Skip to content
Images

MOSS-Image-Ani-NaRA

адаптер аниме-стилистики для MOSS-Image

2 min readupdated

Экспериментальный стилевой адаптер от автора RBXLSP для генератора изображений MOSS-Image-Base, который мягко усиливает аниме-эстетику при генерации. В основе лежит архитектура NaRA (Noise-aware Rank Adaptation — разновидность LoRA, где низкоранговое ядро динамически меняется в зависимости от уровня шума на каждом шаге деноизинга). Адаптер воздействует исключительно на слои внимания (attention) и не требует использования специальных триггер-слов в промпте. Подойдёт пользователям ComfyUI, которые хотят придать генерациям MOSS-Image более выраженный иллюстративный вид без сильного искажения базовой композиции.

Ключевые особенности

  • Работает по методу NaRA (с пресетом attn), динамически адаптируя веса под уровень шума через гиперсеть с кодированием Fourier sigma.
  • Не требует триггер-слов — стиль применяется автоматически при подключении адаптера.
  • Применяется только к модулям внимания proj_q/k/v/o с параметрами Rank 16 и Alpha 16.
  • В редких случаях может слегка ухудшать качество генерации текста на изображении.

Что можно сделать

  • Усилить аниме-эстетику и иллюстративную подачу при работе с базовой моделью MOSS-Image-Base или её версией MOSS-Image-SFT.
  • Подключить адаптер в пайплайны ComfyUI с помощью кастомной ноды RBkreb/comfyui-moss-nara для стилизации сцен без переписывания промптов.

Пример

Для воспроизведения результатов из карточки модели используйте чекпоинт Moss-Image-SFT int8, сэмплер euler с планировщиком sgm_uniform, значение CFG: 3.0 и силу применения адаптера strength: 1.0. Триггер-слова добавлять не нужно.

ПодробнееОбучение: 1 000 шагов (AdamW, LR 1e-4, cosine scheduler, warmup 60 шагов, EMA 0.95) на датасете из 237 иллюстраций с разрешением по длинной стороне от 980 до 10704 px (медиана — 3200 px) и комбинированными описаниями (теги + связный текст).
База: OpenMOSS-Team/MOSS-Image-Base (bf16), гиперсеть c_scale 0.1.
Ограничения: совместимость с другими адаптерами и одновременное смешивание нескольких NaRA не тестировались, хотя теоретически поддерживаются.
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 9.5 GB · ready files12 GB
  • FP8 / INT8recommendedweights 4.8 GB6.7 GB
  • 4-bitweights 2.7 GB4.5 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 5060 Ti 8GB · BF16 — from RTX 5070

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

This is a LoRA: counted together with its base model MOSS-Image-Base.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

A ready script for the machine we advise above: paste it and the model downloads and starts.

  1. 1. Rent A30

    $0.26 an hour with the coupon. Pick the Ubuntu image when you create the machine.

