The base virtual source imagery for this was generated using a clip guided diffusion neural net model hosted on HuggingFace. That was then vectorized in Studio Artist, rescaled up 4X, then re-rendered with additional Studio Artist effects. The paint action sequence timeline was then time expanded, and both the raster and vector imagery was respectively run through a transition context and keyframe animated.
The textural tags for the multi-modal image synthesis was variations of 'people standing in time square guns'.
Normalize the bezier paths and then distance sort them above. Just use the PASeq 4K rescaled images with a flow transition context below.
No comments:
Post a Comment