Just before the holidays, the Pixelmator Pro team have pushed out another update with an impressive new image resizing algorithm, powered by a machine learning model.
Pixelmator Pro are calling it ‘ML Super Resolution’. It’s a new scaling mode that can allow you to increase images without the blurriness you come to expect when you make a bitmap image larger than its original size.
Traditional scaling algorithms like Nearest Neighbor or Bilinear rely on simple mathematical combinations of nearby pixels. The machine learning model tries to understand the content of the image, which lets it fill in semantic details as the image is resized.
ML Super Resolution is available in the Image Size menu or when using the Transform tool. The tradeoff is that the ML method requires a lot of compute power, specifically dependent on the GPU in your system. This is one area where an eGPU connected to your MacBook Pro will definitely make a speed difference.
You can see more examples of the difference between normal scaling and the ML Super Resolution option in the post on the Pixelmator blog. The improvement in sharpness and apparent retention of detail is easily visible.
Pixelmator Pro 1.5.4 also includes updates to the denoiser, and a smattering of bug fixes. Download Pixelmator Pro in the Mac App Store. (The new 1.5.4 update is still propagating at time of publishing.)
FTC: We use income earning auto affiliate links. More.
Comments