UPDATE:
v5: New training method
Full changelog below
This lora has been trained with images of weird fashion trends and random futuristic artists renditions of clothing that will never make the shelves. Apply this to any v1.5 SD model.
Use the prompt 'weird future fashion' in combination the other optional trigger words. At weight 1 it does get pretty opinionated about clothing and styles. Go beyond 1 to get TRULY weird!
Testing and sample images generated using the following models:
Custom checkpoint (similar to Comic Babes v1)
If you like this, check out my other models!
Full Changelog
v4: More images in training, more colorful, generally more functional
v3: Better LoRA training, smaller filesize, more stable (heh) and much better at handling prompts
v2: Expanded dataset, better handling non cloths related prompts
v1: Initial model
Description
V2 preserves backgrounds, styles, and general non clothing prompts better compared to V1 (see examples)