Upscayl

Upscayl Desktop is a free, open-source image upscaler that processes images on your own computer. It uses AI models to enlarge low-resolution or pixelated pictures, with single-image and batch workflows. This cask installs the desktop Mac application, not the separate Upscayl Cloud service.

Features

• Select models suited to photos or other image types, including High Fidelity in 2.15.
• Process one image or a folder of images.
• Choose output location, scale or custom width and supported output formats.
• Compare original and processed images with slider and lens views.
• Use Double Upscayl, configurable tile size and compatible custom models.
• Choose a GPU, inspect logs and system information, and adjust themes and language.

Install and use

Install with brew install --cask upscayl, or download the official Mac DMG and move Upscayl.app into Applications. Select an image, choose a model, choose an output folder and start Upscayl. For batches, first test a representative image and inspect output quality and storage use. Allow access to the input and output folders you intentionally select; ordinary local image processing does not require a cloud account.

Keep original files and use a separate output directory. Large scales and Double Upscayl increase processing time, GPU memory needs and disk usage. Reduce tile size or output dimensions if your GPU runs out of memory, and consult official troubleshooting for device-specific problems. The cask lists no additional formula dependency or explicit conflicting cask. Removing application data separately can remove settings and logs, so review cleanup choices before doing so.

GPU and Mac compatibility

The project requires a Vulkan-compatible GPU and warns that many integrated GPUs and CPUs do not work. It does not promise a general CPU fallback. The Mac backend's official build instructions use MoltenVK and link Apple's Metal framework; this is the Mac implementation of the Vulkan backend, not a requirement to buy an NVIDIA CUDA GPU. Intel and Apple Silicon backend build targets are documented, but GPU and driver compatibility still matter.

The captured cask and official README require macOS 12 or later. The 2.15 release changed the Mac build environment to macOS 13 after GitHub retired Big Sur runners; a build-runner version is not itself a new stated runtime minimum. Do not infer support for Big Sur or every older Intel graphics device. Keep macOS current and test a small input on the intended Mac before relying on a large job. Windows and Linux have separate packages and driver considerations.

Quality and privacy

AI models estimate plausible detail rather than recover an authoritative original. Upscayl's FAQ says it cannot correct an out-of-focus or completely blurred photo. Inspect faces, text, edges and textures, especially for archival or evidence-related images. In 2.11, scales above 4× use a 4× base upscale followed by resizing; a 16× output is not a guarantee of 16× new authentic detail.

Desktop image processing is local and can work without internet once the application and models are available. Downloads, updates, optional custom-model retrieval and Cloud use require connectivity. Version 2.15 adds an optional statistics-sharing control and an option to disable automatic update checks; local processing should not be described as a guarantee that every optional network feature is disabled. Review settings and logs before sharing them with support.

Desktop, App Store and Cloud

The GitHub desktop download used by this cask is free and does not require a paid subscription or account for local upscaling. The separate Mac App Store distribution can be paid; its store price is not a mandatory license fee for the GitHub cask. Upscayl Cloud is a distinct hosted service that requires an account and uploads images for remote processing. Its official pricing offers credit-based purchases/subscriptions and a login-based trial. Cloud storage, limits, features and terms differ from Desktop, and are not included merely by installing this cask. Check current Cloud pricing and privacy terms before uploading sensitive images.

License and download

Upscayl Desktop and its backend are open source under AGPLv3; bundled models and third-party components retain applicable licenses. Review those terms when modifying or redistributing the software. Image-processing software does not grant rights to use or redistribute other people's images. Donations are optional.

The official 2.15.0 Mac DMG is 385,264,901 bytes, confirmed by HTTP metadata without downloading the installer. Installed storage and generated output can be larger. Upscayl also offers Windows and Linux builds; this entry describes the Mac cask.

Sources: Official project and FAQ, Desktop download, Cloud pricing, Mac backend build instructions, Official releases.