Skip to content
WorkflowVerifiedExternal
Download

Wan2.2_I2V_infinity_video_part_by_part(PPP)_max_consistency_FaceReactor_1.0 — v1.4

V1.4 is still not perfect, but it works quite well. Since I am currently working on another project, I am making my current version available to you here. It is not quite finished yet, but it produces good results.

Sourcecivitai.com/models/2396602
TaskMultiCapabilitiesImage edit · Image to image · Text to imageFamilyWanVersionRelease 2936071 — wan22I2VInfinityVideoPartBy_v14.jsonEstimated VRAM16 GiBUnratedgrade

Runtime profile

Access
External workflow
Status
Verified
Version
wan22I2VInfinityVideoPartBy_v14.json
Source
Civitai
Updated
Nov 3, 2026

Source description

V1.4 is still not perfect, but it works quite well. Since I am currently working on another project, I am making my current version available to you here. It is not quite finished yet, but it produces good results.

Info: The best results when extending videos are obtained by assigning the same seed to all extensions as the original video.

Info: If ComfyUI cannot find the saved latent, you need to adjust the path in the "Latent Save" group.

  • fix scale for Video extention

  • fix upscale for Endframe in Landscape

  • fix bug color correktion without video extention

  • some minior fixes and improvements

  • removed colorcorrect after decode (not needed, Color Match is much better)

  • some little changes in settings

  • Videos with more than 81 frames can now be loaded for expansion.

  • Color Match sets color now after Decode too, for maximum results.

  • add Latent Save after video generation

  • add Load latent for video extension with better transitions and prevents plastic look

  • add Load and Save with Name

  • add automatik save to input folder

  • add smooth video extention via prompt shifting

  • add face detailer for testing

  • add latent noise between sampler for better details and textures.

  • add Fast mode switch (Generate 41 frames instead of 81 and interpolate them.)

  • experimental Face detailer (can be enabled if needed)

(Due to time constraints, the documentation was not updated.)

The example video was generated using an RTX 4070 Ti SUPER (16GB VRAM): 10 sec video in ~11min Normal mode or ~6min in Fast mode

I used a photo of my couch, along with an image of a woman generated by Qwen 2512, for this test.

What you need

GPU memory
16 GiB
Model files
4 GiB known subtotal

Remaining hardware or download-size estimates are not available yet.

Requirements

Models

19

  • .safetensors

    Not resolved

    Checkpoint · Unknown

  • codeformer-v0.1.0.pth

    Not resolvedOptional

    Checkpoint · Unknown

  • dw-ll_ucoco_384_bs5.torchscript.pt

    FoundOptional

    Checkpoint · 129 MB · PT · Unknown

  • face_yolov8n.pt

    Not resolvedOptional

    Object Detector · Unknown

  • inswapper_128.onnx

    FoundOptional

    Checkpoint · 529 MB · ONNX · FP32 (unverified) · Unknown

  • models_t5_umt5-xxl-enc-bf16.pth

    Repository matched

    Text encoder · Unknown

  • open-clip-xlm-roberta-large-vit-huge-14_visual_fp16.safetensors

    Found

    Text encoder · 1.18 GB · SAFETENSORS · Unknown

  • Qwen_v2.safetensors

    Not resolved

    Checkpoint · Unknown

  • realisticVisionV60B1_v51VAE.safetensors

    Not resolvedOptional

    VAE · Unknown

  • rife49.pth

    Found

    Checkpoint · 20.4 MB · PTH · Unknown

  • sam_vit_b_01ec64.pth

    Repository matchedOptional

    Segmentation · Unknown

  • test.safetensors

    Not resolved

    Checkpoint · Unknown

  • wan_2.1_vae_fp32.safetensors

    File specified

    VAE · 484 MB · Hugging Face · tuanmeowyu/comfyui-models

  • wan2.2_i2v_lightx2v_4steps_lora_v1_high_noise.safetensors

    Found

    LoRA · 1.14 GB · SAFETENSORS · FP32 + I64 (tensor header) · Hugging Face · Comfy-Org/Wan_2.2_ComfyUI_Repackaged

