New Release: <a target="_blank" rel="ugc" href="https://civarchive.com/models/21134/or010">林忆宁，古户绘梨花 (010, Furuki Edoka)</a><h3>X/Y Tests:</h3>Prompt test:<img src="https://imagecache.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/961d59a2-5412-4470-4935-1b6dca67c800/width=525/961d59a2-5412-4470-4935-1b6dca67c800" />CFG test (w/ 010_v32-000016 model):<img src="https://imagecache.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/8869045b-faf1-41df-9be1-3e62def77700/width=525/8869045b-faf1-41df-9be1-3e62def77700" />Generalization test (w/ 010_v32-000016 model):<img src="https://imagecache.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/9e241e34-90da-4194-6788-db365730fa00/width=525/9e241e34-90da-4194-6788-db365730fa00" /><h3>Dataset:</h3>Images chosen from 010's selfie set, specifically during her days as an acitve member of SNH48's Team X. Training:This LoRA model is trained with <a target="_blank" rel="ugc" href="https://github.com/bmaltais/kohya_ss">Kohya_ss</a> GUI. Please check version descriptions for training parameters <h3>Tips:</h3><ol><li>I suggest using this model at 0.6~0.9 weight for best results. Going above 0.9 usually causes distortion in face.</li><li>Low CFG scale around 2.5 to 5.0 is recommended. Turn it down if you are seeing oversaturated images.</li><li>Using this model to generate NSFW content is possible, but highly dissuaded.</li></ol><h3>Reach Out:</h3>七海电报群：<a target="_blank" rel="ugc" href="https://t.me/nanamivr">https://t.me/nanamivr</a>宁宁肉体厨：<a target="_blank" rel="ugc" href="https://t.me/nanami_ai">https://t.me/nanami_ai</a>Any questions, suggestions and ideas on improving the model are appreciated.

林忆宁，塞纳河小偶像 (010, idol girl from SNH48)

ReleaseDataset contains 20 images, 100 steps each, 2,000 steps in total.Images were resized to 512×512. Max resolution of the LoRA model is set to 768,768.Choose this version to generate portraits with 010 style.

Files with SHA256: d3b20da75824783c00a46c6f2abfae34bbb653f14dc195aa6ded2bcbd9d435ed