OpenAVS

[bond]

opensource implementation of the in china developed/pushed AVS coding standard released:

http://sourceforge.net/projects/openavs/

[Valeron]

the draft is not yet public...

[guada 2]

Just a question:

the draft is not yet public...

Perhaps

AVS coding standard
But, which could be its row?

[Valeron]

But, which could be its row?

according to official introduction, quality wise it should be a great competitor to MPEG-4 AVC, and patent cost wise.... far far far... superior:)

complexity is also lower than AVC, easy to implement

90% patents of AVS belongs to China companies.
they co-state the technology should be very consumer friendly, we should pay little to watch AVS-encode program.

[bond]

obviously marketing and not more...

[Valeron]

obviously marketing and not more...

now there's open source project available, i think some talent guys will contribute to that and the only problem left is compatibility.
DSPs that are AVS decoding capable is available.

hopefully we could see another advanced coding system in the future.

[Sagittaire]

compiled version for win ... ???

[Valeron]

compiled version for win ... ???

there's only decoder implementation available at this moment....
and you've to compile the code by yourself.

i'm asking one of the three contributors to this project for a piece of standard draft.

hopefully they can help

[videomixer9]

http://www.sebone.de/download.php?file=08e52861dd6a7fb3aeb6e61c5805b4ce

compiled avsdecdll but I have no clue if you can use it with anything :P The other package with a player didn't build as it always complains about a file missing that isn't missing.

Honestly the presentation coming with it makes me it's save to ignore any efforts made by this group, it's another of those chinese inventions probably made cause chinese like to have their own way. And if you want to break free of MPEG-LA you can just use patent-free things and not patented in China ...

This thing seems to be a joke to me, and the openavs code honestly doesn't look like someone made real efforts, just the typical China joke.

[LoRd_MuldeR]

according to official introduction, quality wise it should be a great competitor to MPEG-4 AVC, and patent cost wise.... far far far... superior:)

complexity is also lower than AVC, easy to implement

90% patents of AVS belongs to China companies.
they co-state the technology should be very consumer friendly, we should pay little to watch AVS-encode program.

I doubt anything can keep up with x264 until I see a significant Sample !!!
This did not happen yet ^^

[videomixer9]

I bet you can wait another century for one and if you'll get to see one it'll turn out it's no competetion as it always was lately, otherwise it probably turns out most stuff is just a ripoff like typical chinaware, copy successful mechanism, pack them together and sell them as new for a cheaper price like usual. Looks like another scam grabbing fees from MPEG-LA for own profit.

[LoRd_MuldeR]

Well, who though about x264 when the first DivX was released?
The question is not if there will be something superior to x264, the question is when!
Maybe it even will come from China *lol*

But I'm sure it will take a few years from now...

[foxyshadis]

By then we might have the cpu power to harness wavelets in any useful fashion, and the research to understand how to best combine dct and wavelet engines for maximizing efficiency. We might get the holy grail of decent fmv over dial-up that's more than a talking head, and HD streaming. The question is just how long will it take. :p

But then they'll want to send HD over dialup and 3-D holovideo on discs! And we'll be back where we started. ;_;

[LoRd_MuldeR]

By then we might have the cpu power to harness wavelets in any useful fashion, and the research to understand how to best combine dct and wavelet engines for maximizing efficiency. We might get the holy grail of decent fmv over dial-up that's more than a talking head, and HD streaming. The question is just how long will it take. :p

But then they'll want to send HD over dialup and 3-D holovideo on discs! And we'll be back where we started. ;_;

History repeats itself *g*

[Valeron]

@videomixer9:

here is an brief introduction to AVS
http://www.avs.org.cn/en/index.asp

PS: Chinese mostly make jokes but not always.
patent issues is somewhat off topic in this thread, in order to compare with MPEG, especially for TV/IPTV program patent fees, i recommend you check MPEG-LA official page for more info.
All modern multimedia compression technology is very similar in terms of principle. The word "ripoff MPEG" sounds like MPEG is the only one with legitimate.
I've no idea about VC-1(how many MPEG members own the patents?), but the AVS group stated China companies own 90% patents of AVS(the rest 10% maybe MPEG members and other companies).
how an innovative tech can be patented is really hard to figure in terms of law, i'm not going to speak more on this issue.

I know Dirac is patent free, but can't understand why it's not popular? Hard to implement?

[LoRd_MuldeR]

I know Dirac is patent free, but can't understand why it's not popular? Hard to implement?

