do you actually have something to back that up, or is it just speculation (or some spin from the other side)? You can ask paidgeek if Sony insists on 100% automatic, but I'm confident that the answer would be, "No" (and would probably wonder where you got that from). Seriously, did this come from Amir's statement before where I recall he first claimed that he was told that things for Sony had to be done in 2 days, but where my memory is that he later on wouldn't answer the simple question of whether somebody from Sony had really told him that, when it seemed to become clear that he probably never was told that by anybody at Sony?

--Darin

Bear in mind I haven't seen anything of this title other than what's been discussed here, and so I'm speaking from general principals.

But for #2, I absolutely don't want our encodings being in the business of fixing problems with masters, since that's something nearly impossible to usefully program into a codec. There's a rich industry out there of products that do this, and source issues really should be addressed upstream of compression. There's a place for tools that do both (basically including preprocessing and codec in a single GUI), but I don't think that's appropriate for A-list titles as an "always on" feature.

As for PEP/PSE, we even make the 8-bit 4:2:0 conversion its own step with its own QA, in order to make it clear what source the codec itself gets.`

I apologize for mis-communicating on #2. I explained it a little more in my subsequent answer to your post. I was thinking more a long the lines of magnifying the posterization from the source (assuming all things equal between the H264 broadcast and VC-1).

If so, this can be fixed so it better matches the source. Ie, if it runs into posterization off a source, it can better mimic the posterization instead of magnifying it. That's only IF the source's problem was magnified.

Obviously if it was the source material, it's something that should not be fixed by a given codec (as the goal is transparency to the master). Instead as you said, given to restoration specialists.

The issue at this point is that VC-1 actually magnified the problem in the source and was not as transparent to the master as h264. The h264 encode shows better color gradation into the bands surrounding the sun and is not as noticable, whereas VC-1 shows larger and thicker posterization bands and less gradation.

However, if the master (the source) is indeed posterized, then what you are seeing with AVC is a SOFTENING of the step transition -- that is a LESS accurate, not MORE accurate rendition of the master. What you are praising here is what would in reality be a filtering action on the part of AVC -- ergo resulting in LESS faithfullness to the Master (smearing of detail).

And so what you are saying is that VC-1 is in the wrong to accurately portray the step function that is in the master (true, an exaggeration relative to what the master was "supposed" to contain, but not relative to what it actually contains. Yes, an exaggeration relative to what AVC provided, but that's to be expected since AVC also softened it.)

What you are effectively saying is that VC-1 is more accurate in its portrayal of the flawed master content, however, that isn't what you want -- you want it to be sufficiently clairvoyant to render what is "supposed" to be there, not what actually IS there. Equivalently, you are saying that AVC is doing correctly by smearing what is actually in the master. (All of the above presuming the flaw actually is in the master.)

Do you see the contradiction within your argument?

This can be attributed to three things:

1.) The master's for the h264 and VC-1 encodes were completely different, with the VC-1 getting the inferior master.
2.) If the master was identical, the H264 in this instance handled the inherent flaw of the master better than VC-1.

But do you want the CODEC to decide when the master is correct and when the master is wrong? I would say it's perferable such decisions be left to the person monitoring the encoding process. To be sure, the safest solution is to select a CODEC that will always soften -- I suppose one could call that hedging ones bets. But is that really a desirable solution for high definition video encoding?

3.) Neither codec had issues and handled the encodes properly. The compressionist cared more for the H264 broadcast, than Warners VC-1 compressionist.

#1 is quite easy to answer. It's simply something that we have to live with.
#2 There is an issue with VC-1 encoding. It needs to be admited to, corrected, and everyone benefits. It does not need to be spun into damage control.
#3 Warner needs more training with the compressionist, or a better QA staff. The end result is the same as #2, everyone benefits.

Also, explinations of being more transparent to a posterizied master can not be attributed to my original question in this topic. Why if the source is inheritantly flawed, does the Title screen show the exact same issue? The title screen was not part of the master, and should not be effected by any flaws it had. Yet, the issue is there too.

I'm just glad that this is the first time that there cannot be excuses by physical proof everyone can see.

Again, anyway this topic spins the problem remains. It still needs to be fixed, either with a better encoder or more competient staff. We are now a year into one format and nearly a year into another. These types of problems should have been eliminated and worked out a long time ago.

But alas, spin and damage control prevent this from being fixed. It needs to stop so it can benefit us all.

Hey, I don't know the answer. I don't know what the actual story is with the master (though post number #103 was certainly an interesting speculation) and I don't know anything about the title screen and possible issues with that.

From what I've learned, it's entirely possible the problem with both CODEC encodings is lack of operator intervention. Since AVC provides softer transistions than VC-1, it's output was less objectionable in this situation. So is the correct answer to always go for the softer encoder when dealing with HD, or is it to put more emphassis on the operator paying attention?

I would hope the best solution would be whatever it takes to get the most transparent-to-the-master encodes.

Where did this come from? I could say, "If Universal insists on 100% automatic ...", but I wouldn't do that (other than to make this point). Do you actually have something to back that up, or is it just speculation (or some spin from the other side)? You can ask paidgeek if Sony insists on 100% automatic, but I'm confident that the answer would be, "No" (and would probably wonder where you got that from). Seriously, did this come from Amir's statement before where I recall he first claimed that he was told that things for Sony had to be done in 2 days, but where my memory is that he later on wouldn't answer the simple question of whether somebody from Sony had really told him that, when it seemed to become clear that he probably never was told that by anybody at Sony?

--Darin

Neither, I believe it was PaidGeek that indicated Sony Pictures wanted to be able to allow its own encodes to run without fiddling, their focus being throughput (MPEG-2 seems to lend to that) -- the above is my phrasing, but I think it captures the spirit of what was said -- basically, maintenance free encoding. It was from the insider thread. It's really not a critical point either way, but it contrasts with what appears to be required for optimal VC-1 encodes. I'm guessing AVC may also require a little more attention than MPEG-2.

Just a comparison. Too bad Jpeg doesn't have it's own deblocking filter.
http://img263.imageshack.us/img263/5021/pebandingcomparega6.png

Regarding PE, a lot of that material was shot in 8 bit, using DVCPRO HD and HDCAM. This will certainly account for a lot of the banding seen in those samples. As soon as you start to shift the 8 bit levels in the color grade session you introduce banding.
Both are also heavily compressed (For a source format) using 6:1 DCT and 3:1 DCT, and both use subsampling to 1440 horizontally.

Also If those stills are grabbed using a software player you'll have a levels transform and color space conversion both from and to a 8 bit image, this will also introduce banding.

Hey, what are you trying to do? Technical facts used to explain a point are not welcome here. Please check the forum you are in. The agenda here is to discredit VC-1, and by extension, Microsoft. ;)

Actually, welcome to the forum. It sounds like you actually have some knowledge about production in general, and production of the specific documentary being "discussed". As you have seen, posts like yours tend to get lost in the noise. Hang in there, though. There are still enthusiasts here whose agenda is learning and sharing.

Hey, what are you trying to do? Technical facts used to explain a point are not welcome here. Please check the forum you are in. The agenda here is to discredit VC-1, and by extension, Microsoft. ;)

Actually, welcome to the forum. It sounds like you actually have some knowledge about production in general, and production of the specific documentary being "discussed". As you have seen, posts like yours tend to get lost in the noise. Hang in there, though. There are still enthusiasts here whose agenda is learning and sharing.


Good point Robert about a very good post that you quoted.

Cheers,

Grant

What is the color resolution of those DVCPRO HD and HDCAMs? And how much of those underwater shots were shot with these cams? As I asked in the other PE banding thread, I was wondering if that is the cause of the banding (and VC-1 appears to make it worse compared to the h.264 version - which makes sense when you see 19 MB/s CBR versus a lot less assigned to the VC1 encode).

What is the color resolution of those DVCPRO HD and HDCAMs? And how much of those underwater shots were shot with these cams? As I asked in the other PE banding thread, I was wondering if that is the cause of the banding (and VC-1 appears to make it worse compared to the h.264 version - which makes sense when you see 19 MB/s CBR versus a lot less assigned to the VC1 encode).
DVCPRO HD NTSC is typically 1280x1080 8-bit 4:2:2. HDCAM (non-SR) is 1440x1080 8-bit 3:1:1 color (so only 50% as many chroma samples horizontally as HD DVD and BD's codecs can deliver). There is also 720p versions DVCPRO HD.

Either way, both have significantly fewer samples in the source than 1920x1080 4:2:0 can deliver (HDCAM is better in luma, DVCPROHD in chroma).

Yes, I believe a large share of the footage was shot on the Panasonic Varicam, which is very popular in the US.

That's 960x720 in 8 bit 4:2:2 and using a 40 mbit (at 24p) DV intraframe codec.

So your net chroma resolution is down to a 480x720 matrix.

Please note that I don't dismiss that the banding could have occurred at a later stage, just that the origination format and the whole post production process could explain a lot of the occurrences.

Yes, I believe a large share of the footage was shot on the Panasonic Varicam, which is very popular in the US.

That's 960x720 in 8 bit 4:2:2 and using a 40 mbit (at 24p) DV intraframe codec.

So your net chroma resolution is down to a 480x720 matrix.
Ah, Varicam. Great cameras, but well less than half the detail we can deliver to consumers with HD DVD or BD.

So you start with 480x720 chroma, and convert it to 960x540 chroma in 1080p 4:2:0, leaving a net 480x540, half of the 960x540 a true 4:2:2 source would provide.

Knowing the source is Varicam, I'm increasingly liking the theory that the banding is in the source, and that it was either obscured by the H.264 loop filter, or the H.264 encode went through another signal path that wound up softening the image prior to compression.

