Verified: 2 years ago
SafeTensor
Type | |
Stats | 680 16,441 14 |
Reviews | (114) |
Published | Aug 16, 2023 |
Base Model | |
Training | Steps: 14,000 Epochs: 2 |
Usage Tips | Clip Skip: 2 |
Trigger Words | akashi kaoru nogami aoi sannomiya shiho minamoto kouichi |
Hash | AutoV2 C6C1805A48 |
i remade old models that have known problems into multicharacter pacs without this problems.
join discord if you have questions, or want to commision LoRAs.
P.S. previews here is no hiresfix.
I'm continue to focus on anime-like looking -> if you want to use artistic style models decrease weight to something like 0.5, my focus is to made screenshot like images with my model with weight 0.7-0.8 without write any additional tags except character TW+dress(or character only if it lazy model like this one). Why this way? Less tokens(tags)->more control on other things like pose/action, or background, and you don't need to copypaste 10 tags of character colors/styles everytime.