Minimax H3:
v3 dataset, usually used 1.00 strength. Use it for T2VA or start with an image made from an image model version of DIMStyle!
Anima:
v3 dataset. A bit cleaner and removes watermark issues. I recommend going with 0.50 to 0.75 strength. Going all the way to 1.00 can make things look much more dramatic, but increases distortion probabilities.
v1.1
Expanded the dataset to include more images and themes and whatnot. Less restrictive than before
Original Description:
Dramatic stylized art style most appropriate for fantasy/sci fi generations.
Included a couple comparison images in the examples to show differing values of guidance and weight.
Heavy Raymond Swanland influence.