Still working with the same source video of the female singer in front of a band for this SD video processing experiment. I took the optimal settings for my previous post's last example (art walk stylized singer) and changed the non style part of the text prompts to orating politician instead.
I then took the exact same text prompt keyframe and got rid of the video input, doing a straight text prompt embedding interpolation instead. So we removed the spatial anchor modulation influence of the video input on the generative image synthesis.
No comments:
Post a Comment