calcuis

/

ltxv-gguf

like 16

I managed to do it myself in the end. I don't know if it's a good workflow, but it does work for me.

This is using Florence2, but I didn't have much luck.

good try; you could refer to the post; or drag the output clip/video file inside the browser to get the workflow if any; which is similar to the controlnet usage in image-to-image/image-text-to-image model(s); many examples are available