@sherington: Understood and thanks - I know that's often a good solution but in my case, there are many layers of code between where the video is decoded and the buffer used as a texture and a lot of assumptions about how it works. I could always decode video into a fixed buffer size and then scale ...