Latest (2026-11): the fine-tuned full model FFF_MOSS-Alba v1 (https://civarchive.com/models/3034445/fffmoss-alba) and this page's v2 (the same style as a LoRA, for other MOSS-Image Base-family checkpoints). v1 below is the earlier LoRA.
最新(2026-11)は、fine-tune したモデル版の FFF_MOSS-Alba v1 と、このページの v2(同じ画風の LoRA、ほかの MOSS-Image Base 系 checkpoint 向け)です。以下の v1 は以前の LoRA です。
A style LoRA for MOSS-Image that draws people in the style of FFF_ANIMA-Alba. Only the people were trained (backgrounds were masked out), so the scenery keeps MOSS-Image's detail.
人物を FFF_ANIMA-Alba の画風で描くようにする MOSS-Image 用の画風 LoRA です。学習したのは人物だけ(背景はマスクで除外)なので、背景は MOSS-Image の描き込みを生かしています。
Full card on Hugging Face (prompt, negative prompt, settings, samples, license): https://huggingface.co/flyfront/FFF_MOSS-Alba
詳しい使い方(プロンプト・ネガティブ・設定・生成例・ライセンス)は Hugging Face のページにあります。
### Quick start
- MOSS-Image SFT (recommended; Base also works) + <lora:FFF_MOSS-Alba-style-v1:1>. Keep the weight at 1.0.
- Trigger @fyfnt at the head of the prompt (recommended; optional, with it the eyes are a little larger and closer to the teacher).
- Euler, 25-30 steps, MOSS or Simple scheduler (shift 5). About 2.4M pixels in any aspect ratio (1344x1792, 1792x1344, 1536x1536); down to 1M pixels also works. CFG 3 (recommended) for soft shading, 5-6 for a cel-shaded look.
- Negative prompt: the usual MOSS-Image one (full list on the Hugging Face card) plus blush, embarrassed unless you want a blush (SFT blushes by default), and overexposure to keep bright daylight scenes from washing out.
- Works with the 8-step / 4-step / 2-step distilled LoRAs; at CFG 1 use NegPiP negative weights in the prompt instead of the negative prompt.
- No quality tags or background tags needed.
### About
Trained on MOSS-Image Base with the loss masked to the people only (background weight 0), from 1,457 images generated with FFF_ANIMA-Alba. The style transfers to SFT as well. Hands can come out simplified, as with other style LoRAs on MOSS-Image; a hand fixer helps.
MOSS-Image Base を起点に、人物だけに loss を掛けて(背景の重み 0)、FFF_ANIMA-Alba で生成した 1,457 枚から学習しました。SFT に当てても画風は乗ります。指が簡略になることがあるので、手の補正と合わせると安定します。
### License
Apache License 2.0, the same as the MOSS-Image weights. Generated images are yours.
MOSS-Image の重みと同じ Apache License 2.0 です。生成した画像は利用者のものです。