CivArchive
    DaSiWa Wan2.1 I2V Workflows - FastFidelity C-AiO 2.0

    My ComfyUI Workflow for using Wan 2.1 i2v 14B.

    Versions

    ๐Ÿ‘‰ Please read the file descriptions "About this version" for more infos.

    Features

    • ๐Ÿ–ผ๏ธ I2V and FLF2V

    • ๐Ÿงฉ Automatic aspectโ€‘ratio calculation and fitting

    • โœจ Ultraโ€‘fast, colorโ€‘correct upscaling (torchlanc)

    • ๐Ÿค Video resolution matching - Fully automatic scaling and resolution calculations

    • ๐Ÿซฅ Watermark support

    • ๐Ÿงฎ Color match feature

    • ๐Ÿ‘พ MiniMeme feature - Create small gif's

    • ๐ŸŽญ NAG - Negative prompting with CFG1)

    • ๐Ÿช„ Interpolation feature

    • ๐ŸŒ€ Seamless loops

    • ๐Ÿƒ Last Frame Extraction

    • ๐Ÿ”– Bookmark-Shortcuts - with number keys


    Version 1.2 has no dependencies.


    YOU are responsible for outputs as always! If you make ToS violating content and I get aware I WILL report this.

    Description

    Requirements

    • https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite

    • https://github.com/rgthree/rgthree-comfy

    • https://github.com/Artificial-Sweetener/comfyui-WhiteRabbit packaging, torchlanc

    • https://github.com/kijai/ComfyUI-KJNodes

    • https://github.com/yolain/ComfyUI-Easy-Use

    • https://github.com/Fannovel16/ComfyUI-Frame-Interpolation

    • https://github.com/pythongosssss/ComfyUI-Custom-Scripts

    Features

    • ๐Ÿ–ผ๏ธ I2V and FLF2V

    • ๐Ÿงฉ Automatic aspectโ€‘ratio calculation and fitting

    • โœจ Ultraโ€‘fast, colorโ€‘correct upscaling (torchlanc)

    • ๐Ÿค Video resolution matching - Fully automatic scaling and resolution calculations

    • ๐Ÿซฅ Watermark support

    • ๐Ÿงฎ Color match feature

    • ๐Ÿ‘พ MiniMeme feature - Create small gif's

    • ๐ŸŽญ NAG - Negative prompting with CFG1)

    • ๐Ÿช„ Interpolation feature

    • ๐ŸŒ€ Seamless loops

    • ๐Ÿƒ Last Frame Extraction

    • ๐Ÿ”– Bookmark-Shortcuts - with number keys

    Changes

    • Complete Overhaul

    • Resolution + Sampling controls

    • WebM AV1 default

    Fixes

    • none

    FAQ

    Comments (4)

    keraloNov 14, 2025
    CivitAI

    thanks for the WF!
    I have a question though. There is only a high LORA loader ist that normal? The low LORA loader is not needed?

    darksidewalker
    Author
    Nov 14, 2025ยท 2 reactions

    Its WAN 2.1 it only has high checkpoint. It is not MoE like WAN 2.2

    claypress98363Mar 18, 2026
    CivitAI

    How would you slow down the speed of the video?

    fkrone646Mar 19, 2026ยท 2 reactions
    CivitAI

    I fixed the last for workflow for myself. hope this helps.

    Here's a clean forum-ready summary of all fixes made to the workflow DaSiWa WAN 2.2 i2v FastFidelity C-AiO-65:

    ---

    ComfyUI Subgraph Workflow Fix Log โ€” DaSiWa WAN 2.2 i2v FastFidelity C-AiO-65

    ComfyUI version: 0.17.2 | Frontend: v1.41.x | GPU: RTX 5090

    ---

    Error 1: Required input is missing: images (VHS_VideoCombine, node 28)

    The workflow validated but immediately failed โ€” VHS_VideoCombine reported its mandatory images input was unlinked

    despite the canvas showing a connection.

    Root cause: A ComfyUI subgraph bug where nested subgraph instances get their inputs arrays doubled (every slot

    duplicated) with all link values set to null when subgraph definitions are updated. The links table correctly recorded

    all connections, but the instance nodes had the wrong data. Additionally, the post-processing chain

    (BatchWatermarkSingle โ†’ Color match โ†’ Perfect loop โ†’ Upscaler โ†’ Interpolation) had all inner nodes in mute mode (4)

    instead of bypass mode (2), creating dead ends with no pass-through path.

