Generate

​

Key Features

• Text-to-Video Generation: Create videos from detailed text descriptions
• Video Style Transfer: Apply artistic styles and effects to existing videos
• Object Removal & Replacement: Seamlessly remove or replace objects in videos
• Temporal Editing: Manipulate time, speed, and sequence in videos
• High Resolution Output: Generate videos up to 4K resolution
• Custom Model Training: Fine-tune models for specific use cases and styles

​

Response Format

{
  "generated_video_url": "https://api.vlmrun.com/generated/vid_12345.mp4",
  "generation_type": "text_to_video",
  "prompt": "A serene mountain landscape with flowing clouds",
  "style": "cinematic",
  "resolution": "1920x1080",
  "duration": 30,
  "fps": 30,
  "quality": "high",
  "generation_time": "5m 30s",
  "model_version": "v2.1",
  "metadata": {
    "frames_generated": 900,
    "guidance_scale": 7.5,
    "motion_consistency": 0.89
  }
}

​

Generation Types

​

Text-to-Video

• Natural Language Prompts: Describe desired videos in plain English
• Scene Descriptions: Include setting, characters, actions, and camera movements
• Style Specifications: Specify visual styles, lighting, and mood
• Technical Parameters: Control resolution, duration, frame rate, and quality

​

Style Transfer

• Artistic Styles: Apply painting styles like Van Gogh, Picasso, or Monet
• Photographic Styles: Mimic specific camera techniques or film types
• Animation Styles: Convert to cartoon, anime, or other animated styles
• Custom Styles: Use your own reference videos for style transfer

​

Object Manipulation

• Object Removal: Remove unwanted objects from videos
• Object Replacement: Replace objects with different ones
• Background Replacement: Change or enhance video backgrounds
• Content Addition: Add new elements to existing videos

​

Temporal Editing

• Speed Control: Change video playback speed (slow motion, fast forward)
• Time Manipulation: Reverse, loop, or reorder video segments
• Transition Effects: Add smooth transitions between scenes
• Duration Adjustment: Extend or shorten video content

​

Supported Styles

​

Realistic

• Photorealistic: Ultra-realistic videos with fine details
• Natural Motion: Realistic movement and physics
• Professional Quality: Suitable for commercial and professional use
• Multiple Subjects: People, objects, landscapes, architecture

​

Cinematic

• Film Quality: High-end cinematic appearance
• Dramatic Lighting: Professional lighting and color grading
• Camera Movement: Smooth, cinematic camera work
• Narrative Focus: Story-driven visual composition

​

Artistic Styles

• Watercolor: Soft, flowing watercolor painting effects
• Oil Painting: Rich, textured oil painting appearance
• Sketch: Pencil or charcoal sketch style
• Digital Art: Modern digital art and illustration styles

​

Creative Styles

• Cyberpunk: Futuristic, neon-lit urban environments
• Fantasy: Magical, mystical, and fantasy-themed imagery
• Abstract: Non-representational, artistic compositions
• Minimalist: Clean, simple, and focused designs

​

Advanced Features

​

Motion Consistency

• Smooth Movement: Ensure consistent motion throughout the video
• Temporal Coherence: Maintain visual consistency across frames
• Physics Simulation: Realistic physics and motion patterns
• Camera Work: Professional camera movement and framing

​

Multi-Modal Generation

• Audio Integration: Generate videos with synchronized audio
• Text Overlay: Add text and captions to generated videos
• Multiple Scenes: Create videos with multiple connected scenes
• Interactive Elements: Generate videos with interactive components

​

Custom Training

• Domain-Specific: Train models for specific industries or use cases
• Style Learning: Learn from your own video collections
• Character Consistency: Maintain consistent characters across videos
• Brand Adaptation: Adapt to specific brand styles and guidelines

​

Prompt Engineering Tips

​

Effective Video Prompts

• Scene Description: Clearly describe the setting, characters, and actions
• Camera Movement: Specify camera angles, movements, and framing
• Lighting & Mood: Include lighting conditions and emotional tone
• Technical Details: Specify resolution, duration, and quality requirements

​

Example Prompts

• Nature: “A peaceful forest stream with sunlight filtering through trees, gentle camera pan, golden hour lighting, 4K, 20 seconds”
• Urban: “Busy city street at night with neon lights, time-lapse traffic, cinematic wide shot, cyberpunk style, 30 seconds”
• Character: “A person walking through a modern art gallery, tracking shot, natural lighting, documentary style, 15 seconds”
• Abstract: “Flowing geometric shapes in space, smooth camera movement, vibrant colors, artistic style, 25 seconds”

​

Execute Agent

from pathlib import Path
from vlmrun.client import VLMRun

client = VLMRun(base_url="https://agent.vlm.run/v1", api_key="<VLMRUN_API_KEY>")

# Execute video generation
response = client.agent.execute(
    inputs={},
    config={
        "prompt": "A serene mountain landscape with flowing clouds, cinematic wide shot, golden hour lighting, 4K, 20 seconds",
        "capability": "video_generation",
        "style": "cinematic",
        "duration": 20,
        "resolution": "1920x1080",
        "fps": 30
    }
)

# Poll for results
import time
while True:
    execution = client.agent.executions.get(execution_id=response.execution_id)
    if execution.status == "completed":
        print(execution.response)
        break
    elif execution.status == "failed":
        print(f"Error: {execution.error}")
        break
    time.sleep(2)

​

Best Practices

• Clear Descriptions: Use specific, descriptive language in prompts
• Appropriate Duration: Keep generated videos within reasonable length limits
• Quality Settings: Use appropriate quality settings for intended use
• Iterative Refinement: Generate multiple variations and refine prompts