MSU Lossless Video Codecs Comparison - 2007 published!

[Maxim.Smirnov]

The comparison report was finally published (http://www.compression.graphicon.ru/video/codec_comparison/lossless_codecs_2007_en.html). By following the provided link one can find the report as a PDF file. Any comments and suggestions are still welcome anyway.

Many thanks to Dmitriy Kovalev, Zarxrax and akupenguin for reviews.

[MiroLx]

Hello.
Nice comparison. I have one question. Why you didn't add MLC Codec to the comparison? Was there any problem?

[smok3]

lagarith seems impressive, how does it work with alpha channel files?

[Zarxrax]

lagarith seems impressive, how does it work with alpha channel files?
It handles them really well, from my experience. You have to explicitly set it to RGBA mode in the settings though.

[Maxim.Smirnov]

Hello.
Nice comparison.


Thanks.


I have one question. Why you didn't add MLC Codec to the comparison? Was there any problem?

We tested some version available in December using a couple of files and found that it gives relatively medium results worse than some other codecs. That is why it was decided not to include this codec so that not to bloat the comparison list and minimize efforts. Actually, there are a few other codecs in the comparison that do not work very well, but we already had measurements for them, so they were kept.

It is possible that the test set will be made available so that anyone could use it.

Also we are planning to update the comparison within a year using only updated versions or new codecs.

[zambelli]

Nice work!
One request for the next comparison: it'd be great if you could include interlaced sources in your tests. There could be significant differences in performance and efficiency with interlaced sources in codecs which support intrafield encoding. Also, there are plenty of things that can go wrong with chroma mapping in YV12 interlaced encoding - it's crucial that these operations are truly lossless.

[foxyshadis]

As long as the codec itself doesn't perform any translation to or from YV12, there should be no problem with interlaced chroma. I can already tell you that if you force YV12 from non-YV12 in many of these codecs, or force YUY2/RGB output, they will sample incorrectly. (Having tested ffdshow huffyuv, ffv1, and lagarith.) MSU at least has an interlaced toggle, not sure if it affects compression or just conversion.

Can't tell you how they compress, but I suppose I could try a few things when I get a chance.

[System25]

Hello, today I've downloaded the comparison and it seems to be a little missunderstanding. LZOCodec works internally with YUY2 colourspace. Not YV12.

You can test it by compressing a video in YUY2 and YV12 an you will see a very little difference in compression ratio. Or by compressing a video in YUY2 and comparing with the original in order to see no differences between both.

But in general I'm very happy to compete. I will have some holidays soon and I will try to improve the codec.

Thank you for introducing it in the comparisson.

[Maxim.Smirnov]

Hello, today I've downloaded the comparison and it seems to be a little missunderstanding. LZOCodec works internally with YUY2 colourspace. Not YV12.


Hm, it looks like a bug in the comparison. We will check it and try to update the doc.
Thanks!

[NoX1911]

What about TSCC (Techsmith Screen Capture Codec)? Its RGB lossless only, very fast and a good intermediate codec. It comes bundled with Camtasia Studio. Its unbeatable at low delta material (i think). I use it for game recordings like amiga games. Those sources get 10x faster and better compressed than FFV1 for example.

[DmitriyV2]

We tested all available on that moment codecs, so we will test more in next comparison as we test in last one (http://www.compression.ru/video/codec_comparison/lossless_codecs_2007_en.html) more than in previous one (http://www.compression.ru/video/codec_comparison/lossless_codecs_en.html)