文件清单 线路:镜像(hf-mirror.com) · 切换
下载后放进对应的 ComfyUI 目录。models/ 指 ComfyUI 的模型根目录。
说明
MeMaMoss Flat Anime Style - Moss
作者:Fish788
Version 1+2 were trained on Moss Image Base. Version 3 was trained on Moss Image SFT. Example images were generated with Moss Image SFT. These loras were trained on images from Megami Magazine from late 2018 to 2026. Moss already knows the Megami Magazine tag but it’s riddled with japanese text. These loras were trained with the text manually removed and the data curated to present a specific style or look. To get the results you see in the images above, these are the settings I used. Visit My Site For Models I Can't Post On CivitAI: https://qtgurl.com/ The loras for Moss V2+3 was trained at 2048 resolution while Moss V1 was trained at 1792 resolution. So they’re capable of generating at around 2048×2048 to 1792×1792 or lower. (I use 1600×1984) MeMaMoss – V1 + V2 + V3 Trained on 68 new and old images. Dataset was re-collected and edited for the best possible quality. About MeMaMoss V3 Through further testing, I discovered a great deal of my generation problems resulted from a incorrect choice in Skip Layer Guidance. I was using 22, 24, 26 when I should have been using just 22, 26. Furthermore, testing has revealed to me that Moss SFT gives better results overall compared to base. So I decided to train a new lora on SFT. This lora unquestionably produces the most accurate results for the Megami Magazine style I've ever seen. The coloring, the line work, the eyes and the backgrounds are all consistent with the source material it was trained on. I haven't tested this Lora on any other model. I recommend you just stick with Moss SFT. Advice for Prompting Moss When using this lora, use only this in the positive: high quality And these in the negative: best quality, normal quality, low quality, worst quality, photorealistic, realistic, 3d, film grain, bad anatomy, watermark, jpeg artifact, bloom, blurry, blurry background, depth of field, extra legs, too many fingers, too many toes, thick lineart, painterly If you're using a lora trained on Base with Moss SFT, the influence from the model can be very strong and if you use certain keywords like Masterpiece, it will over power the lora. I hope you enjoy using this lora as much as I am. 2d style for life!