The industry prefers proprietary technologys.
Reason is clear: Keep the user stupid and make money :rolleyes:

And since most users will use what they find pre-installed on their home PCs...
...we see so many WMV files out there *sigh*

Guess how many users know the difference between container and codec!
0,5% ???

[Kopernikus]

Dirac is not yet usable.

[bond]

dirac is not patent free

[Kopernikus]

why not?

[bond]

http://dirac.sourceforge.net/faq.html#a6B

[Manao]

From a quick look at the source code, AVS looks a lot like any MPEG-like codec ( that includes VC-1... ). There are intra 16x16 & 8x8, inter ( P&B ) 16x16, 16x8, 8x16 & 8x8. The transform is a 8x8 integer DCT. There is a CAVLC, but no CABAC, there is deblocking - which looks quite like AVC's ( even to the naming of the variables... ), weighted prediction, multireference... There doesn't seem to be any interlaced support.

In the end, it's almost subset of AVC ( much like VC-1 is almost one too ).

[Sharktooth]

usual BS then :)

[Sirber]

sounds like AVC to me... that AVS.

[bond]

guys, lets cool down and wait for the specs

[Latexxx]

guys, lets cool down and wait for the specs

Yes. Without the specs this codec is useless piece of obscure code which you can't improve as there is no specification for bitstream.

[bond]

guys cool down plz! :D

actually its widely acknowledged in business science that one reason why the asian tiger states became big was the copying of existing technologies, learning how the technology works by doing so, and further developing it on their own becoming big competitors to the initial technology developers
i have no doubt in that china is trying the same

similar things are going on in the pharma industry

[bond]

if local law allows it, its nothing bad...
an example is south africa(n companies) not caring about patents owned by companies for anti-aids medicamentations...

of course those general statements about "china" doing this and that are obviously wrong, cause most of the time companies are the actors and not states. states are normally only helping "their" companies achieving their goals for whatever reasons (imho)

about the scientific references? no idea, i learnt this at the university...

[merbanan]

An AVS patch for ffmpeg has been posted. Expect the codec to be added to the repository in a few weeks.

[lionel3]

opensource implementation of the in china developed/pushed AVS coding standard released:

http://sourceforge.net/projects/openavs/

you can try out a windows installer at

http://prdownloads.sourceforge.net/openavs/OpenAVSInstaller.msi?download

after installation, look at the readme file for playback instructions.

have fun!

[videomixer9]

I tried it with the teststream found on mplayer ftp server. Very bad quality, it stutters around like mad and the CPU usage is beyond acceptable for such a small size clip, between 50-70% on Athlon XP 3000+. It has a some weird blocking problem and the sync is lost quite often on that demo player.

Maybe time to try the mplayer implementation, but if it is as bad as their standalone player, I think their version numbering is wrong, it's more like 0.0.0.1.

Well good luck on it but I don't have much hope. But well, at least the picture looked kinda sharp.

MPC played the stream audio only, using MPEG splitter. Well of course, got not the right video filter.

[lionel3]

I tried it with the teststream found on mplayer ftp server. Very bad quality, it stutters around like mad and the CPU usage is beyond acceptable for such a small size clip, between 50-70% on Athlon XP 3000+. It has a some weird blocking problem and the sync is lost quite often on that demo player.

Maybe time to try the mplayer implementation, but if it is as bad as their standalone player, I think their version numbering is wrong, it's more like 0.0.0.1.

Well good luck on it but I don't have much hope. But well, at least the picture looked kinda sharp.

MPC played the stream audio only, using MPEG splitter. Well of course, got not the right video filter.

You are sure you used the right test streams? there are 2 kinds of test streams, one without audio and one with audio and video multiplexed. If you use the windows binary, you are expected to test it with the multiplexed streams.

On my AMD Mobile Sempron 2600+, the player works just fine.

[videomixer9]

As fan of the mplayer samples archive I used this one here: ftp://ftp.mplayerhq.hu/MPlayer/samples/AVS/stream01.avs

Still playback sucked and had too much CPU load. I'd expected more from sth. using SDL. Noone said it doesn't work, it just has way too high CPU usage, stuttering playback and is way too blocky.

see it yourselves:
http://xs203.xs.to/xs203/06293/openavs.png.xs.jpg (http://xs.to/xs.php?h=xs203&d=06293&f=openavs.png)

[lionel3]

As fan of the mplayer samples archive I used this one here: ftp://ftp.mplayerhq.hu/MPlayer/samples/AVS/stream01.avs

Still playback sucked and had too much CPU load. I'd expected more from sth. using SDL. Noone said it doesn't work, it just has way too high CPU usage, stuttering playback and is way too blocky.

see it yourselves:
http://xs203.xs.to/xs203/06293/openavs.png.xs.jpg (http://xs.to/xs.php?h=xs203&d=06293&f=openavs.png)

try download and play this one

http://www.avs.org.cn/fruits/software/stream02.avs

It will look better. stream01.avs is full of night scenes.

