Pixelmator Pro now includes the instrument that has gone down in history to solve most of the cases in the B series investigative films and TV series, or the resolution increase of an image. Taking advantage of a machine learning model, the photo editor for Mac users is able to enlarge photos without drastic quality losses and details, and without the introduction of unpleasant artifacts like blurs and grainy parts. Following some comparisons with traditional enlargement methods (on the official announcement post they are interactive; just follow the FONTE link at the bottom of the article). For reference: Nearest Neighbor it is the simplest algorithm that produces the worst results, while Lanczos it is the best and most advanced. bilinear is in the middle.
All processing operations are performed locally, which means that on some Macs it may even take a few minutes to process a single image. On the latest generation of hardware, which is better optimized for AI functions, processing times fall to just a few seconds. If there are more GPUs (like on one of the more extreme configurations of the new Mac Pro) or external GPU, the wait is very close to zero. This small table summarizes effectively the abyss, in terms of the required computing power that separates traditional systems from those that use machine learning:
This instead illustrates the waiting times on a 300,000 pixel image and in the case of a 300% zoom.
The new function, whose precise name is ML Super Resolution, is available with the update to version 1.5.4, available recently in the App Store.