    Fix 1 โ€” De-duplicate subgraph instance inputs (4 nodes):

    Four subgraph instances inside the Checkpoint settings SG had the doubled/null inputs bug. For each, removed the

    duplicate input slot and restored the correct link ID by cross-referencing the inner links table:

    - Node 1814 (Interpolation SG): 6 inputs โ†’ 3 inputs, restored links 3913/3914/3915

    - Node 1729 (Upscaler SG): 4 inputs โ†’ 2 inputs, restored link 3788

    - Node 1731 (Perfect loop SG): 2 inputs โ†’ 1 input, restored link 3787

    - Node 1730 (Color match SG): 4 inputs โ†’ 2 inputs, restored links 3790/3878

    Fix 2 โ€” Change BatchWatermarkSingle nodes from mute to bypass:

    Nodes 1726 and 1727 had mode: 4 (mute = no output). Changed to mode: 2 (bypass = pass first input through).

    Fix 3 โ€” Redirect IMAGE output path around broken post-processing chain:

    Inner link 3794 was pointing to muted node 1727 (dead end). Redirected it to the active Context Switch node 1589 at

    slot 7 (IMAGE output), bypassing the entire non-functional post-processing chain.

    Fix 4 โ€” Sever broken post-processing chain links:

    Removed inner links 3790, 3878, 3787, 3788, 3915 which were pulling the Color match โ†’ Perfect loop โ†’ Upscaler โ†’

    Interpolation chain into execution even though all their inner nodes were muted with no bypass path.

    ---

    Error 2: TypeError: GetImageSize.execute() missing 1 required positional argument: 'image'

    After resolving Error 1, the workflow would start but crash immediately during execution with a Python TypeError.

    Root cause: Same doubled/null inputs bug as Error 1, but on a different node โ€” node 1588 (Resolution Logic subgraph

    instance) inside the Checkpoint settings SG. Its image input slot was duplicated (two entries, both link: null), even

    though inner link 3464 in the links table correctly recorded the connection from the Checkpoint SG's first-frame IMAGE

    input (origin_id: -10, slot 0). Because node 1588's instance had no live image connection, the Resolution Logic SG's

    internal GetImageSize node was being called with no image argument, causing a Python TypeError.

    Fix 5 โ€” De-duplicate node 1588 image input and restore link 3464:

    Reduced node 1588's inputs array from 2 entries (both null) to 1 entry with link: 3464. This restores the first-frame

    IMAGE flowing into the Resolution Logic SG so GetImageSize can compute width/height for the sampler.

    ---

    Summary table

    โ”Œโ”€โ”€โ”€โ”€โ”€โ”ฌโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ฌโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”

    โ”‚ Fix โ”‚ Node(s) โ”‚ What changed โ”‚

    โ”œโ”€โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ค

    โ”‚ 1 โ”‚ 1814, 1729, 1731, 1730 โ”‚ De-duplicated inputs arrays, restored link IDs โ”‚

    โ”œโ”€โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ค

    โ”‚ 2 โ”‚ 1726, 1727 โ”‚ mode: 4 โ†’ mode: 2 (mute โ†’ bypass) โ”‚

    โ”œโ”€โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ค

    โ”‚ 3 โ”‚ Link 3794 (inner) โ”‚ Redirected from dead-end muted node โ†’ Context Switch 1589 slot 7 โ”‚

    โ”œโ”€โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ค

    โ”‚ 4 โ”‚ Links 3790/3878/3787/3788/3915 โ”‚ Removed โ€” severed broken post-processing chain โ”‚

    โ”œโ”€โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ค

    โ”‚ 5 โ”‚ 1588 โ”‚ De-duplicated inputs, restored link: 3464 for IMAGE input โ”‚

    โ””โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”˜

    Fixed file: DaSiWa WAN 2.2 i2v FastFidelity C-AiO-65v2.json

    All fixes were made by directly editing the workflow JSON โ€” the ComfyUI canvas showed correct connections but the

    underlying instance data was corrupted by the subgraph update bug. Cross-referencing the links array against the

    inputs arrays of each subgraph instance is the key diagnostic step.

    Workflows
    Wan Video 14B i2v 720p

    Details

    Downloads
    1,471
    Platform
    CivitAI
    Platform Status
    Available
    Created
    11/13/2025
    Updated
    5/14/2026
    Deleted
    -

    Files

    dasiwaWan21I2VAnd_fastfidelityComfyflf1.zip

    dasiwaWan21I2V_fastfidelityCAio20.zip

    Mirrors