>>16235
If you do video gen on ComfyUI, let me give you a word of advice - unload the video and clear the execution cache after you finish a video gen. After a video gen, the model tends to stay loaded, along with some reserved RAM and VRAM. Depending on the next gen’s parameters, you can end up hitting an Out Of Memory error. Unloading and clearing to free RAM is useful to help ensure your video gen doesn’t die mid-gen. (This isn’t as much of a problem if you’re running ComfyUI on really high hardware, like H100 or RTX 6000 cards on RunPod, but cloud-genning isn’t really necessary for most day-to-day uses.)