What is Inpainting?
An AI technique that fills in or modifies a masked region of an image while preserving the unmasked area. The standard method for AI-healing seamless tile boundaries.
In detail
Inpainting takes three inputs: an image, a binary mask defining which pixels to edit, and (optionally) a text prompt describing the desired output. The diffusion model runs its denoising process only on the masked region while using the unmasked pixels as fixed context. The output is the original image with the masked region replaced by AI-generated content that visually continues the surrounding patterns. For seamless tile creation, inpainting is used in the offset-and-inpaint workflow: shift the tile so the seam moves to the center, mask the seam region, run masked inpainting to heal it, un-shift to put the heal back at the edge. This produces a seamless tile from any starting image. Modern textile AI platforms use SDXL Inpaint or FLUX Fill for this workflow. Inpainting quality scales with mask precision: a tight mask (covering only the seam strip) preserves more of the original art but limits the model's ability to redistribute motifs across the seam; a wider mask (covering 20% of the tile) gives the model more freedom but may reinvent details the designer wanted preserved. Texloom defaults to a 12-pixel feathered mask for the offset-and-inpaint workflow.
Example
A user has a hand-painted floral with motifs running off the edges. The seam between left and right edges does not match. Texloom's pipeline: shift image by W/2 horizontally → mask center vertical band → SDXL Inpaint heals the band → un-shift. The tile now has invisible seams.