This thread is in no way intended to "bash VC-1" as a codec per se but to make the community aware of one of the single
most annoying and severe picture artifacts we still have to bear with our next generation formats - POSTERIZATION.

This is not a format war topic either as both sides suffer from it equally.
And this artifact is in now way limited to VC-1 as all three available codecs are prone to it.

Why did I choose to state explicitly VC-1 in the topic title then?
Well for once VC-1 is quite well represented here @ avsforum - I sure would love to hear some ("marketing-free") insights from Amir and Ben about this.

Other than that posterization just happens to be a really big issue especially with some of the latest VC-1 releases.
So even with the "very latest", applied generation of VC-1/PEP and considering that we are talking about (most probably highly optimized) high-profile
releases there is strong indication that there is in fact an inherent problem with the VC-1/PEP workflow regarding difficult to avoid posterization.

To put it very simple these are potential causes for the posterization we see all too often (in no particular order):

- it is already present in the source/master

- "flawed" 4:4:4 10/12bit -> 4:2:0 8bit conversion

- unsufficient hand-tuning by the compressionist for problematic sequences. eg. proper dithering can help to hide any obvious banding/posterization artifacts

- unsufficient bitrate/bandwidth for particular sequences. Problem is that the
encoder likely is automatically applying very low bitrates due to the "simplistic nature" of some of these scenes - this "should" be manually corrected in the second pass.
Best case scenario is that the encoder is able to indentify such sequences automatically.

I invite Amir and Ben to complement and/or correct this simplistic list.

To demonstrate the problem I will post the following screenshots of recent VC-1 releases.
My thanks goes to members Xylon and House for all their efforts in providing these.
These screenshots are for demonstration purposes only - results may vary slightly depending on the playback configuration
BUT I personally verified these caps with my XA2 and the PS3 - both times the results were just as bad as shown here :



VERY interesting to say the least is the comparison with the BBC h.264/AVC broadcast:



So again this leads to the conclusion that the source/master is not the issue here (how could it ever be with such horrible posterization as shown here)
but the VC-1 encoding. Everybody is free to draw any conclusions from the comparison with the BBC capture.

These are just single samples of the problem at hand. eg. the recent Planet Earth release shows severe posterization (among other artifacts) not just occasionally
but a several times per 45min episode - some of these shots last for many seconds (like the one shown above) - starting with the initial BBC logo...

As another example I haven't really seen any high-def release with underwater sequences without a hint of posterization
(-> eg. VC-1 releases U571, Superman Returns, Happy Feet,... AVC/h.264 -> Ice Age 2...).

The important question is: What can be done to avoid it in the future?

I sincerely hope that Amir, Ben and other insiders admit this issue and provide insights of their strategies to avoid it.

If that's true then i don't think the thread title is fair to Ben, Amir, Microsoft, or VC1.

Knowing the source is Varicam, I'm increasingly liking the theory that the banding is in the source, and that it was either obscured by the H.264 loop filter, or the H.264 encode went through another signal path that wound up softening the image prior to compression.

So the h.264 version is actually the incorrect and less pure version. That's interesting. ;)

Thanks to Ben, a.holck, and Jeff Williams (post in insider thread (http://www.avsforum.com/avs-vb/showthread.php?p=10417318&&#post10417318) ) for bringing some facts and sense into this thread. Let's face it, we have a bunch of amateurs which have no necessary expertise, no access to the masters, some superficial knowledge how compression is done, yet like to through out terms and draw conclusions. When I saw so many 'ifs' and assumptions in these posts, I know for sure these posts are useless :rolleyes:

I understand lots of members are heavy on specs and want perfect PQ/AQ, it is simply not realistic to expect something like this on a consumer level format, you are NOT paying the studios enough for what you are asking :) More specifically with regard to PE, we should realize how difficult to shoot these footages under the conditions they had and learn to appreciate what they achieved. If the state-of-art equipments only give you 1440*1080, why complaining about lack of 1902*1080p :confused:

Well, having said that I would of course wish we had a >8 bit home screening format someday.
Filmscans, Digital cinematography, HD CAM SR and AVCintra are all >8bit, and most color correction/DI is done at 32 bit floats. The DCI Digital cinema projection standard dictates 12 bit per channel color depth.

But the HD disc formats seem hard to sell to the average consumer as is, so an extra expense for the extra space, DSP power and cost of image circuits to do proper 10 bit output would surely leave it dead in the water. As always, it's a compromise.

I prefer to have a commonly available 8 bit HD format, than not having one at all ;-)

Thanks for the extra information on those cams. More things to ponder. :)

Well, having said that I would of course wish we had a >8 bit home screening format someday.
Filmscans, Digital cinematography, HD CAM SR and AVCintra are all >8bit, and most color correction/DI is done at 32 bit floats. The DCI Digital cinema projection standard dictates 12 bit per channel color depth.

But the HD disc formats seem hard to sell to the average consumer as is, so an extra expense for the extra space, DSP power and cost of image circuits to do proper 10 bit output would surely leave it dead in the water. As always, it's a compromise.

I prefer to have a commonly available 8 bit HD format, than not having one at all ;-)

If an 8 bit 4:2:0 30 GB HD consumer format is the best we can get at this time then I also prefer it to exist, rather than nothing. But if the limitations become too obvious then we should not pretend there could not be a better format coming along, especially if this one does not take off fast enough. It is up to the movie sellers to convince us their format is good enough. They will either manage to do it with this round of technology or they won't.

So I believe in enjoying what is here but but maybe not betting the farm on it long term.

- Tom

So you start with 480x720 chroma, and convert it to 960x540 chroma in 1080p 4:2:0, leaving a net 480x540, half of the 960x540 a true 4:2:2 source would provide.

Knowing the source is Varicam, I'm increasingly liking the theory that the banding is in the source, and that it was either obscured by the H.264 loop filter, or the H.264 encode went through another signal path that wound up softening the image prior to compression.


I don't think it is explained how a shortage in colorspace in the source can translate into a shortage of color depth. If anything, the (relatively) low chroma image resolution would result in more blurring of a color transition, rather than the creation of a sharp "step function" associated with a banding artifact.

Color depth is entirely apart from this phenomenon that takes place in a color space regime. The former is more of a magnitude property, while the latter is a spatial property.

Additionally, if one is to accuse the h264 of softening the image, then it needs to be substantiated if the entire h264 program suffers from such softening with classic screen detail. If you have seen both programs, can you really claim the h264 is visually softer than the vc-1 encode, all around? Similarly, can you apply a softening filter to the infamous vc-1 "sun scene" to duplicate what is supposedly happening in the h264 "sun scene". Then apply that same filter to other detailed scenes...does the detail performance still match between the 2 encodes?

Furthermore, what happens if you breakdown the vc-1 "sun scene" image into its Y and C components? Do the coarse gradients appear in the Y channel or the C channel(s) (or both)? If it is primarily a Y channel behavior, then that presents an additional problem to the "low colorspace" theory (if the artifact ends up not having anything to do with "color", in the first place).

It seems that these ideas need to be fleshed out more to support the "low colorspace and h264 softening" theory to be any more relevant to the classic "low bitrate" theory (Occam's Razor, and such).

I don't think it is explained how a shortage in colorspace in the source can translate into a shortage of color depth. If anything, the (relatively) low chroma image resolution would result in more blurring of a color transition, rather than the creation of a sharp "step function" associated with a banding artifact.
That very much depends on how the color is upsampled.

If the capture is being done via HD SDI, the deck is converting between the native bitstream at 10-bit 4:2:2, which is then captured.

Additionally, if one is to accuse the h264 of softening the image, then it needs to be substantiated if the entire h264 program suffers from such softening with classic screen detail. If you have seen both programs, can you really claim the h264 is visually softer than the vc-1 encode, all around? Similarly, can you apply a softening filter to the infamous vc-1 "sun scene" to duplicate what is supposedly happening in the h264 "sun scene". Then apply that same filter to other detailed scenes...does the detail performance still match between the 2 encodes?
It doesn't need to say anything about the whole encode, just the scene we're talking about.

I don't have Planet Earth in any form yet, so I'm basically answering as if a PSE user had reported a simliar problem without my having seen the source.

That goes w/o saying, that once you go from applying it "very general" to "very specific", it becomes much less assumably plausible, w/o doing some further "proof of concept" testing.

Applied specifically, we are left to wonder if a specific color conversion step can be identified that will turn a smooth gradient into a coarse gradient (like we see in the picture) and then some sort of h264 pass (that seemingly only gets applied to one scene, despite frequent earlier suggestion of h264 having a general "softening" effect on all material) can nearly "restore" the gradient to nearly smooth. Like I said, some proof of concept tests seem in order to really consider this a viable explanation.

...and if we are to accept that the infamous "sun scene" exclusively can get some sort of "special h264 encoding" that makes banding nearly disappear, that ironically opens the door to an equally plausible counter theory that the banding was never in the source and some "special vc-1 encoding" applied exclusively to that scene created the banding artifacts. ;)

...add to that, is it even possible to manually apply some degree of softening to the vc-1 sample scene that would mask the bands while still maintaining the outline definition of the sun, as shown in the h264 scene sample?

Hanky, you can't blur the banding out of the vc-1 shots, unless you don't want to see the sun in any detail. Believe me, I've tried. :D

If Ben's theory is correct then there's a super special clever h.264 encoder that turns nasty banding shots into fairly smooth gradiants. Just like if it was the source. Wow. Screw vc-1. Every BRD/HD-DVD should have AVC if it can work that kind of magic. ;)

Hanky, you can't blur the banding out of the vc-1 shots, unless you don't want to see the sun in any detail. Believe me, I've tried. :D

If Ben's theory is correct then there's a super special clever h.264 encoder that turns nasty banding shots into fairly smooth gradiants. Just like if it was the source. Wow. Screw vc-1. Every BRD/HD-DVD should have AVC if it can work that kind of magic. ;)
Now that you mention it, I think I have heard of one of the H.264 encoders having a debanding preprocessing feature. This isn't architecturally part of the codec, of course.

