Skip to content

TextureFlow#

Create dynamic video animations by mixing multiple style images with fluid motion patterns. TextureFlow transforms static images into mesmerizing animations with optional shape guidance and controllable motion patterns.

Overview#

  • Mix 1-6 style images into smooth video animations
  • Multiple motion mapping modes (circles, rectangles, stripes)
  • Optional shape/depth guidance via ControlNet
  • Upscaling to HD resolution
  • Adjustable motion strength and transition effects

Quick Start#

  1. Visit Eden Create: TextureFlow tool
  2. Upload 1-6 style images
  3. Set video duration and dimensions
  4. Select motion mapping mode
  5. Click "Create"

See API Reference section below for API usage examples.

Ask an Eden agent to "TextureFlow".

Basic Usage#

Required inputs:

  • 1-6 style images
  • Video dimensions (384-1024px)
  • Duration (2-24 seconds)

Optional settings:

  • Motion mapping mode
  • Shape guidance input
  • Motion strength
  • Boundary softness

Technical Guidelines#

Resolution Options#

  • Width: 384-1024px
  • Height: 384-1024px
  • Common aspects: 16:9, 3:2, 1:1, 9:16

Generation Settings#

  • Video length: 2-24 seconds
  • Steps: 4-14 (default: 8)
  • Motion scale: 0.7-1.4 (default: 1.1)
  • Boundary softness: 0.0-0.25 (default: 0.1)

Best Practices for Inputs#

Style Images#

  • Use 3 seconds of animation per style image
  • Choose visually compatible images for smooth transitions
  • Consider color harmony between images
  • Avoid extremely busy or detailed images

Shape Guidance#

  • Use simple, clear shapes for control inputs
  • Keep edges and contours well-defined
  • Match control input aspect ratio to output dimensions
  • Test with lower controlnet strength first (0.4-0.6)

Advanced Features#

Motion Mapping Modes#

  • Concentric patterns (circles/rectangles inward/outward)
  • Rotating segments (clockwise/counter-clockwise)
  • Pushing segments (clockwise/counter-clockwise)
  • Directional stripes (vertical/horizontal)

Shape Guidance Options#

  • Edges (Canny)
  • Depth mapping
  • Line art
  • Human pose
  • Scribble lines
  • Luminance (QR-code regions)

Technical Details#

Base technology:

  • LCM denoising
  • ControlNet v1.1
  • IP-Adapter for style transfer

Supported formats:

  • Input: JPG, PNG
  • Output: MP4 video
  • Optional control input: Image, GIF, video

API Reference#

Endpoint#

curl -X POST "https://api.eden.art/v2/tasks/create" \
  -H "Content-Type: application/json" \
  -H "X-Api-Key: YOUR_API_KEY" \
  -d '{
    "tool": "texture_flow",
    "args": {
      "images": [
        "https://storage.googleapis.com/public-assets-xander/A_workbox/init_imgs/img_00019.jpg",
        "https://storage.googleapis.com/public-assets-xander/A_workbox/init_imgs/img_00021.jpg",
        "https://storage.googleapis.com/public-assets-xander/A_workbox/init_imgs/img_00024.jpg",
        "https://storage.googleapis.com/public-assets-xander/A_workbox/init_imgs/img_00025.jpg"
      ],
      "n_seconds": 4,
      "n_steps": 6,
      "width": 640,
      "height": 640,
      "use_controlnet1": true,
      "control_input": "https://storage.googleapis.com/public-assets-xander/A_workbox/lora_training_sets/xander.jpg",
      "preprocessor1": "DepthAnythingV2Preprocessor",
      "controlnet_strength1": 0.5
    }
  }'

Response Format#

{
  "task": {
    "_id": "task_12345",
    "createdAt": "2024-01-29T21:52:07.171578+00:00",
    "updatedAt": "2024-01-29T21:52:07.171608+00:00",
    "user": "user_id",
    "requester": "requester_id",
    "tool": "texture_flow",
    "parent_tool": null,
    "output_type": "video",
    "args": {
      "images": ["https://example.com/image1.jpg"],
      "n_seconds": 4.0,
      "n_steps": 6,
      "width": 640,
      "height": 640,
      "use_controlnet1": true,
      "control_input": "https://example.com/control.jpg",
      "preprocessor1": "DepthAnythingV2Preprocessor",
      "controlnet_strength1": 0.5
    },
    "mock": false,
    "cost": 48,
    "handler_id": "fc-xxxxx",
    "status": "pending",
    "error": null,
    "result": null,
    "performance": null
  }
}

Common Issues and Solutions#

  1. Duplication artifacts

  2. Reduce output dimensions

  3. Lower motion scale

  4. Simplify input images

  5. Poor style mixing

  6. Adjust boundary softness

  7. Use more compatible style images

  8. Increase steps for smoother blending

  9. Weak shape guidance

  10. Increase controlnet strength

  11. Try different preprocessor types

  12. Ensure clear control input

  13. Motion artifacts

  14. Reduce motion scale
  15. Increase steps
  16. Lower video duration