About this unblur image
Unblur Image estimates what your photo looked like before it was blurred. It models the blur as a spread of light (a Gaussian point spread function) and runs Richardson-Lucy deconvolution, the method used in astronomy and microscopy, which repeatedly re-blurs its estimate and corrects it until it matches your photo. Blur size should match how far edges are smeared, Passes trades detail against noise, and Noise suppression keeps flat areas smooth. It works on brightness only, so colours do not fringe, and runs at full resolution in a background thread of your browser. It recovers mild focus and shake blur well; it cannot invent detail that was never captured, such as a face a few pixels wide.