Removing banding from a source isn't that hard if you can do it explicitly - basically a low-pass filter to soften the boundary between bands. It's that edge that causes the perceptual problem.

That goes w/o saying, that once you go from applying it "very general" to "very specific", it becomes much less assumably plausible, w/o doing some further "proof of concept" testing.

Applied specifically, we are left to wonder if a specific color conversion step can be identified that will turn a smooth gradient into a coarse gradient (like we see in the picture) and then some sort of h264 pass (that seemingly only gets applied to one scene, despite frequent earlier suggestion of h264 having a general "softening" effect on all material) can nearly "restore" the gradient to nearly smooth. Like I said, some proof of concept tests seem in order to really consider this a viable explanation.

...and if we are to accept that the infamous "sun scene" exclusively can get some sort of "special h264 encoding" that makes banding nearly disappear, that ironically opens the door to an equally plausible counter theory that the banding was never in the source and some "special vc-1 encoding" applied exclusively to that scene created the banding artifacts. ;)

...add to that, is it even possible to manually apply some degree of softening to the vc-1 sample scene that would mask the bands while still maintaining the outline definition of the sun, as shown in the h264 scene sample?
Whip up a .avi file with source like you're talking about it, and pour it through your favorite H.264 encoder, and Windows Media Encoder. I'll help you with the WME settings.

I recommend mastering to IYUV codec, since that's native 8-bit 4:2:0, so anything you see on playback there is in the file, and the encoders can read that in native colorspace avoiding any codec-level colorspace conversion.

about the possible bluring from the AVC inloop filter.

Given
1) the easy for compression scene (so that VC-1 consumes only 5.3Mbps for a 1920X1080 picture)
2) the 19Mbps of the BBC encoding for a 1440X1080 picture
3) the fact that the AVC's inloop filter "kicks in" only at larger than q17 quantization
4) the light touch of the inloop filter that Ron (Mr1394) showed in the InsidersThread for a much more complicated scene (BBC 1440X1080 again)

i would be very surpised if the AVC's inloop filter has really "kicked in" at all in this "SunScene"

Whip up a .avi file with source like you're talking about it, and pour it through your favorite H.264 encoder, and Windows Media Encoder. I'll help you with the WME settings.

I recommend mastering to IYUV codec, since that's native 8-bit 4:2:0, so anything you see on playback there is in the file, and the encoders can read that in native colorspace avoiding any codec-level colorspace conversion.

I'm suggesting these ideas for you (or others who are curious) to try out.

The onus is not on me to substantiate your theories.

Hanky, you can't blur the banding out of the vc-1 shots, unless you don't want to see the sun in any detail. Believe me, I've tried. :D

If Ben's theory is correct then there's a super special clever h.264 encoder that turns nasty banding shots into fairly smooth gradiants. Just like if it was the source. Wow. Screw vc-1. Every BRD/HD-DVD should have AVC if it can work that kind of magic. ;)

I anticipated you would try it out. :D Your results are what I would have suspected, as well.

I'm suggesting these ideas for you (or others who are curious) to try out.

The onus is not on me to substantiate your theories.
I can point to hundreds of VC-1 titles on the market that include lots of sky shots without banding, so I consider my theories sufficiently validated for my and my customers needs :).

There's a lot of people making lots of assertions and assumptions around here about how codecs operate. Given that we provide an end-to-end VC-1 encoding solution FOR FREE, I continue to encourage folks to try their hands at it instead of just ruminating about it. Assuming one's got the skills to make a simple .avi with a gradient, doing the encode will only take a couple of hours.

Now that you mention it, I think I have heard of one of the H.264 encoders having a debanding preprocessing feature. This isn't architecturally part of the codec, of course.

...but now we are straying from your original theory of h264's penchant to soften scenes must be masking the banding that you suspect is in the original.

So which is it?...is h264 softening this particular "sun scene" just right so as to mask the banding in the original, while seemingly leaving the detail in other scenes intact, or are they now employing an extra "debanding" filter to achieve the result in the h264 broadcast version? Either way, you are dead set to believe that these bands came from the source, correct?

...But unless we talk about these problems and consider possibilities, no new discoveries will be made, and certainly no new progress in image encoding. So don't pooh pooh non-technical people speculating about plainly visible artifacts in the end product. Like some wags say, I don't care how brilliant you think your algorithm is, but the end result sucks sometimes. And it don't take no phd to see this.

I don't have any beef with raising issues at all, but I do question the value of amateur 'detective' work shown here, and to some degree, unrealistic expectation :) As others put it, I rather have a HD format that gets accepted and survives, than nothing at all. To me, most HDM titles are done reasonably well, except for a few titles (well, we all know what these are). They are so much better compared with broadcast HD (at least in US), I really wish there could be a much better real time encoder, right now live sports broadcast just looks horrible.

I do lament the deteoriating quality of discussion here since format war started, a lot of posts are simply intended to score points for or again a format, these kinds of posts really drown quality posts, noise level is way too high now.

Or maybe ... VC-1 just doesn't like Starburst patterns, specifically those with 14 rays. In fact, isn't it entirely possible that VC-1 might have problems satisfactorily resolving the color gradients of a scene of a rainbow w/ a pot of golden shamrocks at it's base? MSFT better go check to make sure there isn't a malicious bug planted in the encoder by a disgruntled engineer to make that particular scene be a source of drama and debate.

In case some people have missed it, we already have one insider who has worked with the D5 master of "Happy Feet"and states that the banding is in the master.

http://www.avsforum.com/avs-vb/showthread.php?p=10417318&&#post10417318

Sorry to burst everyone's little bubble in this wirtch hunt.

Now - let''s talk a little bit more about the severe and much more disturbing artifacting we are seeing on the Mpeg2 and AVC encodes we are getting on many Bluray titles ;) Such as Fantastic 4 (where the broadcast AVC version is actually much better than the Bluray) and X Men 3.

Sounds pretty far fetched to me.

Professional reviewers are saying the best VC-1 encodes have been done on Blu-ray disc at bitrates beyond the capability of the HD-DVD spec. (Deja Vu etc.)Sounds far fetched to me. Please link the reviews (plural) that say this. Note that I do not know of any VC1 encodes that have needed to exceed 36 mbits per second, since VC1 attains "transparency"to the master at rates WAY below this.

But do tell - I'd LOVE to know ;) Since some BD folks regard VC1 as the antichrist, I'm particularly surprised to hear talk like this :p

Although developers may have calculated in theory that VC-1 did not require a high bitrate, experience shows us that the best, artifact free VC-1 encodes are done at higher bitrates only possible on Blu-ray.Which experience has shown this?

Really? How did this happen? You don't think this needed any more bits?

But folks are discussing a completely USELESS (IMO) section of blurred background, that means nothing whatsoever, from a similarly useless frame of the movie, where the explosion in the picture and the movement of the camera have blurred the scene byond any recognition - and are using THIS as EVIDENCE as to which encoding is BETTER??? Are people serious??

It would be nice to see the master to know how much of the "GRAIN" folks are highlighting in the Bluray version was in the master, since I'll bet most of it is noise created by the Mpeg encode ANYWAY.

BUT - The REAL relevant things to look for in this image is the Mpeg Blocking artifacting that is present in the face in the captures. There is FAR more blocking artifacting in the Bluray version.

Furthermore, as I state in the first line of this response, the frame in question is completely useless in drawing conclusions as to superiority, since the movement and the explosion have reduced the frame to a blurry mess on the film frame itself.

Of FAR greater relevance are the other pictures already posted from this film that show a lot more detail. And the HD DVD VC1 version has amply demonstrated a visible superiority over the BD Mpeg version. Even in those scenes, the BD version has shown a LOT more Mpeg blocking artifacts than the VC1 version.

See the early pages of this thread should anyone need a refresher...

You seem to be implying that vc-1 will not introduce banding at any bitrate, regardless of how low, let alone 5 Mb/s.
No, I did not. Nor is there any reason to encode at 5 Mbps. If the codec simply messed up on a shot and didn't give it enough bits, you tell the encoder to adjust encoding modes for that shot.

Therefore, you are implying the banding is in the source, and vc-1 captured it accurately.

You furthermore imply that the banding in the source was introduced by a color conversion in the mastering/editing of the footage.


I'm stating that's the likely, but not definitively, the source of those artifacts.

Then you go on to imply that the broadcast h264 sample of the very same scene lacks banding because it generally softens images and thus has masked the evidence of these banding artifacts which are supposedly in the source, even though detail appears to be intact throughout the rest of the program.

Then you change course and imply that the broadcast h264 sample may be incorporating a "debanding" filter, as an explanation for how it is possible for the bands to not be there, the detail in the rest of the program can be intact, and the bands still originating from the source.

Those are two possibilities, for which we don't have enough information to discriminate between. Other possibilities exist, like some softening being introduced in the 1920x to 1440x conversion.

Then you imply that the correct settings of your vc-1 codec will prevent banding, but are remarkably coy in then conceding that the incorrect settings could, indeed, result in banding. The banding can still only come from the source.
The wrong settings could make it look like a black and white woodcut if one was sufficiently perverse. If your complaining that our professional compression tool assumes it's being operated by professionals, you're missing the point. We have other solutions for fire-and-forget encoding.

All the while, you haven't been particularly descriptive of what these "correct settings" are or where you are "redistributing bits" which prevent this banding from happening.
Can you let me know what other professioal compression or video processing tools you're familiar with that I can reference? I've mentioned adaptive deadzone and differential quantization as the two big tools applicable in this case. As for redistributing bits, in the professional workflow, 2-pass encodes typically use somewhat fewer than the maximum alloted bitrate to leave some headroom to add bits in where needed. This is standard practice with any codec, going back to DVD.

