This version was retrained from phase 2 using multires noise (instead or normal noise), fixing tails (all I could till a degree with animal bias), making it way more flexible for outfits but at the cost of "maybe" at some seeds, losing original style. It's a give and take and this was the best solution (you guys will tell me if works better or not).
Avoid cat word and use instead c4ttitude in this version. Portrait will enhance your outputs at most generations. If you want to combine LORA's, enable composable LORA (at some will be need) and try the weight in total to sum 1 for easier manipulations and better results. Tested in Dreamshaper v3.32 Clipfix and Colorful v3
This version was retrained from phase 2 using multires noise (instead or normal noise), fixing tails (all I could till a degree with animal bias), making it way more flexible for outfits but at the cost of "maybe" at some seeds, losing original style. It's a give and take and this was the best solution (you guys will tell me if works better or not).
Avoid cat word and use instead c4ttitude in this version. Portrait will enhance your outputs at most generations. If you want to combine LORA's, enable composable LORA (at some will be need) and try the weight in total to sum 1 for easier manipulations and better results.
Tested in Dreamshaper v3.32 Clipfix and Colorful v3