Bug report - Bug in MP VMR9? (1 Viewer)

[hoborg]

Looks like there is bug in MP VMR9. Some videos failed to produce any picture or freeze on first frame. This is not a splitter/decoder issue, must be renderer problem. Graph is created OK. Can be reproduced on all OS, just select VMR9 as renderer.

Sample opened in Graphstudio: (working fine)

Sample opened in MP using VMR9: (not working)

BiHeight reported as "-480" ?!? - can this be the problem?

Sample opened in MP using EVR: (working fine)

And here is video sample

 

[tourettes]

Could you test following dshowhelper.dll? I didnt test nor have looked into VMR9 related code in years so no quarantee that it will even work

 

[tourettes]

doesn't seem to help, I guess I need to debug a bit.

 

[tourettes]

All VMR9 internal height variables are positive 480. So, no clue where that comes. Negative values with bitmap height are perfectly valid, those are just showing the picture must be decoded from bottom to top.

 

[hoborg]

Maybe the negative value is not the problem here, maybe it is not related...

 

[tourettes]

Maybe the negative value is not the problem here, maybe it is not related...

Yep, could be... but I have no clue what else might be wrong.

 

[hoborg]

Hm, i just looking for any diference from other samples i have here, and this one is detected as interlaced - no idea if it can de related...

2010-01-18 18:38:26.923497 [Debug][MPMain]: VMR9: SetDeinterlaceMode - Interlaced frame detected

 

[Scythe42]

Just tried the supplied sample. Plays flawless with VMR on my Win7 machine (used MPC and ffdshow as in the example)

 

[tourettes]

Some findings:

1) Windows 7 video decoder -> black video
2) ffdshow and VMR9 + all settings turned on -> works without any issues
3) Windows 7 decoder in GraphEdit -> video is ok, height is -480

Sounds like some incompatibility with VMR9 and certain video decoders. Also -480 is valid.

Bad thing is that I know not a one developer with C++ and XP combination