Inpaint

Can be one of the following values:

The 'Inpaint Method' property determines which algorithm will be used to restore the masked areas of the image. Each algorithm offers a different approach to estimating the values of unknown pixels:

• Telea: Fast, edge-aware inpainting that propagates pixel values from the boundary inward. Suitable for small or thin regions, such as scratches or lines.
• NavierStokes: Uses a partial differential equation approach to smoothly propagate information from the boundary into the masked region. Often produces more natural results for larger or more complex areas.

Choose the algorithm that best matches the type of restoration required for your documents. For example, use 'Telea' for removing fine lines, or 'NavierStokes' for filling larger damaged areas.

The 'Inpaint Radius' property determines how far the inpainting algorithm will look from each masked pixel to estimate its value. This radius defines the size of the area, centered on each masked pixel, that is used to sample known (unmasked) pixels for restoration.

• Smaller values (e.g., 2px or 3px) restrict the algorithm to using only immediate neighbors, which can help preserve sharp edges and fine details, but may not fully fill larger masked regions.
• Larger values (e.g., 8px or 10px) allow the algorithm to consider a broader area, which can produce smoother, more natural-looking results in large or complex regions, but may introduce blurring or artifacts if set excessively high.

The value is specified as a string with a numeric value and units, such as 3px for 3 pixels. Only pixel units are supported.

Best Practices

• For small marks or thin lines, use a smaller radius (2px–4px) to avoid over-smoothing.
• For larger damaged areas or when a more blended restoration is desired, use a larger radius (6px–10px).
• Test different values on representative images to find the optimal balance between detail preservation and smoothness.

The 'Mask Dilation Factor' property determines how much the mask is expanded (dilated) or contracted (eroded) before being applied to the image. This allows you to fine-tune the dropout effect, making it more or less aggressive depending on your requirements.

• Positive values will dilate the mask, causing the dropout region to expand and potentially remove more content from the image.
• Negative values will erode the mask, shrinking the dropout region and preserving more of the original content.
• A value of zero applies the mask as-is, with no dilation or erosion.

Configuration Guidance

• Increase the value to more effectively drop out the edges of objects or to ensure complete removal of unwanted features.
• Decrease (use negative values) to avoid removing too much content, especially when important features are close to the dropout region.
• Typical values range from -16 to 16, but the optimal setting depends on the specific image and dropout scenario.