Hi, i’m currently trying make an improvement of this sample for my project :
Since DXGI captures in a BGRA format and NvEnc can encode this format, i got rid of the color conversion.
But a thing is bothering me,
I was thinking why should we copy a texture we already have, so in order to optimize it, instead of doing
DXGI capture Texture → Copy into the texture that NvEnc will encode → Encode, why should I not do this :
So I did and called NvEncoder’s GetNextInputFrame() right before capture with DXGI, then immediatly encode it, but the video produced is completely black.
Is DXGI setting a TextureDesc incompatible with the encoding ?
Is there 0 way to get rid of the copy ?