Flux version -
is nothing more than a quick experiment to see how well it can handle different morphs of ball python, trained on a subset of 1000 images (civitai trainer limit)
Suggested weight is 0.3 - 0.4
One common morph (Pastel, Spider, etc) generally works fine, but mixing different morphs or a confusing term may not be ideal.
SDXL -
Trained with 4824 images of different ball python morphs. // Pastel, Spider, Pinstripe, Mojave, Lesser, Albino, Piebald, Axanthic, Ghost, Enchi, Champagne, etc
Sample prompt: ball python, {morph}
Mixing morphs is possible, ball python, {morph1}, {morph2}, etc
Description
FAQ
Comments (1)
props to tagging so comprehensively, out of pure curiosity I wonder if your tagged dataset can be somehow used to identify morphs from an input image?


