Skip to content
Images

MFAF-Net

симуляция косметических операций на фото без GPU

2 min readupdated

MFAF-Net — исследовательский прототип, который по фотографии и анкете пациента (возраст, пол, тип кожи по Фицпатрику, история процедур) строит приблизительное изображение лица после косметической операции. Он находит 468 точек лица через MediaPipe FaceMesh, добавляет 12-мерный вектор данных пациента и с procedure-aware attention (маской внимания к анатомической зоне операции) геометрически деформирует фото через Delaunay-триангуляцию и аффинное преобразование. Поддерживаются ринопластика, блефаропластика, подтяжка лица, гениопластика, уменьшение лба и подъём бровей; на выходе — изображение, тепловая карта различий и метрики SSIM/PSNR/MAE. В отличие от LandmarkDiff, не требует GPU и Stable Diffusion 1.5 + ControlNet, а работает на CPU через Flask. Это proof-of-concept для исследователей и клинической коммуникации, а не медицинский прибор.

Ключевые особенности

  • Мультимодальный вход: фото + возраст, пол, тип кожи и история лечения.
  • 468 landmarks MediaPipe FaceMesh и геометрические измерения лица.
  • Procedure-aware attention: зона, важная для выбранной операции, получает 3× вес.
  • Полностью CPU-пайплайн на Flask: без diffusion-моделей и GPU.
  • Количественная оценка: SSIM, PSNR, MAE, доля изменённых пикселей и heatmap.

Что можно сделать

  • Показать, как примерно может выглядеть нос после ринопластики или веки после блефаропластики, на учебных примерах.
  • Собрать Flask-демо для исследовательской группы: загрузка фото, выбор процедуры и отчёт с метриками.
  • Прогнать набор before/after из датасета и сравнить геометрические симуляции по SSIM/PSNR.
  • Использовать как лёгкий CPU-baseline рядом с LandmarkDiff, когда нет GPU.

Пример

pip install -r requirements.txt
python setup_dirs.py
python run.py
# затем открыть http://localhost:5000
Подробнее6 процедур: ринопластика, блефаропластика, ритидэктомия, гениопластика, операция на лбу и подъём бровей.
468 landmarks, 12-мерный вектор пациента, procedure-aware attention с весом 3×.
Метрики: SSIM, PSNR, MAE, Changed Pixel %; запуск через Flask на CPU.
Ограничение: академический proof-of-concept, не сертифицированное медицинское изделие; симуляция — геометрическая аппроксимация.
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.0 GB · ready files1.5 GB
  • FP8 / INT8recommendedweights 0.0 GB1.5 GB
  • 4-bitweights 0.0 GB1.5 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 2060 · BF16 — from RTX 2060

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

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.

    # haresh2792005/MFAF-Net
    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 haresh2792005/MFAF-Net --local-dir ~/models/MFAF-Net
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to MFAF-Net
  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

New posts about versions of MFAF-Net (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

haresh2792005/MFAF-Net
Author
haresh2792005
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