There's an example code here that uses SDL_UpdateYUVTexture to render video frames generated by ffmpeg. But there's no example doing the same using SDL_GPU.
I tried GPU_UpdateImageBytes. Although it generally works (having an issue where video becomes grayscale), the performance feels slower than using SDL_UpdateYUVTexture. I got frame skips and sound stopped playing after a while for a 720p video. Same video plays fine on my platform with SDL_UpdateYUVTexture.
One thing I noticed is that, with SDL_CreateTexture, I can specify a STREAMING flag. But there's no place to do that with GPU_CreateImage. Could that be the reason for the difference in performance?
Appreciate any help on this. Basically I want a comparable example as done in the test code but using sdl_gpu.
There's an example code here that uses
SDL_UpdateYUVTextureto render video frames generated by ffmpeg. But there's no example doing the same using SDL_GPU.I tried
GPU_UpdateImageBytes. Although it generally works (having an issue where video becomes grayscale), the performance feels slower than usingSDL_UpdateYUVTexture. I got frame skips and sound stopped playing after a while for a 720p video. Same video plays fine on my platform withSDL_UpdateYUVTexture.One thing I noticed is that, with
SDL_CreateTexture, I can specify aSTREAMINGflag. But there's no place to do that withGPU_CreateImage. Could that be the reason for the difference in performance?Appreciate any help on this. Basically I want a comparable example as done in the test code but using sdl_gpu.