Slashdot Mirror
Video Codec Comparison
FonkiE writes "Doom9 wrote a good article: After more than 3 weeks of work and no free time during that period it has been done: The latest codec comparison is online. 7 codecs have been put through one of the hardest tests in the history of codec testing. The results: find out on your own ;) I had planned to change the presentation somewhat but certain events (forum problems and such) prevented me from completing this for the release. I plan to eventually supply an updated version of the comparison."
Along a few of the pictures he says along the lines of "jpeg helped this codec. In this one it hurt this codec.." Granted he probably wanted to save bandwidth but why couldnt he post zipped up uncompressed files to download? Also, I think a single image with side by side comparisons of parts of each scene would be nice as I cant flip back and forth between all the pictures and remember what I liked and disliked about them all.
what is odd is that while the final mp3 shows only 12% rms distortion (actually that's a fair bit if you're an audiophile) the intermediate AIFF shows massive added noise when I convert by imovie. this added noise is spread over the spectrum but has significant components at 10Khz. the final mp3 converted by method 1 or 2 are simmilar and show the same level of distortion 12% rms.
being a signals processing expert but not a codec expert I'm totally at a loss to explain how distortion can show up in the aiff then vanish when reconverted to mp3 (or back to ACC). this makes no sense, and is actually impossible from an information from a (naive) signal/noise point of view unless the noise/distortion is either predictable or out of band form the codec's point of view. never-the-less this is reproiducible.
I'd like to publish this analysis on slashdot but first want to clear this up.
anyone got a clue?
Some drink at the fountain of knowledge. Others just gargle.
It seems that everyone is worried about encoding and what I would like to know is which format takes the least cpu to decode. I'm running on a rather weak laptop and even mpg1 gives me mismatched audio. Are the mpg4/SBC decoders less cpu intensive?
To be sure, he's comparing the performance of the codecs against content that is popular (and typically difficult to compress...)- but he's pulling it from a lossy format, namely the MPEG2 stream off of a DVD.
.OGG's or something like them instead of pulling them from MP3's as source. With a lossy format, you're deliberately losing content, introducing hopefully un-noticable distortions into the reproduction of the sound. Unfortunately, the varying formats use different assumptions, which produce differing kinds of distortions into the result. Because of this, re-encoding from MP3's, your sound files end up with distortions on top of distortions- the quality and compression suffers as a result.
There's a reason why you're really supposed to re-encode from the CD when you're producing
The same applies to video files.
This is not to say that what Doom9's doing isn't important or relevant- it is if you're using the codecs to space shift (i.e. Put several movies on your laptop so you're not carrying the DVD's on your business trip...) or pirating movies. The reality is that the codecs he's reviewing are largely designed for previously untouched video feeds- someone ought to test that as well.
Anything else is not really a proper review- unless you're only caring about ripping and re-encoding DVDs. To me, that's something useful to know about, but I'm as or more interested in the intended usages of this stuff as well.
I am not merely a "consumer" or a "taxpayer". I am a Citizen of the State of Texas
This test is amazing:
RV9 and WMV9 have the most visible tendency to smooth out details
I don't know where he got that (the end of the matrix page) but all the screenshots shows the opposite for WMV9: it shows a lot more details than *ALL* the others codecs.
If your screenshots shows the opposite of what you say, then I just don't know what to believe! Don't show them!
Write boring code, not shiny code!
You make a good point but it's not just DVDs that use compressed video. Digital cable, satellite, digital video cameras, etc. are all pretty common sources for people using these codecs and they're all compressed. Heck, most of the stuff I encode has been through compression/decompression twice allready, once by Directv (mpeg2)and once by my pvr (mjpeg) and when I want to store it long term I use mpeg4.
Europeans don't give a shit about stupid American laws.
>If you mean the people who developed DV... probably because the codec you describe was not developed when DV was.
Yeah, I do mean the developers of DV, however this doesn't appear to be the most complex codec to build in the world (if only the author's website was still up) -- I was just wondering why they never thought of this. Then again, at a decade old, I doubt any hardware could have handled it in realtime, so it makes sense that even if they had thought of this scheme they would have dropped it.
>You can bet the farm on DV's data rate and that is VERY important when writing to tape.
Ahhh, good point. Didn't think about that one. Thanks.
If you could be told what you can see or read, then it follows that you could be told what to say or think - BoC
I can't deny MPEG1 is technologically inferior to MPEG2.v 38/faq_v38.html
.mov recently. H.263...no sweat. Compile new vlc and ffmpeg and poof. Video works. Audio still in an unsupported format. Oops. I guarantee my parents would never have support for this on their computer, let alone the werewithal to figure out how to make it work. Meanwhile MPEG1 just works.
Although it is still not a catch-all. Example...part 8 of the MPEG2 specification--for higher-quality YUV--was dropped entirely for "lack of interest". http://bmrc.berkeley.edu/research/mpeg/faq/mpeg2-
But back to the point. So what? What good is a standard that isn't universal? I would LOVE a super ultra high-falutin 4:2:2, 12-bit YUV video codec to archive precious footage on. After all, VHS is so bad why not salvage every last detail while you still can instead of compromising quality further? Maybe I want studio-quality in my home. But if said codec is non-free and not platform independent it defeats the purpose of archival---preservation for the ages, or, in the case of smaller web files, guaranteed audience accessibility.
I was given a
"Not a spec difference, but in practice, only MPEG4 uses 2-pass encoding."
WTF? Never seen a DVD-Video, then?
This whole test is bullshit because a) all their source material was compressed and b) they left out some of the best GENUINE, commercial codecs. Total waste of time.
That was classic intercourse!
The following commands produce video that looks pretty good, at least to me (from an NTSC DVD source - "tccat -i
You'd need to replace $VOP with some video options (like scale=FINALWIDTH:FINALHEIGHT,crop=WIDTH:HEIGHT:TO
You make a good point but it's not just DVDs that use compressed video.
Good point, but it would have been nice to have multiple sources for the tests, not just DVDs. It could be that particular codecs are tripped up more by the compression techniques of DVDs vs. DV, etc., so your choice of codec might be influenced by your input source.
Ooh, a sarcasm detector. Oh, that's a real useful invention.
What I really don't understand is why he reviewed the videos on an LCD monitor, LCD's that are known to have inferior color to CRTs. CRT's can't reproduce the range of colors that a CRT can.