I've heard VC1 takes up half the space of MPEG2.

Well VC1 encodes the data much better so they can compress the info more while still mainting the look of the video. Were as in mpeg 2 the futher you compress the data the more macroblocking you get.

Most pepole complain becase bluray can use VC1 or mpeg4 which is better yet they chose to use mpeg2 becase it a older proven tech that is eazy to use. Although many of the newer bluray movies are comming out a lot better.

But I'm sure someone can better explain the major differences..

VC-1 can take up a tenth of a percent of the room as an mpeg2 encoding...just not with the same quality. What you are asking about is efficiency, that is, how much space a given compression scheme takes up with the "Same" quality video. Unfortunately, this is very hard to establish empirically (some people find certain artifacts more objection thatn other types). There are to date no published, widely available on efficiency. It is generally agreed that AVC and VC-1 are substantially more efficient than mpeg2 (and rumors have been flying about AVC vs VC-1), but given that a) the people developing VC-1 and AVC are working to improve the encoding process (so efficiency may improve) and b) VC-1, at least, had proprietary IP, it's unlikely we will ever see an objective quantatative study on this.

All of the codecs can produce stunning results given the right amount of space and bandwidth, and time and practice will tell which are the best applications of each (probably with mpeg2 slowing moving the the wayside as the others...and any future codecs...replace it).

It seems it's more about 2/3 for comparable quality from what I've seen (some may argue half, but I've not seen ~10mbps average VC1 encodes looking as good as some of the recent ~20mbps average mpeg 2 encodes).

From what I've seen AVC and VC1 are very comparable at similar bit rates -- they both have small advantages over the other, iirc... should look pretty similar given the same restrictions and an encoder that does it's job well.

To me VC-1 looks still much better than MPEG4/AVC. I'm happy that HD DVD has more VC-1 encodes than MPEG4/AVC.

I can't argue with what looks best to you, but I'll point out that some of the AVC encode I've seen (e.g. X-men 3) look as transparent to the source as anything I've seen on VC-1. Otoh, VC-1 has been used a lot more, so there can be little doubt that more expertise is being developed at the current time, and it is an excellent codec (kudos to MS and Amir personally). But it still comes down to the compressionist; even really talented people on big budget projects can put out bad vc-1 encodes (like Superman Returns) that are not nearly as good as the best AVC and VC-1 encodes.

Was SR a bad encode or a bad source? I think people confuse the two way too often.

VC-1 is just better at compressing than MPEG-2, why stick with something that still compresses but can't do it as well? Don't buy the propaganda that MPEG-2 is "more mature" because that's true but meaningless. I'd rather take a hot 21 year old than a hot 40 year old.

MPEG-2 has major limitations that AVC and VC-1 were created to address such as adaptive motion issues, block size reductions, in-loop deblocking filters and sub-frame optomizations that allow compressionists to craft better products.

BTW, Amir has been hinting that VC-1 can do great at ~10mbps. It certainly has been called reference at ~12.9mbps for titles like BB. My personal fav. pq/aq title is ATL and that fit on a 15GB disc...

Where did you see the source of X-Men 3?

Superman Returns a bad VC1 encode?? That is the first I have heard of that in those terms... Some don't like the articial grainyness but noone that I can recall had badmouthed the VC1?

Superman Returns was recorded on a Panavision Genesis Digital cam. I don't think people have mastered this cam yet but the results are improving (Apocalypto looked great). It's not yet as solid as film IMO but improving by leaps and bounds. I think SR looks good but not necessarily great and I think that's the source rather than the encode.

Scary Movie 4 was also shot on the Genesis IIRC, encoded in AVC and looks spectacular - I think it has much more to do w/ the tweaks Bryan Singer did on the print. In normal viewing cirmcumstances, one would be hard pressed to distinguish one codec from another - what counts is putting out a spectacular encode, and each codec can achieve that in its own context

Ok, roughly in order of posting above:

Forceflow: I'm not even questioning that VC-1 (and AVC) are more efficient compression algorthyms, I was just indicating that compression-artist interaction is still obviously one of the determining factors. I indicated (and support) that everyone will be moving away from mpeg2, I just dojn't think it will happen overnight (in part for political reasons...but that's business for ya).

The banding that is painfully evident in the underwater scenes in SR was not there in theaters, so I don't see how it can be anything other than a bad encode. I don't care about the grainyness at all when evaluating encode quality (I love the transfer of X3 and BHD, and both are very grainy movies).

I agree that the lack of deep blacks (the "solidity" and "depth" of the film) is indeed an issue with the source (the same problem Taledega Nights has), and is a problem that will continue to occur until digital video can take in a gamut that is as wide as film. It's also not something I consider when I say SR is not a terribly good encode.

Amir can hint all he wants that 10 mbps will look great, but until real numbers are released with product I can see, it's all spin. ATL does look awesome, but what is the ABR? I haven't seen BB, so I can't comment (I hope Warner hurries up and releases it on BD).

