merahburam/vectorizer 🖼️🔢❓✓ → 🖼️

▶️ 310 runs 📅 Aug 2025 ⚙️ Cog 0.16.3

About

Professional PNG to SVG vectorization with VTracer

Example Output

Output

Performance Metrics

0.36s Prediction Time

8.22s Total Time

All Input Parameters

{
  "image": "https://replicate.delivery/pbxt/NZZhdsXLHW1Y9cnIsU9BkESCsVgv2loUiDiQVkm0dmQWvmQW/940-2.png",
  "min_area": 50,
  "threshold": 0.5,
  "vectorizer": "vtracer",
  "color_count": 8,
  "optimize_svg": true,
  "simplify_tolerance": 2
}

Input Parameters

image (required) Type: string: Input PNG/JPG image to vectorize
min_area Type: integerDefault: 50Range: 10 - 1000: Minimum area for shapes (pixels)
threshold Type: numberDefault: 0.5Range: 0 - 1: Threshold for bitmap conversion (Potrace only)
vectorizer Default: vtracer: Vectorization method
color_count Type: integerDefault: 8Range: 2 - 16: Maximum number of colors to extract
optimize_svg Type: booleanDefault: true: Optimize SVG output (remove white backgrounds, simplify paths)
simplify_tolerance Type: numberDefault: 2Range: 0.5 - 10: Path simplification tolerance

Output Schema

Output

Type: string • Format: uri

Example Execution Logs

Starting vectorization with vtracer:
- Color count: 8
- Min area: 50
- Simplify tolerance: 2.0
- Optimize SVG: True
Using VTracer (quality mode) - Rust-powered vectorization...
Processing 475x475 image with VTracer...
VTracer config: {'color_mode': 'color', 'hierarchical': 'stacked', 'mode': 'spline', 'filter_speckle': 2, 'color_precision': 8, 'layer_difference': 8, 'corner_threshold': 30, 'length_threshold': 2.0, 'max_iterations': 20, 'splice_threshold': 30, 'path_precision': 3}
VTracer CLI not found, trying Python API...
VTracer Python API successful! Output size: 252460 chars
Optimizing SVG output...
SVG optimization complete:
Paths: 580 → 580 (0 removed)
Size: 252460 → 257033 bytes (-4573)

Version Details

Version ID: c92fbee02cff6eaed6141dca8aef5358c8338f012973dc83d4245d652739cf29
Version Created: August 22, 2025

Run on Replicate →