Skip to content
Video

SadikAI Video Wan 2.2 I2V Fast

видео из картинки за 4 шага

2 min readupdated

Это готовая сборка для ComfyUI на базе Wan-AI/Wan2.2-I2V-A14B — 14B-модели, которая превращает одно изображение и текстовый промпт в короткий ролик. Ускорение дают 4-шаговые LoRA от LightX2V: вместо обычных 20+ шагов сэмплирования хватает четырёх. В комплекте лежат fp8-модели high/low noise, текст-энкодер, VAE и готовый workflow, так что не нужно собирать пайплайн вручную. Важно: это не файнтюн SadikAI, а перепаковка официальных весов Apache-2.0; собственный LoRA обещают в будущих версиях.

Ключевые особенности

  • Всего 4 шага генерации благодаря LightX2V LoRA для high- и low-noise этапов.
  • Комплект «под ключ» для ComfyUI: diffusion-модели, LoRA, текст-энкодер, VAE и JSON-workflow.
  • I2V-режим: изображение + промпт, 81 кадр (~5 секунд при 16 fps), разрешения 640×640, 832×480 или 1280×720.
  • High-noise модель делает первые 2 шага, low-noise — последние 2; в workflow это уже связано через два KSamplerAdvanced.
  • Честная перепаковка официальных файлов Wan-AI, LightX2V и Comfy-Org без заявки на новый файнтюн.

Что можно сделать

  • Оживить персонажа или фотографию: загрузить картинку, описать движение и движение камеры, получить клип на 81 кадр.
  • Сделать короткие видео для соцсетей: выбрать 640×640 или 832×480 и перебирать промпты про поворот головы, наезд камеры, ветер в волосах.
  • Взять готовый workflow за основу в ComfyUI и экспериментировать с промптами и настройками без ручной сборки нод.
  • Проверить I2V на RTX 4090 с 24 ГБ VRAM или в облаке, если на компьютере нет NVIDIA-карты.

Пример

hf download SadikNawazAI/sadik-ai-video-wan2.2-fast --local-dir SadikAI-Video-Fast
ПодробнееОбщий размер файлов — ~35.4 GiB. Рекомендуется 24 ГБ VRAM, комфортно — 32 ГБ+. Официальный бенчмарк ComfyUI: ~97 секунд для клипа 640×640 на RTX 4090 24 ГБ с fp8_scaled и 4-шаговыми LoRA.
Настройки: 4 шага, CFG 1.0, sampler euler, scheduler simple, shift 5.0, сила обеих LoRA 1.0. CPU-only не поддерживается.
Лицензия Apache-2.0, как у базовой модели; при распространении файлов нужно сохранять атрибуцию.
Discuss Telegram 0

Will it run?

Runs in 4-bit

Full precision does not fit, a quantised version does: quality barely changes down to 8 bits, more noticeably at 4 bits.

  • FP8recommendedweights 20 GB · ready files26 GB
  • 4-bitweights 11 GB16 GB

Tip: go 8-bit.FP8 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 — from RTX 5090

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.

    # SadikNawazAI/sadik-ai-video-wan2.2-fast
    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 SadikNawazAI/sadik-ai-video-wan2.2-fast --local-dir ~/models/sadik-ai-video-wan2.2-fast
    # The model is in ~/models. How to run it is in the model card (link above).
  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.
  • LightX2V step distill0.6.1from November 9, 2026new release

    5–10× fewer steps

    Distilled LoRAs and an engine: video in 4–8 steps instead of 30–50.

    How to enable: Load their step-distill LoRA, CFG 1.0, 4–8 steps.

  • 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 Wan2.2-I2V-A14B (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 9All on Hugging Face
LoRAs and adapters 12All on Hugging Face

About the model

SadikNawazAI/sadik-ai-video-wan2.2-fast
Author
SadikNawazAI
Category
Video
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