Quoted By:
So for anyone interested:
the process used to get these is:
prompt =
"Feline facial expression which can be interpreted in many ways depending on context in the style of {art_history_style_name} Art"
The first frame seed being the template image above
the prompt is run through my custom fork of VQGAN called LucidDreamer for 65 cycles
Next, the 64th cycle is then loaded to Dall-E where I erase portions of the image and re-prompt the AI with the original prompt and either the same, different, or additional style.
