Enterprise Plans
AI Application Development
AI Translator API
AI SEO/GEO Service
Web Scraping Service
Log in
const main = async () => {
const response = await fetch('https://api.ai.cc/v2/video/generations', {
method: 'POST',
headers: {
Authorization: 'Bearer ',
'Content-Type': 'application/json',
},
body: JSON.stringify({
model: 'alibaba/wan2.2-vace-fun-a14b-reframe',
prompt: 'Mona Lisa puts on glasses with her hands.',
video_url: 'https://storage.googleapis.com/falserverless/example_inputs/wan_animate_input_video.mp4',
image_url: 'https://s2-111386.kwimgs.com/bs2/mmu-aiplatform-temp/kling/20240620/1.jpeg',
resolution: "720p",
}),
}).then((res) => res.json());
console.log('Generation:', response);
};
main()
import requests
def main():
url = "https://api.ai.cc/v2/video/generations"
payload = {
"model": "alibaba/wan2.2-vace-fun-a14b-reframe",
"prompt": "Mona Lisa puts on glasses with her hands.",
"video_url": "https://storage.googleapis.com/falserverless/example_inputs/wan_animate_input_video.mp4",
"image_url": "https://s2-111386.kwimgs.com/bs2/mmu-aiplatform-temp/kling/20240620/1.jpeg",
"resolution": "720p",
}
headers = {"Authorization": "Bearer ", "Content-Type": "application/json"}
response = requests.post(url, json=payload, headers=headers)
print("Generation:", response.json())
if __name__ == "__main__":
main()
-
AI Playground
Test all API models in the sandbox environment before you integrate.
We provide more than 300 models to integrate into your app.
Product Detail
Wan 2.2 VACE Reframe is an advanced video generation and editing model engineered for sophisticated video-to-video transformations. It offers unparalleled fine-grained control over video style and motion dynamics, all while rigorously preserving subject identity and video quality. Built upon the robust VACE architecture and empowered by Wan 2.2 T2V A14B weights, this model enables creators to generate, reframe, and precisely modify videos, whether locally or via API, with exceptional efficiency and accuracy.
🚀 Technical Specifications
- Architecture: VACE architecture combined with Wan 2.2 T2V A14B diffusion-based model.
- Control Conditions: Supports precise control via Pose, Depth, MLSD (Multi-Line Segment Detection), Canny edges, and Trajectory control.
- Frame Rate: Optimized for smooth video prediction at an impressive 81 frames per second (fps).
- Resolution: Supports 512p, 768p, 1024p output resolutions, with theoretical capability for 1080p in longer videos.
- Input Types: Primarily designed for video-to-video transformations; also supports additional inputs like still images and text prompts for related VACE Fun workflows.
✨ Performance Benchmarks
- Video Quality: Ensures exceptional temporal coherence and subject fidelity across frames, resulting in high-quality, natural video motion.
- Versatility: Features multi-language prediction, broadening its adoption across diverse global user bases.
- Control Precision: Achieves superior generation accuracy by enabling fine control over pose, depth, and motion trajectories, outperforming many open-source alternatives.
💡 Key Features
- Video Reframing Specialization: Tailored to reframe videos by adeptly modifying perspective, motion, or style while preserving the original content’s spatial coherence and subject identity.
- Multi-Condition Control: Offers precise input controls including pose, depth maps, MLSD, Canny edge detection, and trajectory paths for accurate motion and framing adjustments.
- High-Fidelity Temporal Consistency: Maintains seamless frame-to-frame transitions and consistent object positioning, crucial for high-quality video reframing free from flickering or distortion.
- Motion & Scene Stabilization: Incorporates advanced VACE 2.0 engine features for camera movement stabilization, background fixation, and the integration of special effects (e.g., fire, smoke) during reframing.
- Versatile Output Resolution: Capable of generating reframed videos up to 1080p with smooth 24+ FPS playback, optimized for consumer-grade GPU hardware for local usage.
💲 API Pricing
- 360p: $0.0525
- 540p: $0.07875
- 720p: $0.105
✅ Use Cases
- Content Creation: Ideal for video content creation and re-framing for social media, advertising campaigns, and entertainment.
- Advanced Video Editing: Perfect for scenarios requiring specific style or intricate motion adjustments within existing video footage.
- Animation Generation: Create dynamic animations from still images or reference videos with enhanced control.
- Local Video Production: Empower users to generate videos locally, significantly reducing reliance on cloud infrastructure.
- Research & Development: Valuable for academic and industrial research applications in video synthesis and style transfer.
💻 Code Sample
<snippet data-name="alibaba.create-video-to-video-generation" data-model="alibaba/wan2.2-vace-fun-a14b-reframe"></snippet>
📊 Comparison with Other Models
vs Wan 2.1 VACE: Wan 2.2 offers a substantial leap in video quality with significantly higher frame rates (81 fps vs ~30 fps) and supports more precise control conditions like trajectory and MLSD. It also benefits from a more efficient VAE compression, enabling faster and higher resolution video reframing while better preserving spatial coherence than its predecessor.
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 Reframe specializes in video-to-video reframing with multi-condition control. This makes VACE Reframe superior for motion and subject preservation in existing footage, as opposed to generating synthetic scenes from text.
vs Wan 2.2-I2V-A14B (Image-to-Video): The VACE Reframe model is expertly tailored for video reframing with fine trajectory and pose controls. In contrast, Wan 2.2-I2V focuses on transforming still images into videos with expert-guided detail enhancement. Consequently, VACE Reframe is the preferred choice for editing and restyling existing videos, while I2V excels in animating from static images.
🔗 API Integration
The model is readily accessible via AI/ML API. Comprehensive documentation is available here.
❓ Frequently Asked Questions (FAQ)
Q: What is Wan 2.2 VACE Reframe primarily designed for?
A: Wan 2.2 VACE Reframe is a state-of-the-art AI model specifically designed for advanced video-to-video transformations, offering precise control over video style and motion while preserving subject identity and video quality.
Q: What specific control conditions does it support for video transformation?
A: It supports fine granular control through conditions such as Pose, Depth, MLSD (Multi-Line Segment Detection), Canny edges, and Trajectory control, enabling highly accurate motion and framing adjustments.
Q: How does Wan 2.2 VACE Reframe ensure high video quality and temporal consistency?
A: The model maintains superior temporal coherence and subject fidelity across frames, ensuring smooth frame-to-frame transitions and consistent object positioning, which is critical for natural and high-quality reframed videos.
Q: What are the main differences between Wan 2.2 VACE Reframe and Wan 2.2-T2V-A14B?
A: Wan 2.2 VACE Reframe specializes in video-to-video transformations with multi-condition control for existing footage, focusing on motion and subject preservation. Wan 2.2-T2V-A14B, conversely, is optimized for generating cinematic scenes from text prompts (text-to-video).
Q: Can Wan 2.2 VACE Reframe be used for local video generation?
A: Yes, it is adapted for consumer-grade GPU hardware, allowing creators to generate, reframe, and modify videos locally, reducing the dependency on cloud-based solutions.
Learn how you can transformyour company with AICC APIs
