CivArchive
    Wan2.2 Fun Inpaint Video (GGUF) 4 Steps - v1.0

    🖌️ Wan2.2 14B Fun Inpaint Video - GGUF Workflow for Video Editing

    Seamlessly edit and transform videos with AI-powered inpainting using the efficient Wan2.2 GGUF models!

    ✨ Key Features:

    Video Inpainting: Edit specific areas while preserving context

    Start/End Frame Control: Define exact transformation points

    GGUF Memory Efficiency: Q8_0 quantization for lower VRAM usage

    Dual-Model System: High/Low noise for professional results

    4-Step LoRA Speed: LightX2V distilled LoRAs for rapid generation

    Flexible Resolution: 480x480 default, easily adjustable

    🔧 Technical Specs:

    • Model: Wan2.2-Fun-A14B-InP (High/Low Noise variants)

    • Quantization: Q8_0 GGUF format

    • LoRA: LightX2V 4-step distilled (High: 5.6, Low: 2.0 strength)

    • Frames: 81 frames at 16fps

    • Sampling: 6 steps total (4 high + 2 low noise)

    💡 Perfect For:

    • Object replacement in videos

    • Background changes

    • Character transformations

    • Seamless video editing

    • Creative video modifications

    📋 What You Need:

    • Start image (what to change from)

    • End image (what to change to)

    • Wan2.2 GGUF models

    • ComfyUI setup

    Experience the future of video editing with AI precision and creative control!

    #Wan22 #VideoInpainting #GGUF #ComfyUI #VideoEditing #AI

    Description

    FAQ

    Comments (5)

    frosty639Feb 12, 2026· 1 reaction
    CivitAI

    What You Need:

    • Start image (what to change from)

    • End image (what to change to)

    Hang on, you need to supply the end result image? So what is this workflow doing?

    zardozai
    Author
    Feb 12, 2026

    Wait until you find out how a GPS works. It actually asks for where you are and where you want to go. Total scam, right? Why won't it just drive me to a surprise location?

    frosty639Feb 12, 2026

    @zardozai I'm just asking what this workflow does. Since it's inpainting I thought you would mask an area and wan would fill it based on your prompt, but if you are supplying start and end image- by that do you mean it's v2v and you have to supply the first frame or...I just don't understand the usage.

    davemullen68699Apr 9, 2026

    @zardozai So you mislabel a workflow that is FLFV and call it "inpainting" when it's nothing of the sort, and then get snarky when someone asks a genuine question? How does it feel waking up every day and knowing you're still an asshole.

    zardozai
    Author
    Apr 21, 2026

    @davemullen68699 The amount of cope is astonishing.

    Workflows
    Wan Video 2.2 I2V-A14B

    Details

    Downloads
    575
    Platform
    CivitAI
    Platform Status
    Available
    Created
    9/10/2025
    Updated
    6/27/2026
    Deleted
    -

    Files

    wan22FunInpaintVideo_v10.zip

    Mirrors

    CivitAI (1 mirrors)