Is that about right? No doubt, there will be some here who are content to take that hook, line, and sinker. However, please do grant the forum with some level of respect and acknowledge that there are others here who are not inclined to take this on face value, on the count of it being a bit all too convenient, and considering you have a vested interest in presenting it this way.
You've got access to the tools to do you own tests, and continue to imply you've got some domain expertise. I don't really understand what kind of information you're looking for from me. What kind of data do you think I can share that would make sense to you?

regarding Deja Vu:

This high bitrate VC-1 encoding takes the codec to heights previously unseen by my eyes

and

This is very possibly the best VC-1 presentation that’s been released to date on either format.

and

In fact the bitrate on this disc peaks higher than HD DVD spec even provides for.

and other professionals notice what kram sacul mentions above

What impressed me most, is that the encoding attempts to preserve the hard parts – the grainier scenes, without resorting to grain reduction to help it along, which always results in softer details. No obvious grain reduction has been applied to aid with compression, nor did bit starvation appear a problem.

http://www.hometheaterspot.com/fusionbb/showtopic.php?tid/136887/

I am not going to quote every review done of this title, but all reviews point to Deja Vu having great PQ. Many also mention the very hihg bitrates. Remember HD-DVD is limited to 29mbps for video stream. ;)

Rdjam, the point is that if the codec is eating up the grain then it can't be considered "transparent to the master". Yeah, it's a shaky blurry shot with no real detail but the mpeg-2 manages to actually resolve grain while the vc-1 renders it as mush. VC-1 is supposed to be better than mpeg-2 and it sucks in this comparison, no matter how useless you think grain is.

I'm not even going to comment on what you think is grain and what you think is mpeg-2 noise because you've shown you're really bad at that. ;)

See the early pages of this thread should anyone need a refresher...

That's the other threads. lol This thread is about the banding fiasco.

Some may argue that more HD-DVD/Toshiba complain threads mean they are more popular than blu-ray! :p

Back on topic, I think we blu-ray users are save with blu-ray exclusive title using VC1 upto 40mbps! We just need the studios to drop the HD-DVD bandwidth limited encode! :D

regards,

Li On

Krinkle, you only need one thread to complain on bluray players because 99% of them are the PS3.

Chad is a member of AVS. If he clarifies that he has received no financial incentive from the BDA, Sony specifically, I'll give a little more credibility to his forum and his reviews.....until then ;)

I believe that Sony backers are retaliating in this forum against VC-1 since Uni, Warner, Paramount and now Disney are all increasing their use of VC-1 and decreasing use of mpeg--2 for HD disks. Simple explanation.

I cant even believe some people are trying to defend poor quality encodes like this. Total shame, and a dis service to the users here.

Perhaps a new avatar is in order for a member, since posterization is a non issue?

http://img143.imageshack.us/img143/125/image1hj5.gif

"Goodly" HD indeed :rolleyes:

Krinkle, I just clicked on those threads and it appears there are more supporters than complainers even on those threads. Looks like only a couple of people in each of those threads are complaining....so the thread count may be high, but people count is low. The later seems the more important metric to me.

I cant even believe some people are trying to defend poor quality encodes like this. Total shame, and a dis service to the users here.

Perhaps a new avatar is in order for a member, since posterization is a non issue?

http://img143.imageshack.us/img143/125/image1hj5.gif

"Goodly" HD indeed :rolleyes:

I wonder if there is any way to get Warner to stop these shoddy practices?

As for PEP/PSE, we even make the 8-bit 4:2:0 conversion its own step with its own QA, in order to make it clear what source the codec itself gets.`
Maybe slightly out of topic, but just for my interest:

A while ago I suggested in the Industry Insiders Thread that PEP might want to accept sources in their native bit and chroma depth and that PEP should carry the full information through the encoding process as long as possible (while still outputting 8 bit 4:2:0 in the end). My suggestion was supported by one of the insiders (I believe it was Tom McMahon, IIRC). Amir said he wanted to look into this. Did you have a chance to test whether this approach would improve image quality and/or compression ratio?

Thanks!

No, I did not. Nor is there any reason to encode at 5 Mbps. If the codec simply messed up on a shot and didn't give it enough bits, you tell the encoder to adjust encoding modes for that shot.


I'm stating that's the likely, but not definitively, the source of those artifacts.

...


That seems highly likely to me. I believe you are saying that if the section in question was encoded at only 5 mbps then with a little more care the compressionist could have fixed the problem, either by changing some options or increasing the bit rate for that scene. And I'd agree. There is no particular reason to increase the bit rate for the entire movie if only a few identifiable scenes need it. And usually there is some reserve that can be found somewhere even if a few more bits are needed, as long as it is not extreme. Also, in this case the max bandwidth obviously doesn't come into play since the bandwidth seems to be occurring in otherwise easy to compress scenes where the automatic encoding didn't bother to give it much bandwidth.

So maybe more bits (or different options) would be needed, but only for isolated parts of the movie. That's sort of the way operator guided mmulti-pass compression works.

The way I understand it, all the highdef formats are coded with an 8 bit (per component) pixel depth. And this is usually sufficient to hide banding assuming it's not in the source.

Problems can occur when the encoder does not allocate enough bits to the scene. At that point due to encoding quantization the video effectively becomes compressed with 7,6,5 or less bits of precision per component. At some point this becomes visible and annoying. Now with a lot of detail, noise, or grain it is harder to see banding since the demarcations between adjacent areas are muddied by the other detail. But banding is much more visible on smooth clean gradients with very low detail since our eyes may pick out the only detail there as the transition between colors.

I'll wager Microsoft can jigger things a bit in the future so even the default parameters automatically get around this somewhat better in future versions of the encoder.

- Tom

Maybe slightly out of topic, but just for my interest:

A while ago I suggested in the Industry Insiders Thread that PEP might want to accept sources in their native bit and chroma depth and that PEP should carry the full information through the encoding process as long as possible (while still outputting 8 bit 4:2:0 in the end). My suggestion was supported by one of the insiders (I believe it was Tom McMahon, IIRC). Amir said he wanted to look into this. Did you have a chance to test whether this approach would improve image quality and/or compression ratio?

Thanks!

I work with medical imaging, and in my business, it is important that high bit-depth be used for the entire processing chain, until the final output to 8-bits. Otherwise, posterization will definitely occur. Now, it doesn't show up in images with some noise, but it does show up in smooth areas.

Even the act of converting data 8-bit RGB data to YUV 4:2:0 or 4:2:2 is a lossy one, reducing the number of gray levels to 220, and reducing the number of colors. Going the other way is also bad. This is why it is best for TVs to work with high bit depth processing, and never to actually convert the YUV to 8-bit RGB -- ever (always larger number of bits).

The quantization process in the compression scheme can lose gray and color levels, but it is up the encoder or compressionist to prevent posterization in smooth scenes.

If the mastering process consists of multiple phases (pre-processing, etc.), each puting the data into some 8-bit format, then posterization can occur even before VC1 does its work. So, for any particular movie, we can't really know who is at fault.

Sounds pretty far fetched to me.

Professional reviewers are saying the best VC-1 encodes have been done on Blu-ray disc at bitrates beyond the capability of the HD-DVD spec. (Deja Vu etc.)

Although developers may have calculated in theory that VC-1 did not require a high bitrate, experience shows us that the best, artifact free VC-1 encodes are done at higher bitrates only possible on Blu-ray.

I don't think anyone that has seen both King Kong and Deja vu can honestly say Deja Vu is the better looking movie....

TU for good information

I don't think anyone that has seen both King Kong and Deja vu can honestly say Deja Vu is the better looking movie....
I agree. King Kong looks much better than Deja Vu.

"This is not a format war topic either as both sides suffer from it equally.
And this artifact is in now way limited to VC-1 as all three available codecs are prone to it."

This thread title still has not been changed? Lame...

Off-topic and rule-violating posts removed. Warnings sent. Please keep this thread on topic.

"This is not a format war topic either as both sides suffer from it equally.
And this artifact is in now way limited to VC-1 as all three available codecs are prone to it."

This thread title still has not been changed? Lame...

The thread is called this way because we actually can discuss Posterization artifacts as a result of the VC-1/PEP workflow here @ avsforum due to representatives like Ben and Amir. It makes any discussion much more beneficial to actually reveive (non-marketing) feedback from the source.

I'm not quite sure why some of you consider me a notorious VC-1 hater/basher. The only thing I really "hate" about it is the kind of "strong marketing" and promoting (all while goofing on its alternatives) it receives here @ avsforum. So any criticism and discerning discussion may help "to restore some well-balanced views" instead of breeding new lemmings.

Well - there is one other thing I don't appreciate about VC-1 and any other codec for that matter -> encodings with just adequate/sufficient bitrates (12-18MBit/s avg. VC-1/AVC) instead of plenty.

BITRATE MATTERS -> the better and more revealing your setup gets (and who knows what the future will bring...) the more obvious this will become. (IMHO)

So for the record: VC-1 as a codec and the whole PEP workflow is capable of some truely outstanding results - but it still has flaws and shortcomings just like its alternatives. If avsforum is not the right place to discuss them I don't know where to go...

And for the record - my TOP 3 PQ favourites right now just happen to be VC-1 releases - in no particular order King Kong, Chronicles of Riddick and Deja Vu.

So in the end it is highly unlikely that I'm a true "VC-1 hater" afterall...

Ben,

would you care to give me an answer to this question:

"Given the (rather questionable) trend of ever decreasing applied average bandwidth/bitrates for common VC-1 encodings (an approach and "accomplishment" that your team in general and Amir in particular seem to be quite fond and proud of) don't you agree that MS's PEP application "should" be able to indentify sequences that are prone to severe posterization and therefor AUTOMATICALLY apply sufficient bitrates during the first pass. This would make the quality of the encoding regarding posterization somewhat independent from the attentiveness of the compressionist.

