Magic Animate

magic-animate

MagicAnimate: Temporally Consistent Human Image Animation using Diffusion Model

A100 80GB

Fast Inference

REST API

Try in Console API Docs Examples

Model Information

Response Time~70 sec

StatusActive

Version

0.0.1

Updated13 days ago

API Reference

View Full Documentation

Prerequisites

Create an API Key from the Eachlabs Console
Install the required dependencies for your chosen language (e.g., requests for Python)

API Integration Steps

1. Create a Prediction

Send a POST request to create a new prediction. This will return a prediction ID that you'll use to check the result. The request should include your model inputs and API key.

import requests
import time

API_KEY = "YOUR_API_KEY"  # Replace with your API key
HEADERS = {
    "X-API-Key": API_KEY,
    "Content-Type": "application/json"
}

def create_prediction():
    response = requests.post(
        "https://api.eachlabs.ai/v1/prediction/",
        headers=HEADERS,
        json={
            "model": "magic-animate",
            "version": "0.0.1",
            "input": {
  "seed": null,
  "image": "your_file.image/jpeg",
  "video": "your_file.video/mp4",
  "guidance_scale": "7.5",
  "num_inference_steps": "25"
}
        }
    )
    prediction = response.json()
    
    if prediction["status"] != "success":
        raise Exception(f"Prediction failed: {prediction}")
    
    return prediction["predictionID"]

2. Get Prediction Result

Poll the prediction endpoint with the prediction ID until the result is ready. The API uses long-polling, so you'll need to repeatedly check until you receive a success status.

def get_prediction(prediction_id):
    while True:
        result = requests.get(
            f"https://api.eachlabs.ai/v1/prediction/{prediction_id}",
            headers=HEADERS
        ).json()
        
        if result["status"] == "success":
            return result
        elif result["status"] == "error":
            raise Exception(f"Prediction failed: {result}")
        
        time.sleep(1)  # Wait before polling again

3. Complete Example

Here's a complete example that puts it all together, including error handling and result processing. This shows how to create a prediction and wait for the result in a production environment.

try:
    # Create prediction
    prediction_id = create_prediction()
    print(f"Prediction created: {prediction_id}")
    
    # Get result
    result = get_prediction(prediction_id)
    print(f"Output URL: {result['output']}")
    print(f"Processing time: {result['metrics']['predict_time']}s")
except Exception as e:
    print(f"Error: {e}")

Additional Information

The API uses a two-step process: create prediction and poll for results
Response time: ~70 seconds
Rate limit: 60 requests/minute
Concurrent requests: 10 maximum
Use long-polling to check prediction status until completion

Overview

Magic Animate is a model that animates static human images by applying motion patterns extracted from a reference video. This approach ensures temporal consistency, resulting in smooth and natural animations.

Technical Specifications

Magic Animate employs a diffusion model to animate static human images. By leveraging motion information from a reference video, the Magic Animate generates temporally consistent animations, ensuring smooth transitions and realistic movements.

Key Considerations

Temporal Consistency: The Magic Animate ensures that animations are smooth and free from temporal artifacts.

Motion Alignment: The quality of the output heavily depends on the alignment between the input image and the reference video's motion.

Parameter Sensitivity: Adjusting parameters like num_inference_steps and guidance_scale can significantly impact the animation quality.

Tips & Tricks

Input Image:

Ensure the image is high-resolution and well-lit.
The subject should be clearly visible without obstructions.

Reference Video:

Select videos where the motion aligns with the intended animation.
Ensure the video's perspective matches that of the input image for seamless integration.

Parameter Settings:

Number of Inference Steps:
- Range: 1 to 200.
- For detailed and refined animations, consider setting this parameter between 100 and 150.
Guidance Scale:
- Range: 1 to 50.
- A value between 15 and 25 often provides a good balance between adhering to the input image and incorporating the reference video's motion.
Seed:
- Setting a specific seed ensures reproducibility of results.
- If variability is desired, use different seed values for each run.

Capabilities

Realistic Animation: Transforms static images into dynamic animations by applying motion from reference videos.

Temporal Consistency: Ensures that the generated animations are smooth and free from temporal artifacts.

Parameter Control: Offers adjustable parameters to fine-tune the animation process according to user preferences.

What can I use for?

Content Creation with Magic Animate: Enhance static images by adding realistic motion for multimedia projects.

Virtual Avatars: Animate character images for use in virtual environments or presentations.

Educational Tools: Create dynamic visual aids from static images to facilitate learning and engagement.

Things to be aware of

Diverse Motions: Experiment with various reference videos to observe how different motions affect the animation.

Parameter Exploration: Adjust num_inference_steps and guidance_scale to see their impact on the animation quality.

Background Simplification: Use images with simple backgrounds to evaluate the Magic Animate's performance in isolating and animating the subject.

Limitations

Pose Compatibility: The Magic Animate performs best when the poses in the input image and reference video are similar.

Complex Backgrounds: Intricate backgrounds in the input image might lead to less accurate animations.

Motion Complexity: Highly complex or rapid motions in the reference video can sometimes result in unnatural animations.

Output Format: MP4

Related AI Models

Hailuo I2V Director

hailuo-i2v-0-1

Image to Video

Live Portrait

live-portrait

Image to Video

Pixverse

pixverse

Image to Video

SadTalker

sadtalker

Image to Video