ArtificialGuyBR

Home / Blog / AI video

Metalcification LoRA for Wan 2.2: Turn Objects into Metal

Create metalcification video effects with this Wan 2.2 I2V LoRA. Trigger word, ComfyUI setup, high/low noise variants.

2 sources cited AI video

On this page

  1. What Is Metalcification?
  2. Trigger Word
  3. Two Variants: High Noise vs Low Noise
  4. ComfyUI Setup
  5. Recommended Settings
  6. License
  7. Common Errors and Fixes
  8. FAQ
  9. Sources

Metalcification LoRA for Wan 2.2: Turn Objects into Metal

What Is Metalcification?

Metalcification is a LoRA for Wan 2.2 I2V that creates a gradual metallization effect. Objects in the generated video transition from their original material into polished metal — capturing surface reflections, highlights, and texture changes frame by frame. The effect works on any subject: organic shapes, manufactured objects, architectural elements.

The LoRA was trained with sponsorship from Redmond Ai and released under CC0-1.0, making it free for commercial and non-commercial use.

Trigger Word

Use Metalcification in your prompt. The model was trained on this single trigger word — it activates the full transformation sequence. Example:

Metalcification, a wooden chair transforming into polished chrome, cinematic lighting

Two Variants: High Noise vs Low Noise

The release includes two SafeTensor files: Variant | File | Best For

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

Low Noise | [WAN2.2]Metalcification_Redmond_low_noise.safetensors | Clean, subtle metallization, fewer artifacts High Noise | [WAN2.2]Metalcification_Redmond_high_noise.safetensors | Dramatic, aggressive metal transformation Start with low noise at strength 0.8–1.0. Increase strength or switch to high noise if the effect is too subtle.

ComfyUI Setup

  1. Place both .safetensors files in ComfyUI/models/loras/
  2. Add a Load LoRA node after your Wan model loader
  3. Connect model output → Load LoRA model input
  4. Select the variant (low noise recommended first)
  5. Connect Load LoRA MODEL output → KSampler
  6. Set strength_model to 0.8–1.0

Ensure your ComfyUI environment already runs Wan models (see ComfyUI Wan guide).

License

CC0-1.0 — public domain dedication. Free for any use including commercial.

Common Errors and Fixes

Error | Cause | Fix

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

No metallization visible | Trigger word missing or misspelled | Add exact Metalcification to prompt Artifacts / flickering | Strength too high or high-noise variant | Lower strength to 0.7–0.8 or switch to low-noise variant Model not loading | Wrong base model | Use Wan Video 2.2 I2V-A14B (not T2V or 1.3B variants)

FAQ

Q: Does this work with Wan T2V (text-to-video) models? A: The HF card lists Wan2.2-T2V-A14B as base, but Civitai specifies I2V-A14B. Test with I2V first; T2V may work but isn't the primary target.

Q: Can I use only one of the two LoRA files? A: Yes. Low noise is the safer default. High noise is for stronger effect.

Q: Is there a recommended resolution? A: Wan 2.2 I2V works at 480p–720p typically. Higher resolutions need more VRAM.

Sources