Tried to do something a little bit more challenging this time. Generated with fluffyrock_e184VpredE157 on SDNext; anythingKlF8Anime2Vae. Had a lot of trouble with oversaturation and blowout. Fluffyrock clearly has a much better anatomical understanding (and generates high-quality results without artist tags using the quality LoRa) but I haven't quite got the workarounds for vprediction down yet. Some parts of the image - the eyes in particular - I touched up by hand, as the generated ones were hilariously oversaturated and quite bad. FluffyRock is good at low-res generation, but at higher resolutions when upscaling it does not work well. (unlike Yiffymix, which handles them well.)