There are plenty of multimodel models, like Llava, with image input. Or things like BLIP diffusion which will to image+text to image. Pure text-to-video is ver

There are plenty of multimodel models, like Llava, with image input. Or things like BLIP diffusion which will to image+text to image. Pure text-to-video is very hard, and is being worked on.