Skip to main content
OpenComic AI Bin provides a comprehensive library of AI models for image processing, specifically optimized for comic book and manga content. All models are automatically downloaded when needed.

​
Model types

The library organizes models into three main categories:

​
Upscale models

Upscaling models increase image resolution while preserving or enhancing quality. These models can scale images by 2x, 3x, 4x, or higher factors depending on the specific model. Learn more about upscale models

​
Descreen models

Descreening (also called dehalftoning) models remove halftone patterns and dot screens commonly found in printed comics and manga. These models operate at 1x scale and are typically applied before upscaling. Learn more about descreen models

​
Artifact removal models

Artifact removal models clean up JPEG compression artifacts, blur, and other image degradation. These models are essential for restoring quality to compressed comic scans. Learn more about artifact removal models

​
Model organization

Models are accessed through the OpenComicAI.models object, which is organized by type: 
OpenComicAI.models = {
  upscale: { /* upscaling models */ },
  descreen: { /* descreening models */ },
  'artifact-removal': { /* artifact removal models */ }
}
You can also get a flat array of all available model names: 
const allModels = OpenComicAI.modelsList;

​
Automatic model downloading

Model files are not included with the package. Instead, they are downloaded automatically when first used:
  1. Set models path: Configure where models should be stored
  2. Use a model: When you first reference a model in your pipeline
  3. Automatic download: The library downloads the required model files
  4. Local caching: Models are saved for future use
// Set the directory for model storage
OpenComicAI.setModelsPath('./models');

// Use a model - it will download automatically if not found
await OpenComicAI.pipeline('./input.jpg', './output.jpg', [
  { model: 'realcugan', scale: 4, noise: 0 }
], progressCallback, {
  start: () => console.log('Starting download'),
  progress: (p) => console.log(`Downloading: ${Math.round(p * 100)}%`),
  end: () => console.log('Download complete')
});
If you prefer to install all models upfront, use the opencomic-ai-models package.

​
Upscaler types

Models use one of three upscaler engines:

realcugan

Advanced upscaling with noise reduction controls. Supports scales 2x, 3x, 4x with noise levels 0-3.

waifu2x

Flexible upscaling supporting 2x, 4x, 8x, 16x, 32x scales with noise levels 0-3.

upscayl

Fast, general-purpose upscaling. Most models support 2x, 3x, 4x scales. Used by most models in the library.

​
Model properties

Each model in the library includes the following properties:
PropertyDescription
nameHuman-readable model name
upscalerEngine type: realcugan, waifu2x, or upscayl
scalesArray of supported scale factors
noiseNoise reduction levels (or undefined if not supported)
latencyAverage processing time in seconds for a single image
folderDirectory containing model files
filesArray of model file names

​
Performance considerations

Model performance varies significantly. The library includes latency data for each model to help you choose the right balance between speed and quality. View performance comparison

​
Next steps

Upscale models

Explore 30+ upscaling models

Descreen models

Remove halftone patterns

Artifact removal

Clean up compression artifacts

Performance

Compare model speeds

Build docs developers (and LLMs) love

Get started for freeTalk to us