wan-video/wan-2.2-i2v-fast 🔢🖼️📝✓❓ → 🖼️

⭐ Official ▶️ 5.8M runs 📅 Jul 2025 ⚙️ Cog 0.16.9

About

A very fast and cheap PrunaAI optimized version of Wan 2.2 A14B image-to-video

Example Output

Prompt:

"Close-up shot of an elderly sailor wearing a yellow raincoat, seated on the deck of a catamaran, slowly puffing on a pipe. His cat lies quietly beside him with eyes closed, enjoying the calm. The warm glow of the setting sun bathes the scene, with gentle waves lapping against the hull and a few seabirds circling slowly above. The camera slowly pushes in, capturing this peaceful and harmonious moment."

Output

Performance Metrics

32.82s Prediction Time

32.83s Total Time

All Input Parameters

{
  "image": "https://replicate.delivery/pbxt/NRvtedaIOd3pdE0pTE3L9uavxJ53g33THGr0HF81M2olNOce/replicate-prediction-g8gbs3rbk9rme0crbhwatpsq04.jpg",
  "prompt": "Close-up shot of an elderly sailor wearing a yellow raincoat, seated on the deck of a catamaran, slowly puffing on a pipe. His cat lies quietly beside him with eyes closed, enjoying the calm. The warm glow of the setting sun bathes the scene, with gentle waves lapping against the hull and a few seabirds circling slowly above. The camera slowly pushes in, capturing this peaceful and harmonious moment.",
  "go_fast": true,
  "num_frames": 81,
  "resolution": "480p",
  "sample_shift": 12,
  "frames_per_second": 16,
  "lora_scale_transformer": 1,
  "lora_scale_transformer_2": 1
}

Input Parameters

seed Type: integer: Random seed. Leave blank for random
image (required) Type: string: Input image to generate video from.
prompt (required) Type: string: Prompt for video generation
go_fast Type: booleanDefault: true: Go fast
last_image Type: string: Optional last image to condition the video generation. If provided, creates smoother transitions between frames.
num_frames Type: integerDefault: 81Range: 81 - 121: Number of video frames. 81 frames give the best results
resolution Default: 480p: Resolution of video. 16:9 corresponds to 832x480px, and 9:16 is 480x832px
sample_shift Type: numberDefault: 12Range: 1 - 20: Sample shift factor
frames_per_second Type: integerDefault: 16Range: 5 - 30: Frames per second. Note that the pricing of this model is based on the video duration at 16 fps
interpolate_output Type: booleanDefault: false: Interpolate the generated video to 30 FPS using ffmpeg
disable_safety_checker Type: booleanDefault: false: Disable safety checker for generated video.
lora_scale_transformer Type: numberDefault: 1: Determines how strongly the transformer LoRA should be applied.
lora_scale_transformer_2 Type: numberDefault: 1: Determines how strongly the transformer_2 LoRA should be applied.
lora_weights_transformer Type: string: Load LoRA weights for the HIGH transformer. Supports arbitrary .safetensors URLs from the Internet (for example, 'https://huggingface.co/TheRaf7/instagirl-v2/resolve/main/Instagirlv2.0_hinoise.safetensors')
lora_weights_transformer_2 Type: string: Load LoRA weights for the LOW transformer_2. Supports arbitrary .safetensors URLs from the Internet. Can be different from transformer LoRA. (for example, 'https://huggingface.co/TheRaf7/instagirl-v2/resolve/main/Instagirlv2.0_lownoise.safetensors')

Output Schema

Output

Type: string • Format: uri

Example Execution Logs

generating video...
Inference took 21.49 seconds
Interpolating video from 16fps to 30fps...
Video interpolation completed: /tmp/tmpp5k4_fl7/output_30fps.mp4

Version Details

Version ID: febae7d9656309cf8c5df4842b27ae4768c0e47a0e1ce443a5ae81f896956134
Version Created: December 16, 2025

Run on Replicate →