Best local, one-time purchase, image upscaler for Mac/Windows?

Upscayl is the obvious open-source pick — Mac, Windows, Linux, runs Real-ESRGAN locally, no account, free. Quality is solid for most use cases. If you need professional-grade results and don't mind paying once, Topaz Gigapixel is the standard. Pricier but still sells a perpetual license, not subscription. For thousands of images batch processing matters more than per-image quality, and Upscayl handles batches cleanly.

try my Lekh AI Pro - one time cost and you get way more things than you bargained for. I have a commercial license for UltraSharp V2 by Kim2091.

❤️ Thanks for being there from the start. I hope you are enjoying the latest updates.

Upscayl s the way to go

‘brew install ImageMagick’ :)

Outside of ComfyUI, or a subscription to Topaz's software, I don't know of anything on a Mac that really compares.

https://www.reddit.com/r/upscaling/

https://www.reddit.com/r/macapps/s/GqsSz7He3v https://www.reddit.com/r/MacOSApps/s/gGISNUDK3X

Upscayl is the one I use. But not a big fan of their new price module.

For thousands of images I’d test the boring part first: batch handling, output naming, crash recovery, and whether it can resume. Upscayl looks like the closest cross-platform local option, but I’d run a mixed 50-image batch before committing. A lot of upscalers look fine one image at a time and then become painful in bulk.

I love upscayl, it's open source and free and the one I personally use.

Topaz Gigapixel

Pixelmator pro's upscaling is pretty decent.

If local-only is a hard requirement, I’d narrow it to tools that can batch process without sending files anywhere. Upscayl is the obvious first test because it’s local and free. It’s not perfect, especially on faces or text, but for a big mixed folder it’s a good baseline before spending money. For paid one-time-ish options, I’d test with a small sample set before buying: portraits, screenshots/text, artwork, and noisy old photos. Upscalers can look great on one category and awful on another. The workflow I’d use for thousands of files: Sort by type first, not just folder name. Try 2x before 4x. Avoid heavy denoise unless the source is genuinely noisy. Export a small batch and inspect at 100%, not preview size. Keep originals untouched, because bad upscaling artifacts are hard to undo. I wouldn’t use a web upscaler for this particular job if privacy/local processing is the main constraint.

Upscayl as others have said

We're currently working on this for our app TidyImage. It's currently not upscaling, but compressing and downscaling images, but adding a local, fast, and actually good batch upscaler is what we're aiming at. What kind of images will you be upscaling? Graphics or Photographs?

This comment was deleted at the source.

Doesn't look so free anymore, which makes sense if they added reverse diffusion or similar generative upscaling: https://upscayl.org/pricing There is an old version here: https://github.com/upscayl/upscayl but I never found it to be amazing. Real-ESRGAN struggles to compete with reverse diffusion generative upscaling. Leonardo AI's newest offering is among the best I've seen, and probably a better option if the only options are subscription. Everything by topaz continues to be sub-par.

The best dedicated image editing software for upscaling include algorithms such as Lanczos, Sinc, and B-spline. But as far as I know, Photoshop itself and Topaz within Photoshop use Ai to generally do a superior job most of the time… I compare and contrast the kind of work I do and the kind of photos that I’ve taken over and art I’ve created over the years. For instance, Ai does a pretty good job on line Art since I am an illustrator and Cartoonist. Of course, a really horrible photo that is not nearly enough pixels is gonna be a real ordeal no matter what choice you make. Maybe another solution is to use an online upscaling tool that employs those algorithms or Ai?