How to Run Wan2.2 Image to Video GGUF Models in ComfyUI (Low VRAM)

Welcome to the tutorial on How to Run Wan2.2 Image to Video GGUF Models in ComfyUI (Low VRAM). This guide is designed for users who want to leverage the powerful capabilities of the Wan2.2 model to convert images into stunning videos, even on systems with low VRAM (less than 12GB). The Wan2.2 model is a state-of-the-art image-to-video generation tool that utilizes advanced AI techniques to create high-quality videos from still images. In this tutorial, we will walk you through the entire process, from setting up the necessary environment to generating your first video.

By the end of this guide, you will have a comprehensive understanding of how to effectively use the Wan2.2 model in ComfyUI, ensuring that you can create dynamic video content without the need for high-end hardware.

Before diving into the setup process, it’s crucial to ensure that your system meets the requirements for running the Wan2.2 Video Generation Model in ComfyUI, especially if you are working with low VRAM configurations. The GGUF version of the Wan2.2 model is optimized for efficiency, enabling smooth performance on systems with 12 GB of VRAM or less.

Requirement 1: ComfyUI Installed

To begin using ComfyUI, you’ll need to have it installed on your local system. Follow the step-by-step instructions in the article below to get ComfyUI up and running on Windows.

How to Install ComfyUI Locally on Windows?

Requirement 2: Update ComfyUI

To ensure compatibility with the Wan2.2 GGUF models, make sure your ComfyUI installation is up to date.

• Via ComfyUI Interface:
  Click on "Manager" in the top-right corner of ComfyUI, then select "Update ComfyUI".

• For Windows Portable Users:
  Navigate to the folder:
  ...\ComfyUI_windows_portable\update
  and double-click the file named update_comfyui.bat.

Keeping ComfyUI updated ensures access to the latest features, bug fixes, and compatibility improvements.

Requirement 3: Download Wan 2.2 Model Files

Next, we need to download the required model files for Wan2.2. You can find them on the Wan2.2 Hugging Face main GGUF model page. Below is a table listing all the required files along with their corresponding folders where they should be placed.

To wrap up, you'll also notice that we’ll be downloading two LoRA models. This helps accelerate the rendering process by allowing you to reduce the KSampler steps to 4 —speeding up generation without sacrificing quality in the final video output.

Verify Folder Structure

Before we proceed, let’s ensure that all Wan2.2 model files are correctly placed in their respective ComfyUI folders. Having everything in the right place ensures the workflow loads smoothly without errors.

ts
1📁 ComfyUI/
2└── 📁 models/
3    ├── 📁 text_encoders/
4    │   └── umt5_xxl_fp8_e4m3fn_scaled.safetensors
5    ├── 📁 diffusion_models/
6    │   ├── Wan2.2-I2V-A14B-HighNoise-Q3_K_S.gguf
7    │   └── Wan2.2-I2V-A14B-LowNoise-Q3_K_S.gguf
8    ├── 📁 vae/
9    │   └── Wan2.1_VAE.safetensors
10    └── 📁 loras/
11        ├── Wan2.2-Lightning_I2V-A14B-4steps-lora_HIGH_fp16.safetensors
12        └── Wan2.2-Lightning_I2V-A14B-4steps-lora_LOW_fp16.safetensors

Now that all the required files are downloaded and properly organized, let’s move on to downloading and loading the Wan2.2 GGUF JSON Workflow so we can start generating high-quality videos from images!

With the requirements in place, the next step is to download and load the Wan2.2 GGUF JSON Workflow into ComfyUI. This workflow is essential for configuring the model and ensuring it operates correctly.

Step 1: Download the Workflow File

Begin by downloading the workflow file specifically designed for the Wan2.2 GGUF model. This file contains all the necessary configurations and settings required for the model to function properly. You can find the download link below:

👉 Download Wan2.2 I2V GGUF Workflow JSON

Step 2: Load the Workflow in ComfyUI

Once you’ve downloaded the workflow file, launch ComfyUI and simply drag and drop the .json file onto the canvas to load the full setup and begin working with the I2V model.