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    #!/usr/bin/env bash
    # MOSS-Image-Ani-NaRA
    set -e
    # NVIDIA driver: CUDA 13 needs 580+. If older or missing, the current nvidia-open is installed (keep yours: KEEP_DRIVER=1)
    DRV=$(nvidia-smi --query-gpu=driver_version --format=csv,noheader 2>/dev/null | head -1 | cut -d. -f1)
    case "$DRV" in ""|*[!0-9]*) DRV=0 ;; esac
    if [ "$DRV" -lt 580 ] && [ -z "${KEEP_DRIVER:-}" ]; then
      . /etc/os-release && \
      curl -fsSLO "https://developer.download.nvidia.com/compute/cuda/repos/ubuntu${VERSION_ID//./}/x86_64/cuda-keyring_1.1-1_all.deb" && \
      sudo dpkg -i cuda-keyring_1.1-1_all.deb && rm -f cuda-keyring_1.1-1_all.deb && \
      sudo apt-get update && \
      sudo DEBIAN_FRONTEND=noninteractive apt-get install -y nvidia-open || \
      { echo "Could not install the NVIDIA driver: see the error above. To keep the current one, run with KEEP_DRIVER=1."; exit 1; }
      echo "NVIDIA driver updated. Reboot the machine (sudo reboot) and run this same script again."
      exit 0
    fi
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    # CUDA 13+ → int8_convrot (Comfy-Org's advice), else fp8_scaled
    CU=$(nvidia-smi | grep -oP 'CUDA Version: \K[0-9]+' || echo 12)
    [ "$CU" -ge 13 ] && TORCH=cu130 || TORCH=cu128
    [ -d ~/ComfyUI ] || git clone https://github.com/comfyanonymous/ComfyUI ~/ComfyUI
    cd ~/ComfyUI
    uv venv --python 3.12 --allow-existing && source .venv/bin/activate
    uv pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/$TORCH
    uv pip install -r requirements.txt "huggingface_hub[cli]"
    # Base files: MOSS Image Base, 8-bit
    if [ "$CU" -ge 13 ]; then
      hf download Comfy-Org/MOSS-Image diffusion_models/moss_image_base_int8_convrot.safetensors text_encoders/qwen3.5_9b_int8_convrot.safetensors vae/flux2-vae.safetensors --local-dir models
    else
      hf download Comfy-Org/MOSS-Image diffusion_models/moss_image_base_bf16.safetensors text_encoders/qwen3.5_9b_w4a8_gptq.safetensors vae/flux2-vae.safetensors --local-dir models
    fi
    # LoRAs (1)
    hf download RBXLSP/MOSS-Image-Ani-NaRA moss-ani-1000-nara.safetensors --local-dir models/loras/RBXLSP__MOSS-Image-Ani-NaRA
    # Start ComfyUI. From your computer: ssh -L 8188:localhost:8188 user@IP, then http://localhost:8188
    python main.py --listen 127.0.0.1 --port 8188
    Add more LoRAs to MOSS-Image-Base
  3. 3. Use it

    From your computer: ssh -L 8188:localhost:8188 user@machine-IP, then open http://localhost:8188. Or open that address in the browser on the machine's desktop. The model's workflow is usually in its Hugging Face card.

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • SageAttention 3v2.2.0from October 28, 2025

    even faster on RTX 50 / Blackwell

    FP4 attention for Blackwell. Faster than SageAttention 2, but the authors suggest v2 when accuracy matters.

    How to enable: Build from the sageattention3_blackwell folder of the repo; in ComfyUI via KJNodes.

    Earlier generations (2)
    • SageAttention 2++ — Quantised (INT8/FP8) attention with no visible quality loss. The main speed-up for image and video generation.
    • SageAttention 1 — First version: INT8 for Q·K. Still works, but 2++ is faster and more accurate.
  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (3)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
    • xFormers — The older way to speed up attention, works almost everywhere. Sage or Flash are faster where available.
  • cache-dit (DBCache)v1.5.6from November 20, 2026new release

    1.5–3× at nearly the same quality

    Reuses block outputs between diffusion steps. Supports most DiT models.

    How to enable: pip install cache-dit; one line cache_dit.enable_cache(pipe) for diffusers.

    Earlier generations (2)
    • MagCache — Skips steps by magnitude of change. More accurate than TeaCache, no calibration.
    • TeaCache — The first popular diffusion step cache. Works, newer methods are more accurate.
  • Nunchaku (SVDQuant 4-bit)v1.2.1from January 25, 2026

    ≈3× faster and 3.5× less memory

    4-bit versions of popular image models with almost no quality loss (FP4 on RTX 50).

    How to enable: Ready weights for supported models; in ComfyUI use ComfyUI-nunchaku.

  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

Requires the base model

OpenMOSS-Team/MOSS-Image-Base

This model is a LoRA / adapter. Download the base model first, then load this one on top.

New posts about versions of MOSS-Image-Base (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 12All on Hugging Face
Merges 2All on Hugging Face

About the model

RBXLSP/MOSS-Image-Ani-NaRA
Author
RBXLSP
Category
Images
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments