Skip to content
Images

Qwen-Image-Genitalia-Anatomy-Fix-NSFW-LoRA-v1

единый LoRA для анатомии взрослых в Qwen Image 2.1

2 min read

Это LoRA-адаптер от автора wank3r для Qwen Image 2.1: один файл Genitalia-Anatomy-2.1-v1.safetensors объединяет данные по женской и мужской анатомии взрослых. Обучен на 3 689 изображениях, rank 64; предназначен для text-to-image и редактирования по изображению в совместимом пайплайне Qwen Image 2.1. Не загружайте его в Qwen Image 2512.

Ключевые особенности

  • Один LoRA для Qwen Image 2.1; файл весов Genitalia-Anatomy-2.1-v1.safetensors не совместим с Qwen Image 2512.
  • 3 689 training images из прошлых анатомических работ 2512: западная, китайская, японская и корейская женская анатомия, а также мужская без национальной категории.
  • Основной триггер zl_private_v1 и пять опциональных категорийных токенов; автор рекомендует сначала обычное текстовое описание, токены не гарантируют чёткого разделения.
  • Генерация text-to-image и редактирование с conditioning по изображению; можно комбинировать с другими LoRA для той же архитектуры Qwen Image 2.1.
  • Две reference-реализации инференса, включая экспериментальную low-VRAM; low-VRAM скрипт всё равно требует CUDA GPU.

Пример

Начните с естественного описания взрослого человека, позы, анатомии и сцены. Для более явного контроля добавьте zl_private_v1 и один подходящий категорийный токен, затем сравните результат с той же seed без этого токена.

ПодробнееОбучение: 3 689 изображений, rank 64, batch size 8, 16 000 шагов; это новое обучение на Qwen Image 2.1, а не слияние весов 2512.
Тренировка шла на NVIDIA RTX PRO 6000 Blackwell / 96 GB; low-VRAM скрипт выгружает неактивные компоненты в RAM и может быть медленнее.
Лицензия: только некоммерческое использование; коммерческое использование и распространение — с письменного разрешения автора.
Анатомия не гарантируется для каждой seed и сложных ракурсов; перекрытия, экстремальные позы и необычные углы требуют проверки.
Примеры
Discuss Telegram 1 642

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 14 GB · ready files16 GB
  • FP8 / INT8recommendedweights 6.9 GB9.1 GB
  • 4-bitweights 4.0 GB5.9 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 3080 · BF16 — from RTX 4090

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 Qwen-Image-2.1.

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

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    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.

    # wank3r/Qwen-Image-Genitalia-Anatomy-Fix-NSFW-LoRA-v1
    # ===== Tokens: paste yours between the quotes =====
    export HF_TOKEN=""        # huggingface.co/settings/tokens, for gated models
    # ==========================================
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download Qwen/Qwen-Image-2.1 --local-dir ~/models/Qwen-Image-2.1
    hf download wank3r/Qwen-Image-Genitalia-Anatomy-Fix-NSFW-LoRA-v1 --local-dir ~/models/Qwen-Image-Genitalia-Anatomy-Fix-NSFW-LoRA-v1
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to Qwen-Image-2.1
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

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

Qwen/Qwen-Image-2.1

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

New posts about versions of Qwen-Image-2.1 (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 4All on Hugging Face

About the model

wank3r/Qwen-Image-Genitalia-Anatomy-Fix-NSFW-LoRA-v1
Author
wank3r
Category
Images
Released
Posted

On NeyroLearn

Hugging Face

6421

Ask AI about this model

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

Comments