Does MOVA support pure text-to-video generation without reference images?

[kallu87]

Congrats on the release , I wanted to ask if MOVA supports pure text-to-audio video (T2AV) generation without requiring a reference image.

[xiami2019]

Yes. Actually we find that T2AV can be achieved by input a placeholder first frame, like a pure white of a pure black image. The generated video still looks good. We will update relevant code later~.

Thanks for you attention!