[videomixer9]

Also if it fails on night scenes that horrible it is unusable, btw. they should've used a widescreen movie and still it has this stair effect due to the general block pattern that's allover and CPU usage reminds me more of a 720p h264 video.

edit: that one is not better, on bright scenes the blocks are even more obvious, it stutters and the encode is even newbish pseudowidescreen (how can you put that up as a sample? black bars encoded into the video).

[Manao]

videomixer9 : you seem not to be able to stop complaining about the codec. If you don't like it, leave it alone. It's slow because no single line of code have been optimised yet ( no mmx, sse2 ). It's more a base reference decoder than anything else at the moment. That explains easily at least a 4x slowdown.Also if it fails on night scenes that horrible it is unusableIt's not the decoder that is failing, it's most probably the encoder.I think their version numbering is wrong, it's more like 0.0.0.1.Version numbering have never meant anything at all.

[videomixer9]

Well if you name something 1.0 and it's open source I'd expect something finished to a certain degree, usually this careful numbering is applied to most projects to show that it's not a finished product. If it says 1.0.3 I'd think it's in a state that it has optimized code to a certain degree, in correspondence to other serious open source project this should be maybe adjusted a bit to reflect the current state of development a bit more in the number. Version numbers are an indicator usually that reflects how far development is, and it's just sad to see people always use these cheap tricks of using version numbers that suggest a certain state of the project it isn't in yet.

And I never talked about the decoder producing those blocks or anything, I'm talking of course about the result that I just get to see presented by the decoder. Just also noted that the decoder is quite annoyingly slow and stuttering considering the the mentioned improvements on speed etc. it boasts about in the presentations.

And what's the purpose on those samples anyways if you let an lousy encoder do some encodes of pirated movies. I somehow doubt they acquired licenses to distribute those samples :P

Well, maybe I try that non OpenAVS. Might be less stuttering, though it seems to be tricky to go past that funny license thing.

[ia-ia]

guys, lets cool down and wait for the specs

Yes, full specs are not available yet, but all coding tools are well defined.

for example see here:
http://ltswww.epfl.ch/ltsftp/VCIP2005/DATA/5960_73.PDF

here:
http://research.microsoft.com/~fengwu/papers/avs_icme_04.pdf

and here:
http://www.itu.int/ITU-T/IPTV/events/072006/docs/ID/FGIPTV-ID-0082e.doc

Basically it's very close to AVC, faster and slightly less efficient.

Regards,
ila.

[leiming2006]

The problem I want to know is how can I make one AVS file?
The AVS Script?(impossible)

[CruNcher]

be carefull with the official test decoder it can give problems with the .avs (avisynth) extension (tries to render it) and hang the system because of a (dshow chain loop overload) when this happens it's hard to get you windows back under controll memory will fill up like crazy (im in no doubt every Windows user allready made that experience) and the (Kernel) System response becomes very slow the perfectly architectured Windoz Memory Management how nice it is :D And if you want to make that expirience just install 200 dshow filters and try to put a theora.ogg into MPC have fun with the overload yeah (inteligent Graph building) you have to love it sorry for all this oftopic but i couldn't resist after years :d

[videomixer9]

I guess you could just name those files .mpg, mpeg splitters seem to read them fine. The inventors obviously had no clue about avisynth or anything, chinese tend to ignore the rest of the world after all and it seems avisynth is quite well known as an encoding helper. Well, filtering seems to be unknown to the AVS guys anyways if you'll see those example rips on their site :P

Btw. no hangsups here also not with theora oggs though I really encountered them only on some few hardcore OSS fansites so far.

[CruNcher]

it also works another way just Drag&Drop an .exe into the mpc window :P the theora.ogg and many other only happen if you have no filter installed that can handle such files and directshow tries to find a filter that can render it and then it can get very problematic with alot of filters installed :d (Dshow Hell). Thats one reason i prefer integrated players like Mplayer, VLC, tcpmp and soon CorePlayer (they never overloaded Windows for me) :)