  • wan2.2_i2v_lightx2v_4steps_lora_v1_low_noise.safetensors

    Found

    LoRA · 1.14 GB · SAFETENSORS · FP32 + I64 (tensor header) · Hugging Face · Comfy-Org/Wan_2.2_ComfyUI_Repackaged

  • Wan2.2-I2V-A14B-HighNoise-Q4_K_M.gguf

    Not resolved

    Checkpoint · Unknown

  • Wan2.2-I2V-A14B-LowNoise-Q4_K_M.gguf

    Not resolved

    Checkpoint · Unknown

  • wan2.2-t2v-a14b-controlnet-hed-v1.safetensors

    Not resolvedOptional

    ControlNet · Unknown

  • yolox_l.onnx

    Not resolvedOptional

    Object Detector · Unknown

Node packs

10

  • comfyui_controlnet_aux

    PossibleOptional

    Node pack · Registry

  • comfyui_fizznodes

    Found

    Node pack · Registry

  • Found

    Node pack · Registry

  • crt-nodes

    FoundOptional

    Node pack · Registry

  • Fast Bypasser (rgthree)

    Not resolved

    Node pack · Unknown

  • Fast Muter (rgthree)

    Not resolved

    Node pack · Unknown

  • kijai/ComfyUI-KJNodes

    Found

    Node pack · GitHub · kijai/ComfyUI-KJNodes

  • Mute / Bypass Repeater (rgthree)

    Not resolved

    Node pack · Unknown

  • RES4LYF

    Found

    Node pack · Registry

  • was-ns

    Not resolved

    Node pack · Registry

AI-generated commentary

AI-generated explanation based on source and configuration details. Suggestions are clearly labeled.

Purpose

  • Wan2.2_I2V_infinity_video_part_by_part(PPP)_max_consistency_FaceReactor_1.0 — v1.4. V1.4 is still not perfect, but it works quite well. Since I am currently working on another project, I am making my current version available to you here. It is not quite finished yet, but it produces good results.

Flow

  • The current graph contains 11 retained relationships for this resource.

Setup

  • Current retained source (civitai_workflow) is published by Deathkiller1337 at release v1.4.

Dependencies

  • inswapper_128.onnx: retired.

  • bbox/face_yolov8n.pt: retired.

  • RES4LYF: resolved.

  • wan2.2_i2v_lightx2v_4steps_lora_v1_low_noise.safetensors: retired.

  • comfyui_fizznodes: resolved.

  • rife49.pth: retired.

  • Wan2.2-I2V-A14B-HighNoise-Q4_K_M.gguf: resolved.

  • comfyui_fizznodes: retired.

Configuration

  • Sweet Tea records presentation quality as staged with no listed presentation blockers.

Cautions

  • This commentary explains retained evidence; it does not independently assert that the resource is recommended or runtime-tested.

    Suggestion · not verified

Does this need to be edited?

Sign in to send an edit request.

Sources

1 source
CivitaiPrimaryVersion 2936071
civitai.com/models/2396602

Verified Nov 3

Workflow instructions

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

If you dont use face compare, set all to "no" (OPTIONAL) use model for Face ReActor.

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

If you want to generate Image-to-end-image (start/end frame), set Enable Endframe to "yes"

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

How to merge all the videos you created 1. create an txt file 2. Enter this: @echo off setlocal enabledelayedexpansion :: ADJUST PATH: Enter the path to your ffmpeg.exe in ComfyUI here. set FFMPEG_PATH=D:\ComfyUI_Windows_portable\python_standalone\Scripts\ffmpeg.exe if exist list.txt del list.txt (for %%i in (*.mp4) do @echo file '%%i') > list.txt "%FFMPEG_PATH%" -f concat -safe 0 -i list.txt -c copy "OUTPUT_%date%.mp4" if %ERRORLEVEL% EQU 0 ( echo. echo Success! File saved as: OUTPUT_%date%.mp4 ) else ( echo. echo ERROR: FFmpeg was not found or files are incompatible. echo Check the FFMPEG_PATH in the script! ) del list.txt pause 3. save file as VideoJoiner.bat 4. Create a folder and copy the videos you want to merge into it. 5. Copy your Joinvideo.bat file into the same folder and run it. Your Video is joined.

