video-generation

Transform Your Videos with AI-Powered Video Generation

25 Apr 2025
Transform Your Videos with AI-Powered Video Generation

In today's digital landscape, the ability to create and modify videos quickly and effectively can set your application apart. With the Wan 2.1 1.3b Vid2vid model from Alibaba's Tongyi Lab, developers can leverage advanced video-to-video generation capabilities. This service allows you to transform existing videos based on descriptive text prompts, resulting in visually stunning outputs that maintain high quality. Whether you are in the entertainment industry, marketing, or content creation, this tool significantly simplifies video production processes, allowing for creativity and innovation without the need for extensive video editing expertise.

Prerequisites

Before diving into the integration of the Wan 2.1 1.3b Vid2vid model, ensure you have a Cognitive Actions API key and a basic understanding of making API calls.

Generate Video Using Wan

The "Generate Video Using Wan" action is designed to facilitate video-to-video generation, enabling you to create new video content from an existing video file by utilizing a textual description. This action enhances the visual quality of the generated videos, making it an ideal choice for applications requiring high-definition video outputs.

Input Requirements

To utilize this action, you need to provide several inputs:

  • Prompt: A text description of the content you wish to generate or modify (e.g., "a puppy wearing black sunglasses").
  • Input Video: The URI of the existing video that will serve as the base for transformation.
  • Seed: An optional integer to create a random seed for reproducibility.
  • Tiled: A boolean to enable tiled sampling for improved quality on larger videos, defaulting to true.
  • Aspect Ratio: Choose between "832x480" or "480x832" for the output video.
  • Negative Prompt: A list of elements to avoid during generation, such as "low quality" or "watermark".
  • Number of Frames: Specify the total frames for the output video, ranging from 1 to 100.
  • Frames Per Second: Set the desired frame rate between 5 and 24.
  • Denoising Strength: Determine the level of denoising from 0 (none) to 1 (maximum).
  • Number of Inference Steps: Control the quality of generation with a range from 10 to 50 steps.
  • Classifier Free Guidance Scale: Adjust adherence to prompts from 0 to 20, with higher values strengthening this adherence.

Expected Output

The expected output is a newly generated video that reflects the modifications specified in the prompt while maintaining high visual fidelity. You will receive a URI link to the generated video once the process is complete.

Use Cases for this Specific Action

  • Creative Content Creation: Perfect for artists and content creators looking to produce unique video visuals based on their concepts without extensive editing.
  • Marketing Campaigns: Use this tool to create engaging advertisements that resonate with specific target audiences by customizing existing promotional videos.
  • Entertainment: Generate entertaining clips for social media or streaming platforms that adapt existing footage to current trends or themes.
  • Education and Training: Transform instructional videos to better fit different learning scenarios, enhancing viewer engagement with tailored content.
import requests
import json

# Replace with your actual Cognitive Actions API key and endpoint
# Ensure your environment securely handles the API key
COGNITIVE_ACTIONS_API_KEY = "YOUR_COGNITIVE_ACTIONS_API_KEY"
# This endpoint URL is hypothetical and should be documented for users
COGNITIVE_ACTIONS_EXECUTE_URL = "https://api.cognitiveactions.com/actions/execute"

action_id = "78d0eb7d-5e7e-4d7d-b32d-4aed95ecf0bf" # Action ID for: Generate Video Using Wan

# Construct the exact input payload based on the action's requirements
# This example uses the predefined example_input for this action:
payload = {
  "tiled": true,
  "prompt": "a puppy wearing black sunglasses",
  "inputVideo": "https://replicate.delivery/pbxt/MbUZEHGJFGoTaCAlbxPCAHvkmQrWOwAWGhb3o6xwdYRtoV7C/replicate-prediction-6jxv92xhadrma0cncbabf652xm.mp4",
  "aspectRatio": "832x480",
  "negativePrompt": "low quality, blurry, distorted, disfigured, text, watermark",
  "numberOfFrames": 81,
  "framesPerSecond": 16,
  "denoisingStrength": 0.7,
  "numberOfInferenceSteps": 30,
  "classifierFreeGuidanceScale": 6
}

headers = {
    "Authorization": f"Bearer {COGNITIVE_ACTIONS_API_KEY}",
    "Content-Type": "application/json",
    # Add any other required headers for the Cognitive Actions API
}

# Prepare the request body for the hypothetical execution endpoint
request_body = {
    "action_id": action_id,
    "inputs": payload
}

print(f"--- Calling Cognitive Action: {action.name or action_id} ---")
print(f"Endpoint: {COGNITIVE_ACTIONS_EXECUTE_URL}")
print(f"Action ID: {action_id}")
print("Payload being sent:")
print(json.dumps(request_body, indent=2))
print("------------------------------------------------")

try:
    response = requests.post(
        COGNITIVE_ACTIONS_EXECUTE_URL,
        headers=headers,
        json=request_body
    )
    response.raise_for_status() # Raise an exception for bad status codes (4xx or 5xx)

    result = response.json()
    print("Action executed successfully. Result:")
    print(json.dumps(result, indent=2))

except requests.exceptions.RequestException as e:
    print(f"Error executing action {action_id}: {e}")
    if e.response is not None:
        print(f"Response status: {e.response.status_code}")
        try:
            print(f"Response body: {e.response.json()}")
        except json.JSONDecodeError:
            print(f"Response body (non-JSON): {e.response.text}")
    print("------------------------------------------------")

Conclusion

The Wan 2.1 1.3b Vid2vid model revolutionizes the way developers can approach video content creation. By integrating this powerful action, you can streamline your video production process while enhancing the quality and appeal of your outputs. Whether for creative projects, marketing efforts, or educational purposes, the possibilities are vast. Start exploring the potential of AI-driven video generation to elevate your projects to new heights!

← Back to All Articles