Sure, you can always argue that there is no "problem" with the VC-1/PEP workflow that causes posterization - it is just the responsibility of the compressionst to hand-tune these sequences. BUT in the end we as customers have to agonize over many releases with severe posterization artifacts.

I think high-profile releases like Planet Earth, Happy Feet and Superman Returns clearly demonstrate that relying on the proper actions of any given compressionist in order to avoid posterization is certainly not an effective approach - at least not with our 12-15MBit/s avg ~5Mbit/s minimum encodings."

"Given the (rather questionable) trend of ever decreasing applied average bandwidth/bitrates for common VC-1 encodings (an approach and "accomplishment" that your team in general and Amir in particular seem to be quite fond and proud of) don't you agree that MS's PEP application "should" be able to indentify sequences that are prone to severe posterization and therefor AUTOMATICALLY apply sufficient bitrates during the first pass. This would make the quality of the encoding regarding posterization somewhat independent from the attentiveness of the compressionist.
Is there a downward trend of VC-1 data rates? Certainly we can go lower than before, but I haven't been tracking the averages in any particular level of granularity. There was a big initial drop since some early pre 1.0 PEP titles used 20+ Mbps just because they didn't have anything else to put on the discs

Anyway, the task of varying the bitrate in order to achive a constant visual level is what 2-pass VBR encoding is all about. We've been doing it in codecs for over a decade now. Certainly, it's nothing that's ever able to get perfect - we've refined it a lot up to now, and continue to refine it, in order to improve compression efficiency and reduce the number of segment reencodes required.

Sure, you can always argue that there is no "problem" with the VC-1/PEP workflow that causes posterization - it is just the responsibility of the compressionst to hand-tune these sequences. BUT in the end we as customers have to agonize over many releases with severe posterization artifacts. And the ones you're agonizing over are mainly source based AFAIK.

I think high-profile releases like Planet Earth, Happy Feet and Superman Returns clearly demonstrate that relying on the proper actions of any given compressionist in order to avoid posterization is certainly not an effective approach - at least not with our 12-15MBit/s avg ~5Mbit/s minimum encodings."
Another insider who's seen the Happy Feet source clearly stated that the banding was in the source. Note that both Superman Returns and Planet Earth were also shot on CCD, not film, and filmmakers are still working on workflows to avoid banding when the source is 8-bit or even 10-bit with heavy processing.

I'm not sure what your point is about minimum bitrate - no reason to spend even 1 Mbps when the screen is black, and credits can be done with very high qualtiy at much lower rates than 5 Mbps.

Do you have the connections to actually examine the PE master for this infamous "sun scene", to ultimately verify that banding does or does not exist in the master?...or is this not of concern, regarding the capabilities of your codec?

Do you have the connections to actually examine the PE master for this infamous "sun scene", to ultimately verify that banding does or does not exist in the master?...or is this not of concern, regarding the capabilities of your codec?
I can try - I wasn't at all involved in that particular project. I'll doubt I'll have a chance for the next few weeks with both Mix '07 and Streaming Media East going on.

Is there a downward trend of VC-1 data rates? Certainly we can go lower than before, but I haven't been tracking the averages in any particular level of granularity. There was a big initial drop since some early pre 1.0 PEP titles used 20+ Mbps just because they didn't have anything else to put on the discsOne of your bosses (Amir) claimed that bitrates would be falling to what works out to about 9-12Mbps here:

http://www.avsforum.com/avs-vb/showthread.php?t=718689
Given this breakthrough, we are going to see more titles appear at average rate of less than 10 mbit/sec, bringing the general range lower by 20% to 30%. So no longer will I use the 12-15 mbit/sec for rule of thumb . :)based on encodings for a couple of titles, where he also said:
And this is not some random, special case, easy to encode content.That was now almost 8 months ago. I think your implication is probably right that bitrates aren't falling (at least not much), which is pretty much contrary to what some people were made to believe there. I freely admit that it bugs me when an insider misleads people coming here to learn, like happened there. There are probably still people who believe the content wasn't special in any way.

--Darin

I work with medical imaging, and in my business, it is important that high bit-depth be used for the entire processing chain, until the final output to 8-bits. Otherwise, posterization will definitely occur. Now, it doesn't show up in images with some noise, but it does show up in smooth areas.

Even the act of converting data 8-bit RGB data to YUV 4:2:0 or 4:2:2 is a lossy one, reducing the number of gray levels to 220, and reducing the number of colors. Going the other way is also bad. This is why it is best for TVs to work with high bit depth processing, and never to actually convert the YUV to 8-bit RGB -- ever (always larger number of bits).


Medical imaging is a much more demanding area. If you have a digital mammogram, the resolution is much higher (way higher) than 1920x1080. Shadow detail is so critical, they would laugh at what we settle for with HD and whatever codec we use. I have also examined film x-rays, and if you look at the information available at 1:1 it is incredible. As we all know, the negatives are huge.

That being said, I would compare HD encoding to a form of aesthetic compression vice scientific accuracy. Trying to present more with less, in a hopefully artistic manner. What we have with movies are the following- creating an illusion, producing a sense of motion, or highlighting a scene. In many movie frames 75% or more of the image is out of focus. How good the defocused area looks is a matter of bokeh and encoder quality. Making the focused area look good is a forte of VC-1. However, sometimes the eye drifts towards the out of focus area of the picture. There, I don't know if any encoder is perfect (or perfectible due to quantization losses), but assuming h.264 is better than VC-1 or mpeg-2, why not create a hybrid encoder that uses the best of all worlds? This is the challenge for producers of HD media. Let's hope they succeed.

Well, with "Happy Feet" and "Planet Earth" it *is*.

Only the HD DVD folks get their panties in a wad when someone dares criticize VC-1 because it has a monopoly on those encodings.

BR did a better job : they managed to have flaws in every codec's execution.

For those wanting apples to apples material to test VC-1 vs AVC results, the new Dreamgirls HD DVD has the feature done in VC-1, while the HD extras on disc 2 (including clips from the film) are encoded in AVC. I assume that's true for the BD version too, though I don't know that for a fact.

I just wanted to add my opinion here. I honestly think its not so much the codec(VC-1) as it is the Quality control. I just receiver planet earth and have seen banding in other scenes aside from the popular sun shot. Every single time, the ps3 meter reads at most 5 mbps. Thats like DVD bitrate or worse.
If any of you have seen Deja Vu on blu-ray, it uses VC-1 and has great picture quality and its because Disney used the VC-1 codec in a better manner keeping the bitrate nice and high, up to 40 mbps i've seen ! Don't blame the codec, blame Warner

I just wanted to add my opinion here. I honestly think its not so much the codec(VC-1) as it is the Quality control. I just receiver planet earth and have seen banding in other scenes aside from the popular sun shot. Every single time, the ps3 meter reads at most 5 mbps. Thats like DVD bitrate or worse.
If any of you have seen Deja Vu on blu-ray, it uses VC-1 and has great picture quality and its because Disney used the VC-1 codec in a better manner keeping the bitrate nice and high, up to 40 mbps i've seen ! Don't blame the codec, blame Warner

Exactly right. There's nothing wrong with VC-1 if it is used properly. Unfortunately Warner doesn't use it properly.

There's nothing wrong with VC-1 if it is used properly. Unfortunately Warner doesn't use it properly.

If anyone might be wondering why more companies like Warner don't participate in places like AVS, above is a big part of your answer.

:rolleyes:

One of your bosses (Amir) claimed that bitrates would be falling to what works out to about 9-12Mbps here:

http://www.avsforum.com/avs-vb/showthread.php?t=718689
based on encodings for a couple of titles, where he also said:
That was now almost 8 months ago. I think your implication is probably right that bitrates aren't falling (at least not much), which is pretty much contrary to what some people were made to believe there. I freely admit that it bugs me when an insider misleads people coming here to learn, like happened there. There are probably still people who believe the content wasn't special in any way.

--Darin

Darin, I have noticed in many of your posts that you seem to have a very high knowledge of encoding, bitrates, bandwith, etc. Your TL51 posts were made by someone who obviously knows about speed rates and related bitrates/bandwith. Are you the same Darin I know that does MPEG2 and now MPEG4 encoding for a living ? Not as a compressionist, I mean, but as an actual encoding processes developper ?

I just wanted to add my opinion here. I honestly think its not so much the codec(VC-1) as it is the Quality control. I just receiver planet earth and have seen banding in other scenes aside from the popular sun shot. Every single time, the ps3 meter reads at most 5 mbps. Thats like DVD bitrate or worse.
If any of you have seen Deja Vu on blu-ray, it uses VC-1 and has great picture quality and its because Disney used the VC-1 codec in a better manner keeping the bitrate nice and high, up to 40 mbps i've seen ! Don't blame the codec, blame Warner

This is exactly why I believe the title of this thread to be very misleading. Wouldn't expect anything less from an obvious BR fanboy. Funny thing is, once they start cranking up the VC-1 releases, all their threads questioning VC-1's posterization will be gone. This is not an inherent VC-1 flaw, it's an encoding execution flaw. Very different matter.

This is not an inherent VC-1 flaw, it's an encoding execution flaw. Very different matter.

Isn't that what I said? I said there was nothing wrong with VC-1 just the execution

Isn't the "execution" the key factor in all codecs, not just vc-1?

Isn't the "execution" the key factor in all codecs, not just vc-1?

