Wan 2.2 Vace Reframe overview
Wan 2.2 Vace Reframe by Alibaba Cloud
Its innovative architecture supports multiple input control conditions and offers significant compression advantages without sacrificing output fidelity.
Wan 2.2 Vace Reframe specs
- Model ID
wan-22-vace-fun-a14b/reframe- Developer
- Alibaba Cloud
- Input
- Text, Video
- Output
- Video
- Endpoint
POST https://api.aimlapi.com/v2/video/generations
Wan 2.2 VACE Reframe is a state-of-the-art video generation and editing model designed for advanced video-to-video transformations. It supports fine granular control over video style and motion dynamics while preserving subject identity and video quality. Built on the VACE architecture and paired with Wan 2.2 T2V A14B weights, this model empowers creators to generate, reframe, and modify videos locally or via API with efficiency and precision.
Technical Specifications
- Model Architecture: VACE architecture combined with Wan 2.2 T2V A14B diffusion-based model.
- Control Conditions Supported: Pose, Depth, MLSD (Multi-Line Segment Detection), Canny edges, Trajectory control.
- Frame Rate: Trained for smooth video prediction at 81 frames per second (fps).
- Resolution Support: 512p, 768p, 1024p output resolutions; theoretically capable of 1080p output in longer videos.
- Input Types: Video-to-video transformation; supports additional inputs such as still images and text prompts (for related workflows using VACE Fun).
Performance Benchmarks
- Video Quality: Maintains temporal coherence and subject fidelity across frames ensuring high-quality, natural video motion.
- Versatility: Supports multi-language prediction enhancing adoption across diverse user bases.
- Control Precision: Excels in generation accuracy by enabling fine control over pose, depth, and motion trajectories, surpassing many open-source peers.
Key Features
- Specialized for Video Reframing: Designed to reframe videos by modifying the perspective, motion, or style while preserving the original content’s spatial coherence and subject identity.
- Multi-Condition Control: Supports precise control inputs including pose, depth maps, MLSD (Multi-Line Segment Detection), Canny edge detection, and trajectory paths for accurate motion and framing adjustments.
- High-Fidelity Temporal Consistency: Maintains smooth frame-to-frame transitions and consistent object positioning, critical for high-quality video reframing without flickering or distortion.
- Motion and Scene Stabilization: Includes advanced VACE 2.0 engine features for camera movement stabilization, background fixation, and integration of special effects (e.g., fire, smoke) during reframing.
- Output Resolution Support: Capable of generating reframed videos up to 1080p with smooth 24+ FPS playback, adapted for consumer-grade GPU hardware for local use.
Comparison with Other Models
vs Wan 2.1 VACE: Wan 2.2 offers improved video quality with higher frame rates (81 fps vs ~30 fps) and supports more precise control conditions such as trajectory and MLSD. It also benefits from a more efficient VAE compression, enabling faster and higher resolution video reframing while preserving spatial coherence better than Wan 2.1.
vs Wan 2.2-T2V-A14B (Text-to-Video): While Wan 2.2-T2V is optimized for text-to-video semantic grounding and cinematic scene generation, Wan 2.2 VACE Fun A14B specializes in video-to-video reframing with multi-condition control, making it better suited for motion and subject preservation in existing footage versus synthetic scene creation.
vs Wan 2.2-I2V-A14B (Image-to-Video): The VACE Fun model is tailored for video reframing with fine trajectory and pose controls, whereas Wan 2.2-I2V focuses on turning still images into videos with expert-guided detail enhancement. This makes VACE Fun preferable for editing and restyling existing videos, while I2V excels in animation from images.
API Integration
Accessible via AI/ML API. Documentation: available here.
Use cases
- Video content creation and re-framing for social media, ads, and entertainment.
- Video editing where specific style or motion adjustments are required.
- Creating animations from still images or reference videos.
- Local video generation avoiding cloud dependency.
- Research applications in video synthesis and style transfer.
Wan 2.2 Vace Reframe API pricing
| Usage | Price | Unit |
|---|---|---|
| Video generation | $0.05 | per generation |
| Additional | $0.013 | per generation |
Prices from the AI/ML API price list, checked on Nov 4, 2026.
Wan 2.2 Vace Reframe vs other video models: prices
AI/ML API rates for 5 models, per generation.
| Model | Per generation |
|---|---|
 Wan 2.2 Vace Reframe | $0.05 |
| Wan 2.2 Vace Depth | $0.05 |
| Wan 2.2 Vace Inpainting | $0.05 |
| Wan 2.2 Vace Outpainting | $0.05 |
| Wan 2.2 Vace Pose | $0.05 |
How to use the Wan 2.2 Vace Reframe API
View docs
Build with your agent
Give your agent the model details, code and documentation.
1Get your API key
Create API keyexport AIMLAPI_KEY="8f2b4c7d9a1e6b3f…"2Make your first request
curl 'https://api.aimlapi.com/v2/video/generations' \
-H "Authorization: Bearer $AIMLAPI_KEY" \
-H 'Content-Type: application/json' \
-d '{
"model": "wan-22-vace-fun-a14b/reframe",
"video_url": "https://example.com/input.mp3"
}'import os
import requests
headers = {
"Authorization": f"Bearer {os.environ['AIMLAPI_KEY']}",
"Content-Type": "application/json",
}
payload = {
"model": "wan-22-vace-fun-a14b/reframe",
"video_url": "https://example.com/input.mp3"
}
response = requests.post("https://api.aimlapi.com/v2/video/generations", headers=headers, json=payload, timeout=120)
response.raise_for_status()
print(response.json())const response = await fetch("https://api.aimlapi.com/v2/video/generations", {
method: "POST",
headers: {
Authorization: `Bearer ${process.env.AIMLAPI_KEY}`,
"Content-Type": "application/json",
},
body: JSON.stringify({
"model": "wan-22-vace-fun-a14b/reframe",
"video_url": "https://example.com/input.mp3"
}),
});
if (!response.ok) throw new Error(`HTTP ${response.status}`);
console.log(await response.json());The request returns a generation id. Poll GET https://api.aimlapi.com/v2/video/generations?generation_id=<id> until the status is completed.
Models similar to Wan 2.2 Vace Reframe
Related articles
Wan 2.2 Vace Reframe FAQ
What inputs and outputs does Wan 2.2 Vace Reframe support?
Wan 2.2 Vace Reframe takes text, video as input and returns video.
How much does Wan 2.2 Vace Reframe cost?
Wan 2.2 Vace Reframe costs $0.05 per generation on AI/ML API, billed as you go with no subscription.
How is Wan 2.2 Vace Reframe billed?
Wan 2.2 Vace Reframe is billed per generation, a fixed charge per output rather than by prompt length.
Who created Wan 2.2 Vace Reframe?
Wan 2.2 Vace Reframe was built by Alibaba Cloud.
How do I call Wan 2.2 Vace Reframe on AI/ML API?
Send a request to v2/video/generations with wan-22-vace-fun-a14b/reframe as the model id.
Is Wan 2.2 Vace Reframe available through a single API key?
Yes. Wan 2.2 Vace Reframe is served through AI/ML API, so the same key and endpoint format used for other models applies.
Which API endpoint is used to access Wan 2.2 Vace Reframe?
It is accessed via the v2/video/generations endpoint.
What is the Wan 2.2 Vace Reframe model ID?
Use wan-22-vace-fun-a14b/reframe as the model parameter. Requests go to https://api.aimlapi.com/v2/video/generations with your AI/ML API key.