Linux 20.04
MSDK -> 20.5.1
This is my VPP.In ->
This is my VPP.Out ->
When performing a Query I get -> MFX_WRN_INCOMPATIBLE_VIDEO_PARAM
If I process i get no error in my 'RunFrameVPPAsync' however in the 'SyncOperation' i get ->
If I hardcode NV12 as my input format instead of P010 the warning and error goes away.
The documentation states conversion from P010 to NV12 is supported.
Thanks for helping.
/Anders.
Link Copied
My issue was related to buffer management and I found the problem.
Please ignore this 'issue'.
/A
Setting the parameters
lVPPParams.vpp.In.BitDepthChroma = 10; //P010
lVPPParams.vpp.In.BitDepthLuma = 10; //P010
lVPPParams.vpp.In.Shift = 1; //P010
Got rid of the warning in the query but not the -> MFX_ERR_UNDEFINED_BEHAVIOR
/Anders
My issue was related to buffer management and I found the problem.
Please ignore this 'issue'.
/A
Hi,
Thanks for reaching out to us.
Glad to know that your issue is resolved. Could you please let us know whether you need further help or can we close this thread?
If you have any feedback please share with us.
Thanks.
Hi,
Could you please give us an update?
Thanks.
Sure..
It's the way Intels MSDK want's us users to define the pixel buffers that to me is inconsistent and I made a mistake because of that. The P010 format explicitly wants us users to define the bit-depth where as for NV12 there is no need. So when I defined P010 and bit-depth 10 then I got it working.
Since you ask.. And I got the chance to comment it would be great if Intel simplified the pixel buffer definition for example NV12 is a 4:2:0 8-bit format by the fourCC definition and P010 is a 4:2:0 10-bit format.
What's the reason for having users specify FourCC + ChromaFormat + BitDepthChroma + BitDepthLuma when the only thing needed is FourCC? To me it ads unnecessary complexity and room for errors.
There might be other reasons that to me is unknown however just wanted to mention this.
/Anders
Hi,
Thanks for the valuable feedback.
We have informed the concerned team regarding this.
Could you please let us know whether we can close this thread?
Thanks.
Athira,
Yes, please close this thread.
/Anders
Hi,
We are listening customer’s comment on memory management. We have forwarded your request to dev team and keep you updated.
Thanks.
For more complete information about compiler optimizations, see our Optimization Notice.