Exactly thats why i said that for people complaining that planet earth has crappy PQ or too much banding or whatever, blame Warner, not VC-1.
All codecs can look bad as well if not done properly.
I'm looking forward to the POTC movies coming out. I couldn't care less what codec they use. As long as its not soft, not over saturated and no compression artifacts, i'll be happy :)

I did not say that Warner does not know what it is doing. I am quite sure that Warner would be capable of producing outstanding PQ if they wanted to. For some reason, they have other priorities.

I would say that my "qualifications" are that I have spent money on Warner products and been dissatisfied with what I received for my money.

It's not impossible that Warner has the best intentions with their vc-1 encodes, but for one reason or other, "using vc-1 properly" for this title entails some additional expertise/familiarity that are not so crucial for the typical encoding.

It's not impossible that Warner has the best intentions with their vc-1 encodes, but for one reason or other, "using vc-1 properly" for this title entails some additional expertise/familiarity that are not so crucial for the typical encoding.

Perhaps, but Warner did an excellent job with VC-1 on the departed. Both versions look great.

It would be nice to finally get a few samples (pq and problem shots) from that movie to show up in Xylon's screenshots topics. ;)

So it's possible that there's a weak link in the compression chain at Warner.

I can imagine some exectutive guy hanging over the shoulder of some nervous compressionist yelling "Go lower! Below 10mbps! The guy from Microsoft told me it's possible!". :D

Perhaps, but Warner did an excellent job with VC-1 on the departed. Both versions look great.

Some people just might disagree with you about that! ;)

Perhaps, but Warner did an excellent job with VC-1 on the departed. Both versions look great.

Actually, there has been a lot of discussion here about The Departed. Some of us think that The Departed is an example of what Warner is doing wrong. Some of us think that many scenes in The Departed look soft, and at least one poster has stated that this softness was not in all cases intended by the director.

Actually, there has been a lot of discussion here about The Departed. Some of us think that The Departed is an example of what Warner is doing wrong. Some of us think that many scenes in The Departed look soft, and at least one poster has stated that this softness was not in all cases intended by the director.

I haven't read that discussion about the departed but I agree, it did look a tad soft. I thought it was supposed to be like that. But I may see the other side as well. I recently saw Batman Begins on HDDVD and it had great picture but did look a tad soft. I also recently saw Ladder 49 and some scenes looked soft as well. Being soft i suppose has its benefit, as you can't see the film defects as much.

[...]
I can imagine some exectutive guy hanging over the shoulder of some nervous compressionist yelling "Go lower! Below 10mbps! The guy from Microsoft told me it's possible!". :D
LOL :D ... thanx for that, was a very good laugh ;)

but sometimes I really ask myself, why sometimes the bitrates are kept so low, although there's really so much space on the HD-Discs ... and why not max out the bitrate to it's limits and e.g. especially Warner could spend their movies a kind of "Superbit"-treatment on Blu-ray, instead of just "copying" the HD DVD-filesize- & bandwith-limited encodes (with sometimes adding a "little" PCM-track) ... ok, it's a little bit more work, but there are nearly 20GB more to use on BD50-Discs one surely can do better things with then leaving them empty & unused ;)

I haven't read that discussion about the departed but I agree, it did look a tad soft. I thought it was supposed to be like that. But I may see the other side as well. I recently saw Batman Begins on HDDVD and it had great picture but did look a tad soft. I also recently saw Ladder 49 and some scenes looked soft as well. Being soft i suppose has its benefit, as you can't see the film defects as much.

Softness may also result from techniques to eliminate grain; it has been my belief that someone at Warner thinks that viewers don't want to see grain.

Thanks for being open to the point of view that The Departed and BB show softness. I haven't seen Ladder 49. Softness is not unique to Warner; The Devil Wears Prada and Enemy of the State both look soft to me.

Some of us think that The Departed is an example of what Warner is doing wrong. Some of us think that many scenes in The Departed look soft, and at least one poster has stated that this softness was not in all cases intended by the director.

Anyone can think anything they want. There will be variations in just about every part of a playback system that can affect what one sees on the screen, not to mention any given individual's own experience in interpreting what they do see. Needless to say, unless the source master is available for direct comparison, anything anyone says is simply an opinion, and in the absence of a source master, an uninformed one.

Anyone can think anything they want. There will be variations in just about every part of a playback system that can affect what one sees on the screen, not to mention any given individual's own experience in interpreting what they do see. Needless to say, unless the source master is available for direct comparison, anything anyone says is simply an opinion, and in the absence of a source master, an uninformed one.

That's right! So we should all just keep our peon pie holes shut! How dare we discuss our <gasp> opinions <gasp> on picture quality without having the actual master playing right next to it.

I think that every single thread that has been started that discusses picture quality issues should be deleted by the mods here at AVS. Except of course for the ones where people did have access to the original master, as those are the only opinions that are "informed".

The one idea that is needing eradication is the idea that "vc-1 is inherently transparent, and is henceforth the defacto evidence of the source". With that, we can dispense with the assumption that anything else that looks different from the "reference vc-1" must then be in error. That will at least put our collective "uninformed opinions" on a level playing field.

The one idea that is needing eradication is the idea that "vc-1 is inherently transparent, and is henceforth the defacto evidence of the source".

Eradicating ideas? Interesting approach.

Darin, I have noticed in many of your posts that you seem to have a very high knowledge of encoding, bitrates, bandwith, etc. Your TL51 posts were made by someone who obviously knows about speed rates and related bitrates/bandwith. Are you the same Darin I know that does MPEG2 and now MPEG4 encoding for a living ? Not as a compressionist, I mean, but as an actual encoding processes developper ?I missed this post before and I'm not entirely sure it isn't a trick question :), but no, I an engineer on a medical product and not related to this stuff. But, I do find it very interesting and try to learn what I can. On an only somewhat related note, on the hobby side related to this site I am also interested in things related to contrast ratio, which is why I wrote this article:

http://www.hometheaterhifi.com/volume_13_2/feature-article-contrast-ratio-5-2006-part-1.html

I freely admit that I am picky about some of this stuff and push for improvements that many people out there wouldn't care about (which is how I think the bitrate stuff and contrast ratio stuff is related to some extent).

--Darin

The one idea that is needing eradication is the idea that "vc-1 is inherently transparent, and is henceforth the defacto evidence of the source". With that, we can dispense with the assumption that anything else that looks different from the "reference vc-1" must then be in error. That will at least put our collective "uninformed opinions" on a level playing field.

The idea that no matter how low the bitrate, VC-1 will produce perfect results. Or at least results that Warner apparently thinks consumers should be happy to pay their money for, and not complain, despite the fact that Universal releases consistently look far better.

Actually the source is the most important thing. No codec in the world can make s16 look like s35 on HD.

As for bitrates. There is no problem for a highdef movies to drop to DVD bitrates during certain scenes. Thats what makes VBR so effective compared to CBR encoding, dont waste bandwiths on low motion scenes.

As for Warner, I recommend KissKissBangBang. It beats most stuff in highdef and it isnt very large file. a VC1 encoding that really stand out.

Actually the source is the most important thing. No codec in the world can make s16 look like s35 on HD.

As for bitrates. There is no problem for a highdef movies to drop to DVD bitrates during certain scenes. Thats what makes VBR so effective compared to CBR encoding, dont waste bandwiths on low motion scenes.

As for Warner, I recommend KissKissBangBang. It beats most stuff in highdef and it isnt very large file. a VC1 encoding that really stand out.

KKBB is a very early Warner release; before they adopted their current practices. Another one in the same category is Good Night and Good Luck. The issue is not the capability of Warner compressionists but the orders they are receiving from higher-level Warner policy-makers.

The idea that no matter how low the bitrate, VC-1 will produce perfect results. Or at least results that Warner apparently thinks consumers should be happy to pay their money for, and not complain, despite the fact that Universal releases consistently look far better.

What?

Just so I'm clear... are you stating that Amir (anybody else) really has stated that VC-1 will produce perfection no matter what the bitrate? Or is this hyperbole on yourpart?

Why does everybody seem to take a low-quality product as evidence of what the tools are capable of?

Given what Michelangelo produced, I think I know what oil paints are capable of. I wouldn't look at a 1st-graders efforts to judge the tool.

What?

Just so I'm clear... are you stating that Amir (anybody else) really has stated that VC-1 will produce perfection no matter what the bitrate? Or is this hyperbole on yourpart?

Why does everybody seem to take a low-quality product as evidence of what the tools are capable of?

Given what Michelangelo produced, I think I know what oil paints are capable of. I wouldn't look at a 1st-graders efforts to judge the tool.

I was paraphrasing what an earlier poster had said, for the sake of friendly clarification, for the purpose of making clear what exactly is the mistaken idea that truly does need to be eradicated, regardless of its source.



I am not an authority on what Amir would or would not say.

KKBB is a very early Warner release; before they adopted their current practices. Another one in the same category is Good Night and Good Luck. The issue is not the capability of Warner compressionists but the orders they are receiving from higher-level Warner policy-makers.

If im not wrong BB were encoded before KKBB. The real difference is that KKBB is a S35 movie were movies like BB and Enemy of the state is anamorfic 35mm.

And I would like to se any evidence that Warner has some policy of softening their movies. Its is very unlikely.

I was paraphrasing what an earlier poster had said, for the sake of friendly clarification, for the purpose of making clear what exactly is the mistaken idea that truly does need to be eradicated, regardless of its source.



I am not an authority on what Amir would or would not say.

Gotcha.

regarding Deja Vu:

and

and

and other professionals notice what kram sacul mentions above

http://www.hometheaterspot.com/fusionbb/showtopic.php?tid/136887/

I am not going to quote every review done of this title, but all reviews point to Deja Vu having great PQ. Many also mention the very hihg bitrates. Remember HD-DVD is limited to 29mbps for video stream. ;)
Well I'll be, I have just realized that my original response to this has been deleted.

