VLC 3.0 and Chinese AVS+ codec

[apsattv]

I'm trying to play back a .ts recording from Chinasat 6A it appears that VLC 3.0 still doesn't work with CAVS format. The video window opens to the correct size and the audio plays back fine. However there is no video playback.

I also have a modified version of VLC 2.26 which does play avs+ codec
I have it uploaded here
http://apsattv.com/files/VLC-AVS2.26.rar

and a small 500mb recording
http://apsattv.com/files/125.0E_3760.30 ... )_dump.rar

Hopefully this is something the VLC team can fix as VLC is often used as the player for various satellite softwares such as DVB Dream , Prog DVB and others.

I'm happy to supply furthur recordings if needed.

[InTheWings]

What's the spec / difference ?

[apsattv]

The modified 2.26 that I have can play the AVS+ video fine.

The V3.0 doesn't , i don't know what was done to to modify the 2.26 to get it to play them, but it was provided by a Chinese satellite enthusiast possibly they recompiled or added support in some way?

I'm wanting someone to look at the modified 2.26 and hopefully the developers can see what needs to be done. I'm happy to submit any .ts recordings of the channels in any size. To nail down any compatibility issue to get AVS+ to work.

Feel free to download from the links i posted. See what you can discover

[apsattv]

I guess i could do a download of the 2.26 Release then do a file compare with the modified version to see what files were changed?

[InTheWings]

Sorry, but we can't do anything from a binary, and even not from an unknown source code.
If you can contact the author, try to get source code. Just hope he doesn't use/steal the reference decoders.
If you read/speak chinese, try to get AVS spec.

[apsattv]

I asked my Chinese friend and he said he got the modified 2.26 VLC from a forum. I'm not a programmer so i don't really know much more about it other than that. Is it not possible to play the example file I uploaded, using the modified version to see exactly what its actually doing to process the avs+ format?

There is also a thread here which could have usefull information

http://www.dvbviewer.tv/forum/topic/521 ... deo-codec/

[Rémi Denis-Courmont]

Is it not possible to play the example file I uploaded, using the modified version to see exactly what its actually doing to process the avs+ format?

In theory, the code can be reverse engineered. But the number of living humans capable of reverse engineering a video codec can probably be counted with your fingers. I only know one and he is not a VLC developper.

[apsattv]

VLC already supports Chinese AVS both version 2.26 and 3.0 identify it on the sample above. The video windows opens to the correct resolution. Audio plays. Just that the video is black. So i'm not sure why anything needs to be reverse engineered?

please check the link to another thread that I posted which mentions some ffwdshow filters

[apsattv]

I hope someone can at least look into this
with the examples here and request source codes to add this codec support it
http://www.avs.org.cn/english/

It would help VLC add many users in China as many users there use C band satellite tv and tuner cards for television and the services use AVS+ codec

This was just released in the past month. This source code might be of use?

PKU-VCL AVS2 Reference Software xRD1.0 is released.(31-01-2018)

http://www.avs.org.cn/AVS2_download/xRD-1.0.zip.zip

[stefan.gehrer]

I have written the decoder for Chinese AVS in ffmpeg (and by extension in VLC) in 2006. Since then a lot has changed. While that was a "simplified H264", what is in use now in broadcast - whether called AVS2 or AVS+ - is a completely different thing. From hearsay it has more similarity to HEVC but I do not have any specification for it. If someone could find out more about that modified VLC 2.26 I would be interested to hear. Ffmpeg contains support for the external library libxavs2 but that is only for encoding, maybe someone has authored something similar for decoding?

> The video windows opens to the correct resolution. Audio plays. Just that the video is black. So i'm not sure why anything needs to be reverse engineered?

I believe the bitstream between the different AVS versions has just enough compatibility for that. The decoder for the "old" AVS can therefore read the video resolution from the headers but basically fails immediately afterwards - it cannot understand how to decode the content itself. So while hopefully there is an easier approach than reverse engineering - basically impossible - it is not something that just requires a bugfix. Thank you for posting that reference software, I was not aware of it before.