Mikhail_AMD

I used FFmpeg 6.1.1 full release build from here: https://www.gyan.dev/ffmpeg/builds/ My command line was the same as the original one. AMF in this command line will get D3D11 device from...

There are two separate issues here: 1. PA fails with D3D11VA decoding enabled. Reason: decoder output texture allocation needs to add D3D11_BIND_SHADER_RESOURCE flag to be able to run PA shaders....

Command line parameters for flags needed to mix D3D11VA decoder and AMF PA will be submitted soon. The patches with AMF HW decoder, context and video processor merged into FFmpeg...

OK, I added code to skip corrupted SEIs. But cannot blindly skip parsing errors without testing. Thanks.

From HW perspective not much difference but differences in driver and firmware can produce somewhat different results.

This is known issue and caused by multithreaded tracing. The fix will come in a driver release. A a way-around you may minimize AMF tracing for particular AMF runtime version...

The decoder would keep allocation and call SetCrop on the output surface to indicate decoded size. Note, that the same happens when H264 or HEVC stream resolution is not aligned...

> I expect the output texture size will changes from {w1,h1} to {w2,h2} after 'reinit' called. The expectation is incorrect. The whole point of ReInit is to avoid texture reallocation...

To save some time could you please share clips from RX 9070 and RTX: full and nominal range?

Several thoughts: 1. In OBS range parameter is processed by OBS color converter. It does not depend on encoder. 2. All AMF parameters you mentioned (except range) are used only...