In the next section, we’ll go over all the essential settings needed to run the Wan2.2 GGUF image to video model with the use of the Lighting (lightX2V) LoRA's smoothly and efficiently.

Now that the Wan2.2 GGUF workflow is loaded into ComfyUI, it’s time to configure the settings to optimize video generation. Proper configuration is crucial for achieving high-quality outputs, especially when working with low VRAM systems.

### Step 1: Load Models (via Unet Loader)

Use the Unet Loader node to load both high and low noise models:

• High Noise: Wan2.2-I2V-A14B-HighNoise-Q3_K_S.gguf

• Low Noise: Wan2.2-I2V-A14B-LowNoise-Q3_K_S.gguf

Also load:

• CLIP Text Encoder: umt5_xxl_fp8_e4m3fn_scaled.safetensors

• VAE: Wan2.1_VAE.safetensors

Step 2: Upload Base Image

Import the base image that will act as the first frame of the video. This defines character design, pose, lighting, and scene layout.

Step 3: Load LoRA

Use a LoRA Loader node to load:

• Wan2.2-Lightning_I2V-A14B-4steps-lora_HIGH_fp16.safetensors

• Wan2.2-Lightning_I2V-A14B-4steps-lora_LOW_fp16.safetensors

These LoRAs provide faster rendering while maintaining high visual quality.

Step 4: Enter Prompt and Negative Prompt

Write a detailed prompt to describe the animation behavior and camera motion.

Step 5: Set Video Resolution and Length

Match the aspect ratio of your base image and set the desired animation duration. If you encounter Out of Memory (OOM) errors, try reducing the resolution (e.g., to 840 × 480) or shortening the length (e.g., to 81 frames).

• Width: 1280

• Height: 720

• Length: 121 frames

• Batch Size: 1

Step 6: Configure High Noise Ksampler Settings

Settings for motion-rich, expressive animation:

• Add Noise: Enabled

• Control After Generate: Randomize

• Total Steps: 4

• CFG: 1

• Sampler: Euler

• Scheduler: Simple

• Start at Step: 0

• Start end Step: 2

• Return with Leftover Noise: Enabled

Step 7: Configure Low Noise Ksampler Settings

Stabilizes animation and refines details:

• Add Noise: Disabled

• Control After Generate: Randomize

• Total Steps: 4

• CFG: 1

• Sampler: Euler

• Scheduler: Simple

• Start at Step: 2

• Start end Step: 10000

• Return with Leftover Noise: Disabled

Step 8: Set FPS and Video Output Settings

• FPS: 30
  Creates a smooth ~4-second clip (121 ÷ 30 ≈ 4s).

• Configure the video output:

  • File Prefix: videoComfyUI

  • Format: auto or mp4

  • Codec: h264

With everything in place, click RUN to generate your animated video.

After configuring the settings and clicking "RUN", the Wan2.2 model will begin generating your video. Depending on your hardware capabilities, the generation time may vary, especially if you are working with longer or higher-resolution videos.This generation was run on an RTX 3060 (12 GB VRAM) and took 900 seconds to complete at 840 × 420 resolution, producing 81 frames.

Example 1:

Example 2:

Tips for Success

• Start with shorter video lengths to gauge performance on your system.

• Experiment with various prompts to see how they influence the final output.

• Keep an eye on the resource usage to ensure smooth operation on low VRAM systems.

With these examples and tips, you are now ready to explore the capabilities of the Wan2.2 GGUF image-to-video model in ComfyUI.

This tutorial has guided you through running the Wan2.2 image-to-video GGUF models in ComfyUI, optimized for low VRAM usage. From model setup to workflow configuration and video generation, you now have a complete understanding of how to bring still images to life.

By integrating the lightx2v LoRA, we've significantly reduced rendering time while maintaining visual quality—making the process more efficient and accessible, even on modest hardware.

As AI tools continue to evolve, we encourage you to experiment, refine, and stay up to date with new improvements. Thanks for following along—now go create something amazing!