Verified: a year ago
SafeTensor
Type | |
Stats | 172 236 |
Reviews | (21) |
Published | Nov 26, 2023 |
Base Model | |
Training | Steps: 1,080 |
Usage Tips | Clip Skip: 1 Strength: 1 |
Trigger Words | art by JM |
Hash | AutoV2 C8A360C3BE |
Art style LORA for the Chinese artist JM who unfortunately doesn't have an online presence anymore. He has a sort of painterly style and mainly drew pinups of women with big racks. I trained this as a Dreambooth model for a friend using EveryDream2trainer - 20 images trained into the SD 1.5 checkpoint and then extracted into a LORA. There's some quality loss but I suppose that's expected. Very rough, prone to nightmare fuel, but my friend was happy with it. It might work with checkpoints other than base SD 1.5, but I haven't tested.