The “PatchMatch” method is a randomized algorithm for editing images based on the coherence of an image’s content. Being inclined to care a great deal about the coherence and incoherence of images, and being fairly unsure what exactly makes an image coherent at all, the algorithm seemed like a great object of study. The algorithm is the basis of a more commonly used version, the Content Aware Fill feature introduced in Adobe Photoshop CS5. The tool is generally used to edit out by filling in a selected region with ‘matching’ content. PatchMatch’s basic premise is that some good content matches can be found via random sampling and that ‘natural coherence’ in images can be used to quickly propagate content to surrounding areas. Or put a different way, the utility of this feature is that editing in is the same as editing out.
Drawing In-Painting is a software. Like the Content Aware tool, it doesn’t introduce entirely new things, it just reshuffles existing content. We thought it fitting to use it for a renovation, in this case, an office. To force some comparisons with this method of working, it could be considered a form of painting. Indeed, the larger field of adjacent techniques and tools is called in-painting. Also known as image interpolation, in-painting is used to reconstruct deteriorated parts of images or to paint-in missing pieces based on surrounding imagery of materials. However, it also bears resemblance to what could be considered drawing. Moving a cursor across a screen to select a region of pixels relies on lines to enclose an area. If imaging can be described as a form of drawing with material, then drawing in-painting might serve to give material some form of coherence according to image logic.
Team: Hong Bae Yang and Yiran Chen