The generative AI boom has led to more tools for upscaling and enhancing photos. I’ve been using the free Upscayl app to improve photos and screenshots, and the results have been (mostly) impressive.
Upscaling On Your PC
Upscayl is an image upscaling and enhancement application, with the ability to boost the resolution of images, as well as remove some compression artifacts and noise. You can choose between a few different tuned models, based on the Real-ESRGAN project, with the ability to scale to 1x-16x the original size.
The cool part is that Upscayl runs locally on your own computer, as long as you have a GPU compatible with Vulkan. You definitely don’t need a powerful desktop or laptop, the app will just be a bit slower with underpowered hardware. On my M1 Mac Mini with 16GB RAM (an entry-level Mac from four years ago) with a few other apps running, upscaling a photo from 1024×768 to 4096×3072 took about 36 seconds.
Since Upscayl is running on your own computer, there’s no subscription required to offset server costs. The images should also be staying on your own device, though Upscayl doesn’t seem to guarantee that anywhere in the documentation. Upscayl is working on a cloud version that will work on any device, but it’s not widely available yet. The official documentation mentions that many integrated GPUs are not currently supported, so some laptops and lower-end desktop PCs might not work.
As of the time of writing, Upscayl works on macOS 12 and later, Windows 10 and later, and most desktop Linux distributions. On Linux, you can get it from Flathub, the Snap Store, AppImageHub, and the built-in repositories on some distros.
How It’s Helping Me
Image upscaling can feel like magic sometimes, but there are limitations. It won’t magically turn a dark and grainy photo into a bright and sharp image, and bad smartphone pictures with blown-out highlights will still look bad. It can also sometimes create wavy lines where there should be sharp edges, and some fine detail can be removed entirely (as we’ve seen from some recent upscaled movies on 4K Blu-ray). Still, I have found Upscayl incredibly helpful in some workflows.
Many of the articles I write or edit here at How-To Geek need images provided by companies, but unfortunately, many companies don’t readily provide high-resolution photos. For example, when Raspberry Pi announced its official monitor recently, all the photos on the blog were compressed and low-resolution. I used Upscayl to create a 2x version more suitable for the header images in How-To Geek articles, which improved the resolution and removed the compression artifacts. There’s more distortion when I crank it up to 4x upscaling.
I have also tried using Upscayl for improving the quality of my old digital camera photos, with mixed results. A few photos captured on my old Samsung PL20 14MP camera on a trip to Washington D.C. in 2012 looked sharper after the upscaling, but still had washed-out colors and focus issues.
With photos, Upscayl works best as one step in a larger image editing workflow. For example, I can take my old digital photos, run them through Upscayl to boost the resolution, then import them into Google Photos or Adobe Lightroom to change the color saturation and brightness. Check out the original and edited versions of one of the photos on my Flickr.
I’ve also used Upscayl several times to upscale screenshots, especially when I’m writing guides that need to show buttons and other small UI elements at a larger size. I use the ‘Digital Art’ model option for this use case.
Upscayl and other similar upscaling tools aren’t complete solutions for making low-resolution images look good, but they can be fantastic in certain situations. Hopefully, these tools will keep getting better over time, and more image editing software will build similar features using these models.