Verified: 2 years ago
SafeTensor
Type | |
Stats | 537 137 |
Reviews | (104) |
Published | May 6, 2023 |
Base Model | |
Training | Steps: 3,780 Epochs: 20 |
Trigger Words | gudaguda chibi o_o |
Hash | AutoV2 913DE17293 |
Lora was trained on official comic on https://fate-go.us/gudao/ but with removed speech bubbles. Then was hi-res with realesrgan-ncnn-vulkan
Lora power around 0.8-1
Trigger words are gudaguda, chibi, o_o
o_o is optional but really recommended
Picture res set around 512x512 more higher = more errors. I was using hi-res fix all the time
Used LORAs
https://civitai.com/models/52314/jakuzure-nonon-all-outfits-or-kill-la-kill-klk
https://civitai.com/models/23070/zoe-league-of-legends
https://civitai.com/models/6363/milim-nava-or-character-lora-235
https://civitai.com/models/40640/fumo-like-anime-style-lora
https://civitai.com/models/53552/astolfo-fate-series