Wan2.2-Fun-Camera-Control is a next-generation video generation and camera control model developed by Alibaba PAI. By introducing innovative Camera Control Codes and combining deep learning with multimodal conditional inputs, it generates high-quality videos that adhere to predefined camera motion conditions. The model is released under the Apache 2.0 license, allowing for commercial use. Key Features:
  • Camera Motion Control: Supports various camera motion modes, including Pan Up, Pan Down, Pan Left, Pan Right, Zoom In, Zoom Out, and combinations thereof.
  • High-Quality Video Generation: Based on the Wan2.2 architecture, it outputs cinematic-quality videos.
Here are the relevant model weights and code repository:

​
Wan2.2 Fun Camera Control: Video Generation Workflow Example

Make sure your ComfyUI is updated.Workflows in this guide can be found in the Workflow Templates. If you can’t find them in the template, your ComfyUI may be outdated.(Desktop version’s update will delay sometime)If nodes are missing when loading a workflow, possible reasons:
  1. Not using the latest ComfyUI version(Nightly version)
  2. Using Stable or Desktop version (Latest changes may not be included)
  3. Some nodes failed to import at startup
The workflow provided includes two versions:
  1. Using the Wan2.2-Lightning 4-step LoRA via lightx2v: This may result in reduced video dynamics but offers faster generation.
  2. The fp8_scaled version without the acceleration LoRA.
Below are the timing results tested on an RTX4090D 24GB GPU for 640*640 resolution and 81-frame length:
Model TypeResolutionVRAM UsageFirst Generation TimeSecond Generation Time
fp8_scaled640×64084%≈ 536 seconds≈ 513 seconds
fp8_scaled + 4-step LoRA640×64089%≈ 108 seconds≈ 71 seconds
While the 4-step LoRA improves initial user experience, it may reduce video dynamism. By default, the accelerated LoRA version is enabled. To switch workflows, select the nodes and press Ctrl+B.

​
1. Workflow and Asset Download

Download the video or JSON file below and drag it into ComfyUI to load the corresponding workflow. The workflow will prompt you to download the models.

Download JSON Workflow

 Please download the image below, which we will use as input. Input Starting Image The following models can be found in Wan_2.2_ComfyUI_Repackaged: Diffusion Model Wan2.2-Lightning LoRA (Optional, for acceleration) VAE Text Encoder File save location 
ComfyUI/
├───📂 models/
│   ├───📂 diffusion_models/
│   │   ├─── wan2.2_fun_camera_low_noise_14B_fp8_scaled.safetensors
│   │   └─── wan2.2_fun_camera_high_noise_14B_fp8_scaled.safetensors
│   ├───📂 loras/
│   │   ├─── wan2.2_i2v_lightx2v_4steps_lora_v1_high_noise.safetensors
│   │   └─── wan2.2_i2v_lightx2v_4steps_lora_v1_low_noise.safetensors
│   ├───📂 text_encoders/
│   │   └─── umt5_xxl_fp8_e4m3fn_scaled.safetensors 
│   └───📂 vae/
│       └── wan_2.1_vae.safetensors

​
3. Complete the Workflow Step-by-Step

Wan2.2 Fun Camera Control Workflow Steps
This workflow uses LoRA. Ensure the Diffusion model and LoRA are consistent; high noise and low noise models and LoRA must be paired accordingly.
  1. High noise model and LoRA loading
  • Ensure the Load Diffusion Model node loads wan2.2_fun_camera_high_noise_14B_fp8_scaled.safetensors
  • Ensure the LoraLoaderModelOnly node loads wan2.2_i2v_lightx2v_4steps_lora_v1_high_noise.safetensors
  1. Low noise model and LoRA loading
  • Ensure the Load Diffusion Model node loads wan2.2_fun_camera_low_noise_14B_fp8_scaled.safetensors
  • Ensure the LoraLoaderModelOnly node loads wan2.2_i2v_lightx2v_4steps_lora_v1_low_noise.safetensors
  1. Ensure the Load CLIP node loads umt5_xxl_fp8_e4m3fn_scaled.safetensors
  2. Ensure the Load VAE node loads wan_2.1_vae.safetensors
  3. Upload the starting frame in the Load Image node
  4. Modify the Prompt (both Chinese and English are acceptable)
  5. Set camera control parameters in the WanCameraEmbedding node:
    • Camera Motion: Select the camera motion type (Zoom In, Zoom Out, Pan Up, Pan Down, Pan Left, Pan Right, Static, etc.)
    • Width/Height: Set video resolution
    • Length: Set the number of video frames (default is 81 frames)
    • Speed: Set video speed (default is 1.0)
  6. Click the Run button or use the shortcut Ctrl(cmd) + Enter to execute video generation