Too true.


Yes, banding is an issue but I don't think we can say whether it was an encode issue or a soure issue. In other words, who knows if high bitrate MPEG-2 would have solved the banding problem or introduced other issues.



Click took a beating for using Panavision Genesis cameras that got noisy during dark scenes. The same is being brought against SR. Frankly, the camera is important to the aesthetics of the movie. Some film stock is grainier than others, so one has to be cognizant of those factors when assessing an encode. Grain is good if it belonged there due to these factors, but certainly additional grain or noise is bad if it shouldn't be there.


I assure you that BB looks excellent. I dunno what the ABR of ATL is, but I'm sure someone else can either do the math or figure it out. I believe it was a Universal release and had DD+ 1.5mbps audio.

I tend to believe Amir when he says that VC-1 can make HD look great at 10mbps because of the innovation that has gone into VC-1 to make it superior to other codecs. When AVC gets the same toolset as VC-1 then we can see whether MSFT continues innovating or rests on its laurels.

But it wasn't in the original, so it's an encode problem, whether mpeg2 would have solved it or not. I'm really, really not bashing VC-1, which is clearly more efficient than mpeg2, and thanks to MS's investment is more polished than AVC (as I said in my first posts). I just don't think that any codec has gotten good enough (or automatic enough) that compression artists can fall asleep at the wheel and still tunr out flawless results.


Agreed, on all accounts.


I don't doubt it, I just can't comment on it sinse I haven't seen it. Hopefully this will be rectified soon


I don't doubt that Amir believes it, and I don't doubt that more investment has been made into VC-1 to make it the "codec of tomorrow, today!" and whatnot. But Amir and MSFT have an obvious (and totally justified) bias towards their product, and I simply will not accept their claims without being able to verify them independently. Sort of like I can't say anything one way or another about BB, because I have seen it (or Crank, for that matter, even though I don't entertain any serius doubt that either has fantastic PQ).

So back to the OP, I certainly would not accept the claim (at this time) that VC-1 is twice as efficient as mpeg2, although I wouldn't rule out the possibility that is may eventually be. And I see no evidence that VC-1 is in any significant way superior to AVC, although (again like I said originally) the tools are clearly under-developed compared to VC-1.

I'd agree although I feel AVC's lack of sub-frame optomization makes it a "shade" worse at handling darker movies, but as you (Ottscay) point out the compressionist can't fall asleep and expect a good encode with AVC or VC-1. MPEG-2 (on BD-50) does give the possibility of lacksidasical encoding and still getting a good quality product.

I'd say VC-1 is about as efficient as AVC, they share many common characteristics. Hopefully someone like Ben or Amir could clarify the differences but they are biased towards their own products. Its like "proud papa" syndrome, they love what they created and like to brag.

Actually, MPEG-2 is at least as prone to problems that require hand-tweaking as the other codecs. A-list DVD titles still get weeks of MPEG-2 tweaking to get optimum quality.


I'd say the single biggest difference is in the quality of our implementation - we have features that have never existed for any codec ever. One big difference in the code itself between H.264 and VC-1 is that VC-1 uses a less aggressive loop filter, which helps preserve film grain detail. VC-1 is also less complex to decode, meaning that VC-1 discs should perform better on HTPC compared to H.264.

LOL

I knew it...we will never have enough "My codecs kicks your codec arse" threads.....and i think SR looks that way because the director intended it to look that way....just like some other titles i can name like The Aviator with its greenish/reddish tint....Hi Def transfers are not going to change a films Intended look.....Maybe it will be so good it will enhance these so called "flaws"......

IMO , every Mpeg 2 encode I have seen including xmen3 and the transporter2 look artificially enhanced to give the effect of good encodes....

I will end this thread and answer your question:

VC-1 encodes are superior to MPEG2 and AVC.....I have 24 HD DVD's and 4 BD and every BD to me falls short in PQ compared to my HD DVD's....which mostly use VC 1...for the life of me i fail to understand Sony stubborn stance on VC 1...unless it has something to do with MS....for example, I can almost guarantee if the Fifth Element or House of The Flying Daggers were encoded using VC 1 they would be in tier 0 for PQ, plain and simple...Thats a test I wish we could do......

PS-I have also rented many BD as well just dont buy them....

Sorry, I own many HD DVD and BD disks, and you have failed to callibrate one input as well as the other if you honestly think that BD disks still fall short in any way from HD DVD disks in terms of PQ. Of course, that's not really any part of the discussion, since the OP was asking about efficiency, and what you were referrig to is indirectly coupled to codec efficiency.

Likewise, the better toolset of VC-1 (which only someone stubborn or foolish would contend at this juncture) does not make VC-1 a more efficient codec than AVC, although it certainly may make it more attractive to studios.

Ben, you guys should be proud of your work (and obviously are...deservedly); I agree that the that the loop-filter may be playing a significant role in preserving (at least some forms) of detail, but that doesn't make it more a more efficient codece per se.

