Verified: 2 years ago
SafeTensor
Type | |
Stats | 803 |
Reviews | (135) |
Published | Apr 27, 2023 |
Base Model | |
Trigger Words | lum |
Hash | AutoV2 F1E32E03CB |
Please keep it civil and only post SFW images in the reviews!
The LoRA was trained at 768x768. Make sure your image is at least this tall/wide when generating (bigger is fine).
Since the outfits are tagged and not implicit, you will want to add "bikini, strapless, tiger print" to get her default outfit. I also recommend you use EasyNegative with it.
This LoRA was trained on 114 screenshots from the Rumic World special and the Obstacle Course Swim Meet OVA. I figured it would be a nice balance between the new and classic styles. Base model for training was Anythingv3.
V1.2 could use a bit of improvement, I might do a 1.3 (or maybe other characters) if there's interest.