LoRa training testWork best on 0.3-0.5 strength on model with its own style (Anything, Aom, etc.)
Version 1
Arthur_-_Granblue.safetensors
Same model published on other platforms. May have additional downloads or version variants.