I can only assume it is because I made my conclusion about that reviewer in the open, so I'll repost my reply without my thoughts and allow anyone to make their own conclusion.

This is not an ad hominem attack. You have raised the issue that "multiple" proffessional reviewers have said x, y, or z, yet only cite ONE reviewer. So I think I am entitled to respond to your statement with some excerpts from his reviews to let people draw their own conclusions about the reviewer you have used. I also note that you cited no other cources, despite saying there were multiple.

The reviewer you are citing mostly reviews Bluray, and there are only two HD DVD reviews up from him at that site. It occurs to me that some of the statements in these reviews may be relevant to other readers' consideration of what is said.

Here are a few quotes to give you an idea of his thoughts, starting with the review you are citing:

I’ve been a little critical of the low bitrate VC-1 that HD DVD has content itself with. ... While moderate to low bitrate encodings generally look great, showing vast improvement based on a measure of DVD; with more challenging material, the codec hasn’t impressed me as much as AVC and even MPEG-2 at higher bitrates, particularly in the preservation of grain.He seems to cite a standard BD line that VC1 does not impress him as much as Mpeg2, which seems to fly in the face of reality.

In fact this one had me fooled the whole way thru. I assumed I was watching AVC (Buena Vista’s norm) and never saw anything that made me suspicious of anything else. Grain isn’t excessive in the visual design outside of some flashbacks and dark sequences. But it always looked well maintained. It was something of a surprise when I later popped the disc into the PS3 to find that it was encoded using VC-1.

In dealing with the audio, he again (as you will soon realize is common) states that LPCM soundtracks are better than TruHD, which he seems to claim do not sound as good. The reality, as we know, is that TruHD tracks are identical to the master. The encoding leaves little room for criticism. PCM currently offers the best of the best in terms of home theater sound. And this mix is more compelling than the best I’ve heard from TrueHD.

But let's take a look at the two HD DVD reviews by this reviewer shall we? He's only done two, it seems, with the first being "Feast":
The video reminds me a lot of all the cheesy, but great, low budget horrors from the 70s and 80s - very grainy a little on the soft side, with dark interiors filmed naturally, rather than lit up and corrected. Blacks are naturally elevated by soft, warm lighting and look especially noisy. It would be interesting to see what Weinstein might be able to do with the film with another 20GBs of space and a higher bit budget.Seems like a standard BD reference, here :rolleyes:

But both releases this week feature TrueHD. And thankfully it’s one of the rare TrueHD encodings that didn’t result in the temperamental Toshiba XA1 stuttering the sound every 10 minutes.Interesting, I must be the only one to have not heard of this problem. I've only heard that certain BD players stuttered when playing some DTS HD tracks.

The mix seemed a little constrained, possibly by it budget. It seemed a little underachieving in way of detail and suspense. Even in THD the sound didn’t approach the sharply articulate, completely immersive experience commonly found in PCM on BD. Silent Hill and The Descent delivered a richer, more profound sonic experience even at comparably lower volumes. But there’s still a chance that maybe something is being lost in the Toshiba’s conversion of PCM, which is one of the negatives of having such a small sample and only one real manufacturer of the HD DVD hardware market, unlike BD that has players from a number of different manufacturers already. Again, as you know, TruHD doesn't "bit constrain" an audio track, since it takes everything it gets and spits it out bit for bit. Suggesting that the Toshibas may not do this correctly is nothing but FUD in my book, as they use chips certified to do so by Dolby themselves. And then the side-swipe at Toshiba being the "only" vendor seems, again, to be a classic BD line.

Now, let's look at his review of "School for Scoundrels on HD DVD":
I almost feel a little guilty not giving the video a slightly higher score. Most of the time, it truly looks phenomenal in detail and depth. Anyone should be easily pleased. Like all good high def, the video goes places standard DVD couldn’t even dream. But, it’s still a little spotty.... But, it’s better than much of what I’ve seen from VC1 from Warner. And there’s plenty of detail to go around. The image can be a tad unstable. Though, this was never distracting. Again, another seeming knock at VC1, and Warner also, who has actually had some of the best Hi Def material around, yet here it almost appears that they are mentioned as an example of the worst.

What did burden the experience a little was some sporadic stuttering of the video. I’ve had the audio stutter on TrueHD encoded titles with the XA1. But, this is the first time I’ve seen it stutter with the video. It did so about every ten minutes and is probably a player issue more than a software one. Rewinding, I was never able to reproduce the problem in the same spot. But about ten minutes later it would happen again; finally straightening out about midway thru the film.Again, this mention of a supposedly common problem that many have never heard of. It almost seems designed to create the impression that this problem exists or is common.

We have an XA2 on order. But, it won’t be here for another few weeks. And from what I’m hearing it has problems uniquely its own as well.Very interesting that he would add this tidbit, since it would seem irrelevant to the review and also again seems designed to cast doubt over the player. I would LOVE to know what problems he is referring to, since I have had none.

And there appears to be the usual "faint praise" for TruHD:
For a comedy/romance, the TrueHD track sounds great. ... But, while the audio hardly leaves you unimpressed by DVD standards and should be transparent to the master being a lossless codec, I find myself thinking how it still doesn’t compare to the abundance of articulate detail of the lossless PCM tracks of similarly or even less action oriented films like Little Man from Sony..... But similar films in PCM have given me pause to be too quick to pass it off as a lackluster mix. Surround and effects aren’t nearly as distinctive or lavishly used to support the experience. Even up front, often times dialogue completely dominates what you hear. The TrueHD track is without a doubt a worthwhile alternative to lossy DD though.
Again, he is making a format-specific comparison in a review for a specific title, and can't help but expose what appears to be a bias, when we all know that TruHD would present exactly the same bits as LPCM when given the same material - but of course, TruHD would let you present MORE in the same space, since LPCM is a space hog.

Now, I could go on and on, citing other BD reviews by said reviewer, where problems and inconsistencies are glossed over, but I won't. I think the quotations cited above speak for themselves.

Did you want to present the other multiple sources you mentioned??

Let's just take stock of where we are in this discussion.

1) Happy Feet, which is undoubtedly a STELLAR presentation for PQ, with sharpness and detail beyond compare, has two scenes where hitting "pause" shows that there is minor banding in these two scenes.

2) Insiders here have testified that they have seen this banding in the master, indicating that it is not caused by VC1.

3) BD supporters have taken this as the subject of a thread complaining that VC1 is such a terrible codec, when VC1 has produced the best PQ releases on both formats.

Yep - doesn't make much sense to me either... :rolleyes:

KKBB is a very early Warner release; before they adopted their current practices.
What are Warners' current practices? How do you know?

Another one in the same category is Good Night and Good Luck. The issue is not the capability of Warner compressionists but the orders they are receiving from higher-level Warner policy-makers.
What are the orders that Warners' compressionists are receiving from higher-up? How do you know?

I'm genuinely interested to know why Warner might be compromising the quality of their releases; you seem to assert that you know this, so I'd appreciate hearing about it.

undoubtedly a STELLAR presentation for PQ, with sharpness and detail beyond compare,

Hyperbole much? :rolleyes:

You sure about that? Since the BD uses the identical encode it can't possibly be that good since according to you nothing on BD is as good as HD DVD! :D

What are Warners' current practices? How do you know?


What are the orders that Warners' compressionists are receiving from higher-up? How do you know?

I'm genuinely interested to know why Warner might be compromising the quality of their releases; you seem to assert that you know this, so I'd appreciate hearing about it.

A Warner compressionist used to post here but was ordered by his superiors to stop doing so. One can only speculate as to why they ordered him to stop.

How do I "know" what the instructions to Warner compressionists are? I don't "know." My belief as to what those instructions are is based on my evaluation, and the evaluation of others here, of the results.

The instructions appear to be: no grain; low bitrate; don't worry if the result is soft, as long as it looks a little better than the SD.

Why do I think they are compromising quality? With respect to the grain issue, I think someone at Warner believes viewers don't like grain.

Why the super-low bitrate? Perhaps the plan is to release "improved" versions at some later time and get everyone to buy everything again.

He stopped writing after he gave the critic to the PQ of a bluray title.

That people complain warner titles look soft is because they dont have any knowledge in the film medium.

There is no way the instructions have been no grain, low bitrate because a number of titles of warner has proven the opposite. Just because you cant see grain doesnt mean others cant. Grain comes in number of flavors and not only Pearl harbor grain.

How do I "know" what the instructions to Warner compressionists are? I don't "know." My belief as to what those instructions are is based on my evaluation, and the evaluation of others here, of the results.

The instructions appear to be: no grain; low bitrate; don't worry if the result is soft, as long as it looks a little better than the SD.
cjplay has clearly stated that this is not true and that (Warner) compressionists don't apply any noise reduction on masters prior to VC-1 encoding.

Let's just take stock of where we are in this discussion.

1) Happy Feet, which is undoubtedly a STELLAR presentation for PQ, with sharpness and detail beyond compare, has two scenes where hitting "pause" shows that there is minor banding in these two scenes.

2) Insiders here have testified that they have seen this banding in the master, indicating that it is not caused by VC1.

3) BD supporters have taken this as the subject of a thread complaining that VC1 is such a terrible codec, when VC1 has produced the best PQ releases on both formats.

Yep - doesn't make much sense to me either... :rolleyes:

1. Happy Feet has excellent Tier 0 PQ. Almost unanimous from both camps. There is one poster that indicated that he noticed interlacing artifacts on crowd scene. I did not notice any but I'm still waitng for a timestamp. I really want to know what he meant by it.

Curious.

2. I love the see the master prints. Insiders PM please ;)

3. As long as you ignore FUDsters including kids who just got their PS3 from their mommy. Aside from that there are members who don't give a crap about the format war but cares about the HD PQ they see on their viewing set. Their opinion I value. I just look at their post history.

