Wiggle LoRA for Wan 2.2 I2V: Subtle Motion for Static Images

• What it does: Adds a gentle "wiggle" motion to static images without changing their content
• Base model: Wan-AI/Wan2.2-I2V-A14B
• Trigger word: Wiggle
• License: CC0-1.0 (free for commercial use)

What This LoRA Does

Wiggle Redmond WAN2 I2V 14B is a LoRA that specializes in creating subtle animation effects for Wan 2.2 Image-to-Video models. Unlike full animation LoRAs that transform an image into a completely different scene, this LoRA preserves the original image while adding a gentle, almost imperceptible movement — making photographs and illustrations feel alive.

The effect is described as a "wiggle" — a natural, subtle motion that brings static content to life. It's ideal for:

• Adding life to portrait photographs
• Animating illustrations with gentle movement
• Creating subtle motion for social media content
• Enhancing static assets without dramatic transformation

Training and Sponsorship

This LoRA was trained on the Wan-AI/Wan2.2-I2V-A14B base model (14B parameters). The training was sponsored by Redmond AI, which provided the GPU compute resources. The model is released under CC0-1.0, placing it in the public domain for unrestricted use.

How to Use

Trigger Word

Include Wiggle in your prompt when describing the image you want to animate. The model card shows that simply using "Wiggle" as the prompt is sufficient.

ComfyUI Setup

1. Download both LoRA files:

• [WAN2.2]Wiggle_Redmond_low_noise.safetensors
• [WAN2.2]Wiggle_Redmond_high_noise.safetensors

(Total size: ~146 MB)

1. Place them in ComfyUI/models/loras/

1. In your ComfyUI workflow:

• Add a Load LoRA node after your base model loader
• Load your base Wan I2V model (Wan2.2-I2V-A14B)
• Connect model loader output to Load LoRA node input
• Select both LoRA files in the node
• Connect LoRA MODEL output to your KSampler
• Set strength_model to 0.8-1.0 (recommended starting range)

Example Prompt

Wiggle

The model card indicates that the single word "Wiggle" is sufficient as a prompt.

Technical Details

Property | Value

|---|---

Base Model | Wan-AI/Wan2.2-I2V-A14B Type | LoRA (Low Noise + High Noise pair) File Format | SafeTensor Total Size | 146.34 MB Trigger Word | Wiggle License | CC0-1.0 Pipeline | diffusers, image-to-video

Common Errors and Fixes

Error | Cause | Fix | Source

|---|---|---|---

LoRA not loading | Files placed in wrong directory | Ensure both .safetensors files are in ComfyUI/models/loras/ | Model card No motion effect | Missing trigger word | Include "Wiggle" in your prompt | Model card Excessive motion | Strength too high | Reduce strength_model to 0.6-0.8 | Model card ComfyUI version mismatch | Incompatible Wan I2V base model | Use Wan2.2-I2V-A14B specifically | Model card

FAQ

Q: Does this work with other Wan versions? A: This LoRA is specifically trained for Wan2.2-I2V-A14B. Compatibility with other Wan versions is not guaranteed.

Q: Can I use only one of the two LoRA files? A: The model card recommends using both low_noise and high_noise files together for best results.

Q: Is this really free for commercial use? A: Yes, CC0-1.0 places the model in the public domain with no restrictions.

Q: What resolution does it support? A: The HF widget shows examples at 1096x832 and 664x1000, matching Wan I2V's native resolutions.

Sources

• HuggingFace model card: https://huggingface.co/artificialguybr/Wiggle-Redmond-WAN2-I2V-14B
• Civitai model page: https://civitai.com/models/2212361
• Base model: https://huggingface.co/Wan-AI/Wan2.2-I2V-A14B
• ComfyUI: https://github.com/comfyanonymous/ComfyUI