WMV9 Results in Codec Shootout

[pwimmer]

Hello,

Thanks very much for your great codec comparison. But there's one point where I disagree with your results: When I did my own codec comparison (DivX, Xvid, MainConcept H264 Beta, WMV9), WMV9 was by far the superior codec and there were no color problems at all.

I guess that you get best WMV9 performance only when using the Windows Media Encoder. No idea why Microsoft recommended to use the VfW-Codec, maybe Microsoft didn't handle your request with care and just sent you some default answer.

Well, I know your rule that you always use the settings recommended by the codec manufacturer, but in case the manufacturer obviously recommends stupid settings, you should really use your own (if a discussion fails). Else, you only compare the skills of the manufacturer's support department, not the codec capabilities.

By the way, it would have been better to test the recent advanced profile codec instead of the old one (you also used the recent versions of all other participants). Using the new features is explained here:

http://www.microsoft.com/windows/windowsmedia/howto/articles/codecadvancedsettings.aspx

In Windows Media Encoder, there's a speed/quality tradeof slider in the Tools->Option menu, which many people overlook...

Peter

[Doom9]

Well, you'll always find people who cannot accept the results.

I'm sure the guy who spent hours upon hours and sacrified parts of his holidays to have my decoding problems checked and encoded movies on his own (I think I made this perfectly clear that discolorations and shitframes were not a codec problem but some decoding problem that happened to ocurr on my computer at this particular time - and that I had not previously seen with any WMV9 content) will appreciate comments like this but in case the manufacturer obviously recommends stupid settings

I've had a long series of mail exchanges so there absolutely is no way that those settings were given do not make sense, and we had a discussion on which encoding tool to use. And had you bothered to read everything, you'd have noted that I've used one of the features of the advanced profile: the in loop filter - thus it's not necessary to use the WME after all.

And, you have the settings, the codecs are available so you can make your own comparisons.. but don't tell me you do not see blocks that are clearly visible even in the screenshots.. the only reason to make such claims would be if you cannot accept the results because they don't coincide with your world view.

You know who are the people who I am most likely to listen to: those who don't have any vested interests in any of the codecs. And you've made it perfectly clear which one is your favorite.

[Sagittaire]

IMO WMV9 VCM is very better than WMV9 encoder especialy with high quant encoding ...

[pwimmer]

Doom9, your replay is not fair. I DID do my own comparison (not with your files, but with my own samples) and the results were that WMV9 outperforms DivX and Xvid. And now I'm looking for an explanation (my guess was that the VCM codec is much worse than WME, but maybe I was wrong).

I never said that I wish that WMV9 is best nor that I prefer any other codec. I'm not in any way related to Microsoft nor do I write codecs myself. And I never said I don't accept the result. I just want to give some hints which you might want to consider in future tests, but of course it's your decision how you do your codec tests.

I'm a bit disappointed in the way you answer feedback. You should be more open to suggestions, and even if you don't think my ideas are any good, it would be better to be more polite.

Peter

[Neo Neko]

Originally posted by pwimmer
Doom9, your replay is not fair. I DID do my own comparison (not with your files, but with my own samples) and the results were that WMV9 outperforms DivX and Xvid. And now I'm looking for an explanation (my guess was that the VCM codec is much worse than WME, but maybe I was wrong).

Keep in mind after every comparison there is generally many identical posts saying you should have done something different or you are wrong. To me your initial post is a bit hard to fully understand. You come off kind of antagonistic if that was not you intent.

I would submit that your results are due simply to the source material. Or quite possibly to improper settings on your part or lack of codec knowledge. What settings and codecs did you use and on what sources? WM9 isn't any better than MPEG4-ASP in general and definatly not better than MPEG4-AVC which uses much of the same technology that WM9 uses. Microsoft has never really made a superior product. They simply make products that are "good enough" and then use their desktop monopoly to push said products into their market of choice. Having said that WM9 is not bad. Just not superior either.

[Manao]

pwimmer : I'd say that theorical coding efficiency of WMV 9 is higher than the one of MPEG4-ASP ( but lower than AVC, since it's roughly a lightweight AVC ).

However, the Microsoft implementations aren't good, and they may lack a good rate control ( at least, it doesn't hit the requested filesize, and perhaps it also distribute bits suboptimaly )

That would explain your results : were your samples a whole movie or only short clips ? If it's short clips, then WMV9 isn't hindered by its rate control.

[pwimmer]

I also noticed that WME never reached the specified file size. Therefore I only used the quality VBR mode which works fine (there was no need that the resulting file fits on a CD or DVD). My files were very short, they are avaiable here:

http://mitglied.lycos.de/stereo3d/3dmovies.htm

By the way, these are stereoscopic video files, therefore left and right half of each image are almost identical. Might it be possible that this special kind of content influenced my codec comparison results (I did all test with the "Zoom" clip)?