Here lies the dilemma. The VC-1 when used properly will show how HD really look. Sharp,detailed and 3-D looking. No one is denying that but when we see something less than that then its understandably we get dissappointed. We hold HD DVD and VC-1 to a very high standard. This what made HD DVD a success since last year and hopefully continue to do so. I don't want to see anymore Traffic or The Game kind of releases. I do want to see more King Kong, The Hulk, Seabiscuit, Riddick, etc. kind of HD PQ.

He stopped writing after he gave the critic to the PQ of a bluray title.



I missed this what Blu-ray title?

I missed this what Blu-ray title?

There was a lot of contoversy about the exact circumstances of his being ordered to stop posting. I believe the comments in question were about Talladega Nights, but serious questions have been raised concerning whether those comments were the real reason for his being silenced.

post(s) deleted, a member suspended

Here lies the dilemma. The VC-1 when used properly will show how HD really look. Sharp,detailed and 3-D looking. No one is denying that but when we see something less than that then its understandably we get dissappointed. We hold HD DVD and VC-1 to a very high standard. This what made HD DVD a success since last year and hopefully continue to do so. I don't want to see anymore Traffic or The Game kind of releases. I do want to see more King Kong, The Hulk, Seabiscuit, Riddick, etc. kind of HD PQ.

Well said!
Having both formats, compared the best of both formats, VC-1 wins in the PQ department almost every time. I purchased into HD to see movie's look like KK and for me, mpeg 2 and 4 just does not deliver. I know the original film has some thing to do with the end product, but look what the studios using VC-1 has done with majority of vc-1 encodes.

I had some free machine time while I was at Mix, so I did a demo encode showing how gradients come from source. To avoid arguments, I used the freely available Elephant's Dream source, and am including project files for how I converted from the PNG sequence to a Lagarith encoded .avi file I used as source.

The encode itself was done with Windows Media Encoder and the current released version of the WMV9 Advanced Profile codec. The only thing not freely available is Premiere Pro which I used for the encode.

The clip uses HD DVD data rates (10 Mbps average, 25 Mbps peak), so you'll need a hot machine for full frame rate playback.

If you make a .avi, or look at the PNG sequence, you'll see that the output gradients very closely match what was in the source.

I did this at home on my own gear, so I'm putting it up as a .torrent.

http://216.99.212.230:6969/torrents/Elephants_Dream_WMV.torrent?9C267F763BFE9570D5C439ADD8B6D0D7 3C7708E6

Thanks Ben - appreciated!

What happens to gradiants when the encoder is forced to go below 10mbps to around 5mbps like that's been reported in the Happy Feet and Planet Earth clips?

With 25Mbps peaks and 10Mbps avg, it's going to dip below 10MBbps.

What I'm curious about is how does vc-1 break down when it's forced to do the impossible? Does it begin to look like a bad wmv clip (soft) or it like mpeg-2 where it just blocks to death?

What happens to gradiants when the encoder is forced to go below 10mbps to around 5mbps like that's been reported in the Happy Feet and Planet Earth clips?
There are lots of drops below 5 Mbps with gradients that look just fine, if the gradients are smooth in the source. And there's some banding gradients where there's banding in the source.

What I'm curious about is how does vc-1 break down when it's forced to do the impossible? Does it begin to look like a bad wmv clip (soft) or it like mpeg-2 where it just blocks to death?
WMV uses VC-1, so you'll see a similar kind of degradation, initially soft, but eventually with the block pattern becoming visible.

I chose 10 Mbps for average just to match what the .avi files up on the Elephant's Dream web site were using with H.264, but that looks to be a sufficient bitrate for this content. I could probably lower the peak rate without hurting quality as well, but I wanted to make sure I got some deep valleys in bitrate.

I had some free machine time while I was at Mix, so I did a demo encode showing how gradients come from source. To avoid arguments, I used the freely available Elephant's Dream source, and am including project files for how I converted from the PNG sequence to a Lagarith encoded .avi file I used as source.Being the kind of animation that it is, I would expect that source to be easy to compress. From what I've heard, there is pretty severe banding in the source (like where they are jumping across on the steps that come up, which I can see on the HD DVD), but can you tell us whether "Planet Earth" had the same banding in the source as in the recent VC-1 release? We could look at all sorts of other sources with banding, but seems like the real question is whether this source had the banding or not.

--Darin

I agree.

Regretfully, I am not hooked up for torrents, so I am unable to examine the video. I hope others examine and post screenshots...

I agree.

Regretfully, I am not hooked up for torrents, so I am unable to examine the video. I hope others examine and post screenshots...
Torret support's pretty trivial to get set up on any OS.

I see a couple of folks have already completed downloads, with several more in progress, so hopefully we'll have some indepentant weighing-ins soon.

Being the kind of animation that it is, I would expect that source to be easy to compress. From what I've heard, there is pretty severe banding in the source (like where they are jumping across on the steps that come up, which I can see on the HD DVD), but can you tell us whether "Planet Earth" had the same banding in the source as in the recent VC-1 release? We could look at all sorts of other sources with banding, but seems like the real question is whether this source had the banding or not.

--Darin

I think the source is the most likely cause for the banding.

Point any camera to the sun and the ccd and ad converters are likely to clip.
Even clamping down the iris or using filters will not prevent overloading the optics.

Evidence of this is that the observed banding shows that bands have a purple tint.
This indicates that one or two of the RGB ccd or ad converters goes into overload sooner.
If the encoder is having the problem I would expect banding without these (obvious) color shifts.

Filmscans, Digital cinematography, HD CAM SR and AVCintra are all >8bit

Via the HD-SDI connector and "EE" operation, both the Varicam and HDCAM f900 will deliver a relatively uncompressed signal which is coming via a more direct route from their imagers. It is 10 bit and 4:2:2

If it's the source that shows banding why is it so much less in the broadcast h.264 version?

If it's the source that shows banding why is it so much less in the broadcast h.264 version?
I'm looking forward to new posts by Xylon to see just how much difference there actually is.

Xylon, how are you coming along with this?

Rdjam, we've seen the difference in the sun shot. It's been reported and confirmed by many others. The HD formats have banding, the broadcast doesn't. It's inexplicable.

1) Happy Feet, which is undoubtedly a STELLAR presentation for PQ, with sharpness and detail beyond compare, has two scenes where hitting "pause" shows that there is minor banding in these two scenes.


Actually, the banding was obvious for nearly a minute starting at the 28 minute mark (some time after penguins jumped over cliff into the water). Using a timestamp of 00:28:15 should be good enough for screengrab purposes.

So the requirement to hit PAUSE is simply untrue. It is not a display issue on my end from the tests I've done - screengrabs, viewing on CRT, etc. However, no post-processing was done to the image - perhaps some displays apply a filter to hide banding?

I'm looking forward to new posts by Xylon to see just how much difference there actually is.

Xylon, how are you coming along with this?

Its coming. Lots of pictures going up. Planet Earth is going to be a dozzy.

If it's the source that shows banding why is it so much less in the broadcast h.264 version?

It's already been suggested that if the banding is in the source, the AVC version may be filtering the shot, thus reducing detail but also reducing the banding.

I think the source is the most likely cause for the banding.You could be right, but we know that Microsoft has close contacts with Warner and could ask them if the banding that is on the discs is also in the original master. If it is, then why not just say so? If it isn't in there, then I could understand going through the work of showing how VC-1 can maintain banding that is in a source (like in "Elephant's Dream"), but I don't think there is anybody here who would argue that VC-1 can't maintain banding that is in a master. Of course it can. But that isn't the question. I wonder if there is a real reason they haven't or can't just tell us if the banding is the same in the source for this one.

--Darin

...not only should it be asked if banding is in the source, but what degree of banding. Is the banding just minor in the source (just like as presented in the avc broadcast) or is it major (as it is shown in the vc-1 encode)? The authoritative word on the source material could easily just say that banding is present in the source, but they may be referring to degree shown in the former case, as opposed to the latter case. If that is the case, then the banding in the vc-1 encode is still not absolved (banding in the source is there, but not to the degree that the vc-1 encode is showing). The particular vc-1 encoding may still be responsible for some image mangling.

This is just to prevent someone from saying "ha-ha, I told you the source has banding", while conveniently failing to reveal that the banding in the vc-1 encode is still on a totally different level, and it is still the avc encoding that has been the most faithful to the source wrt the degree of banding.

If it's the source that shows banding why is it so much less in the broadcast h.264 version?
Likely a different master or different preprocessing was applied.

Why wouldn't the HD formats use the best master available? How many HD masters of Planet Earth can there be?

Why wouldn't the HD formats use the best master available? How many HD masters of Planet Earth can there be?
You have no idea :).

Also, it isn't always an issue of good/bad master. For broadcast, running through some kind of real-time noise reduction hardware, or using an encoder with that built in, is pretty common. This will remove detail (both correct and incorrect detail) making the encode look better at lower data rates.

I don't think 20 Mb/s avc is really hurting from "lower data rates" or noise from hd-cam sourced material. :D

I guess for the record, we should note yet one additional theory being floated that there are potentially 2 master shots of the sun scene- the avc getting the intact version and the vc-1 getting the mangled one? ;)

Rdjam, we've seen the difference in the sun shot. It's been reported and confirmed by many others. The HD formats have banding, the broadcast doesn't. It's inexplicable.
Can you link it? I guess I was hoping that Xylon would post these, since I have not seen them, nor seen any other posts confirming this to be the case.

Rdjam, the first couple of pages are chock full of people discussing banding in Planet Earth, Happy Feet, and other titles we haven't seen shots from, yet. Xylon's shots are on the way.

I'm going to post this comparison again because... it's sunny outside. ;)
http://img263.imageshack.us/img263/5021/pebandingcomparega6.png