This thread was not predicated on a "which codec is better?" question, and I don't think it's fair for one or two people to jack the thread and make it into some form of codec war. The question was a simpe one about efficiency, and as Forceflow answered (and Ben as well) it's clear that mpeg2 is not as efficient as VC1 or AVC, although it's essentially impossible to quantify the degree of difference (at least with publicly released data), and the latter two are very similar in terms of efficiency, albeit with a more robust implimentation atm for VC-1.

I am sorry your confused.....there is a bottom line and that is when the codec is used on the same movie for both formats they look exactly the same as far as PQ is concerned.....I have seen avc and while i think it is better than Mpeg 2 to my eyes VC 1 has given more consistent quality transfers....

I work with an HDV camera and I have to say that MPEG-2 isn´t a bad codec, the problem is that it takes a lot of memory and I work only in 720p mode, not in 1080p.

I have to say that I prefer this to the AVC when you have enough disk space to put the files, the reason is simple, is less CPU intensive than AVC but in a disk where you have less space you are forced to use things like AVC.

The problem is that I cannot take better PQ than the original video compressed directly to MPEG-2 with the camera, because is that "master" that I get, I never touch VC-1 as a professional because all works around MPEG-2, VC-1 doesn´t exist and AVC is used only in the cases that the disk space isn´t enough.

Some comparisons between MPEG2 and VC-1





source: http://video.ldc.lu.se/pict/WM9V-MP4...on-Goldman.pdf

What have weak blacks to do with camera gamut??? All footage is graded and as long as you have enough resolution in the data (and these cameras have more than 8 bit resolution) you can have your blacks at code 16 (black hole black) or any shade of gray you want.

As far as mpeg 4 not being equal to VC-1 I think it has more to do with the film itself. At the momment on HD_DVD on weinstein uses Mpeg4 many of their movies are not top box office multi million dollar movies. So they will not look as great. So a bigger budget film using VC1 will look better.


So as far as mpeg2 it requires more space to equal vc1s quility so that why a BR 50 may be needed and only a HD30 is needed for the same transfer using VC1.

But I think BR may have leared a leson and hopfully will start using VC1 so that we start getting good tranzfers. Although it make me wonder if the older BD movies will get a re-release. I'd think yes as sony loves to do this. Wonder if their will ever BE a SB blu-ray movie prehapps they can do the 5th element right as well as jay and bob strike back.

If they do I might have to get a BD player as those are some of my favorite movies..

I'm sure T2 will be redone as well that movies been on every single format mutiple times.

Fair enough, except that I said "deep blacks", and was not a reference to absolute IRE, but rather the ability to have a lot of dark shadows with detail still in them. The narrower light gamut of even hi def video means that you have to expose of lighting a bit differently to make sure all parts of the iamge the DP wants to show up aren't clipped to white or black. Regardless of how much color informatiuon you are grabbing, you are still getting less lumanence information in video. This tends to result in exposures that either have less contrast, or else are artificially contrasty at the expense of midrange detail.

I love digital production and am no film purist, but the wider light gamut of film does usually result in a different look (better or worse being aesthetic preferences) than HiDef video.

http://forum.doom9.org/showthread.php?t=114928

"~ HD-DVD Benchmark Challenge ~

You think that your standard is the best : prove that ... !!?



A - Introduction

The purpose of this challenge is to determine which is the best codec format encoding.
HD-DVD & BD can use MPEG2, H264 and VC-1 for video codec."


If you want to go further into the tech stuff, then i recommend this thread.

The usual tool for comparing video codec performance is "P-SNR plots". As you see in the figure below, there is a set of curves where x represents bitrate, while y represents "quality". MPEG2 will typically be one of the lower curves, while h264 and VC1 will be one of the upper curves.

Basically, this means that for most codecs, the quality improve as the bitrate increase. It also means that MPEG2 can give just as good PQ as h264 or VC1. The difference lies in what bitrate is needed for a given PQ, as any practical storage medium or internet connection has finite bandwidth.

Note that this comes at a cost: increased complexity means that more expensive hardware is needed for eahc generation of video codecs. Luckily, Moores law seems to ensure that PCs and dsps are getting more powerful for each generation.



Note that psnr may be thought of as a measure of "codec accuracy". It is not a perfect 1:1 mapping of human perception of quality (as humans we may even disagree).


BTW, I believe that Microsoft built their VC1 very much on the principles of h264. I havent seen the spec though (I believe that it was only published recently, Microsoft isnt really known for its willingness to document their "standards") If you look broadly on things, the general principles of video encoding havent been fundamentally changed since the late 80s. The methods have been refined, and all claims in media and scientific papers about "revolutionary concepts" seems not to have led to practical products :-)

-k

VC-1 is a SMPTE standard.

http://www.smpte.org/news/press_releases/003_06.cfm