Wan-2.2-i2v-480p-lora-ultra-fast

Wan 2.2 (Ultra-Fast 480p, with LoRA) is a high-speed multimodal video generation model. It delivers cinematic-quality results at ultra-fast inference speed, with support for up to 3 LoRAs per job for flexible style and character control.

• Wan-2.2-i2v-480p-lora
• Wan-2.2/i2v-480p-ultra-fast

Key Features

• Cinematic-level Aesthetic Control: Professional camera language, multi-dimensional control over lighting, color, and composition.

• Large-scale Complex Motion: Smoothly restores natural motion, supports multi-subject dynamics, and enhances controllability.

• Precise Semantic Compliance: Excels at complex scene understanding and multi-object generation, ensuring faithful creative intent.

• LoRA Integration: Import up to 3 LoRAs per job for both high-noise and low-noise experts, with adjustable blending scale.

Limits and Performance

• Resolution: 480p

• Duration options: 5s or 8s

• Input types:

  • Prompt
  • Image (First Frame)
  • Last Image (Last Frame)

• LoRAs: up to 3 high-noise LoRAs + 3 low-noise LoRAs or just 3 LoRAs

• Seed: reproducibility control

Pricing

Duration	Cost
5 seconds	$0.10
8 seconds	$0.16

How to Use

1. Upload an initial image.
2. Write a prompt describing the video scene.
3. (Optional) Add a last_image for smooth transitions.
4. Select duration (5s or 8s).
5. Add LoRAs (up to 3 for high-noise experts, 3 for low-noise experts).
6. (Optional) Set a seed for reproducibility.
7. Run the job and preview/download your video.

Pro Tips

• Use image + last_image for storyboarding transitions.
• Apply high-noise LoRAs for global style changes, and low-noise LoRAs for subtle refinements.
• Keep LoRA scale values balanced (0.5–1.0 recommended) for natural blending.
• Choose 5s for quick iterations and 8s for polished results.

Note

• If you did not upload the image locally, please ensure that the image URL is accessible! A successfully accessible image will display a preview in the interface.

Reference

• Use your LoRA
• Train your LoRA