Video Generation
Generate high-quality videos using state-of-the-art AI models from OpenAI, Google, Runway, and more
Overview
Wazza Engine provides access to 16 video generation models across 6 providers:
OpenAI (Sora)
Sora 2 and Sora 2 Pro - cinematic text-to-video
Pricing: 10-15 credits per generation
Google Gemini (Veo)
Veo 3 and 3.1 models with fast variants
Pricing: 8-12 credits per generation
Runway
Gen-4 Turbo and Gen-4 Aleph
Pricing: 10-25 credits per generation
Qwen (WAN)
7 models including T2V, I2V, and editing
Pricing: 5-15 credits per generation
ByteDance (Seedance)
Seedance 1.0 Pro, Lite T2V, and Lite I2V
Pricing: 8-20 credits per generation
KlingAI
Kling v1.6, v2.1, and v2.1 Master
Pricing: 8-15 credits per generation
Text-to-Video Generation
Generate videos from text descriptions using the Chat Orchestrator:
import WazzaEngine from '@wazza/engine';
const wazza = new WazzaEngine({
apiKey: process.env.WAZZA_API_KEY
});
// AI automatically chooses the best model (Sora, Veo, etc.)
const response = await wazza.chat({
message: "Create a cinematic video of a drone flying over a tropical beach at sunset",
parseIntent: true
});
console.log('Job ID:', response.jobId);
// Wait for completion
const result = await wazza.waitForCompletion(response.jobId);
console.log('Video URL:', result.output.url);Sora 2 Pro Example
// Using Sora 2 Pro for highest quality
const response = await wazza.generate({
provider: 'openai',
model: 'sora-2-pro',
prompt: 'A time-lapse of a blooming flower garden through the seasons',
parameters: {
duration: 10, // seconds
resolution: '1920x1080',
fps: 30
}
});Veo 3.1 Example
// Using Google Veo 3.1 Fast for quick generation
const response = await wazza.generate({
provider: 'google-gemini',
model: 'veo-3.1-fast',
prompt: 'A futuristic city with flying cars and neon lights',
parameters: {
duration: 5,
aspectRatio: '16:9'
}
});Image-to-Video Generation
Animate static images using WAN, Seedance, Runway, or KlingAI:
// Image-to-video with Runway Gen-4
const response = await wazza.generate({
provider: 'runway',
model: 'gen4_turbo',
prompt: 'The person turns and walks away',
parameters: {
image: 'https://example.com/portrait.jpg',
duration: 5,
motion: 'medium'
}
});
// Image-to-video with KlingAI
const response2 = await wazza.generate({
provider: 'kling-ai',
model: 'kling-v2-1-master',
prompt: 'Camera slowly zooms in',
parameters: {
image: 'https://example.com/landscape.jpg',
duration: 8
}
});Video Editing
Edit existing videos using WAN or Runway models:
// Video editing with WAN VACE
const response = await wazza.generate({
provider: 'qwen',
model: 'wan2.1-vace-plus',
prompt: 'Change the weather to snow',
parameters: {
video: 'https://example.com/original.mp4',
editStrength: 0.7
}
});Lip Sync (Hedra Character-3)
Create talking head videos with synchronized lip movements:
// Lip sync with Hedra Character-3
const response = await wazza.generate({
provider: 'hedra',
model: 'character-3',
parameters: {
image: 'https://example.com/portrait.jpg',
audio: 'https://example.com/speech.mp3',
// Or provide text for TTS
text: 'Hello, welcome to our presentation',
voice: 'en-US-female-1'
}
});Best Practices
1. Model Selection
- Sora 2 Pro: Best for cinematic, high-quality videos (slow)
- Veo 3.1 Fast: Best for quick iterations and prototyping
- Runway Gen-4: Best for image-to-video with precise control
- KlingAI: Best for realistic motion and camera movements
- WAN: Best for video editing and transformations
2. Prompt Engineering
- Include camera movements: "slow zoom", "dolly shot", "pan left"
- Specify motion: "subtle movement", "dynamic action", "static"
- Describe lighting and mood for consistent atmosphere
3. Performance Tips
- Start with shorter durations (5-8 seconds) for testing
- Use "fast" variants for rapid prototyping
- Consider webhooks for long-running generations
4. Quality vs Speed
- Fastest: Veo 3.1 Fast, WAN 2.2 Flash (~2-3 min)
- Balanced: Veo 3.1, Runway Gen-4 Turbo (~5-8 min)
- Highest Quality: Sora 2 Pro, Gen-4 Aleph (~10-15 min)