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

To generate new Videos, just set switches to "No" and "false" To extend an existing Video, setz switches to "Yes" and "true" (If you use video extension, you don't need a start frame.)

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

If you have an frameskip after ReAktor then go to: ...\ComfyUI\custom_nodes\comfyui-reactor\scripts\reactor_sfw.py open it with Texteditor or Notepad++ Set: "SCORE = 0" and enter in line after: "def nsfw_image(img_data, model_path: str):" this: "return False" this will disable the NSFW filter. Example: SCORE = 0 logging.getLogger("transformers").setLevel(logging.ERROR) def nsfw_image(img_data, model_path: str): return False

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

1. Load 2 comparison images. If none are available, simply load the start image twice. 1.2 (OPTIONAL) Load Face Model for ReActor if available 2. Load start image. (The image to animate) 3. (OPTIONAL) Set an Endframe 4. (OPTIONAL) Extend an existing Video. (this will create the next 81 Frames of an existing video) (You have to assemble them manually) 5. Enter prompts. 6. Set landscape/portrait orientation. ​portrait format: 480 x 832 / Landscape format: 832 x 480 7. (OPTIONAL) change seed or steps per sampler. 8. (OPTIONAL) Load loras 9 & 10 (OPTIONAL) Setup Controlnet and Pose image (Work in progress) 11 (OPTIONAL) Choose your favorite model and text encoder. (Optional) enable color correction. Everything you need, will be in Output section. Lesser scrolling needed ;) (You don't need to change any wire manually. You don't need to manually create a bypass. Everything is done via switches.)

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

Disable if you want to save storage space.

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

Enter the Name of latent you want to save. You need this for Video extention with load latent.

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

If the colors in your video are tinged with dark or red, activate color correction. (Standart strenght: 0.85 - 1.00) (I only use this when extend videos is enabled)

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

To use Controlnet: yes false yes (Enter Value for control) Disable Controlnet: No true no _______________________________________ Image as Pose: Upload an Image to copy Pose Skelett_Image as Pose: Upload an Posefile(colored stickman)

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

Work in progress It's not working properly yet.

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

Work in progress It's not working properly yet.

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

1 = Portrai 832 x 480 2 = Landscape 480 x 823 3 = Manual value

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

FAST MODE Generate only 41 frames instead of 81 and double them.

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

Set seed and Steps for Sampler 3 Sampler standart: 3 Sampler Fast: High 2 High 1 (Base) High 4 High 2 (Movement) Low 3 low 3 (Textures)

Author notes from wan22I2VInfinityVideoPartBy_v14.json · selected release

To change the latentfile standardpath, you need to "Expand" the "Path Node" and enter your path to Input folder

Download files

Model artifacts

2 artifacts

wan22I2VInfinityVideoPartBy_v14.json

json · 208 KB · SHA-256 b47dd24f1f99…130b · Civitai

Download file

wan22I2VInfinityVideoPartBy_v14.json

json · 208 KB · SHA-256 b47dd24f1f99…130b

Download file

Source excerpts

3 excerpts
NoteNov 3civitai.com/models/2396602

Source context: 639 downloads · Type Workflows · Base model Wan Video 2.2 I2V-A14B

NoteOct 2civitai.com/models/2396602

Source context: 612 downloads · Type Workflows · Base model Wan Video 2.2 I2V-A14B

NoteJul 28civitai.com/models/2396602

Source context: 552 downloads · Type Workflows · Base model Wan Video 2.2 I2V-A14B