Upscayl 2.15: setup, 4x image upscaling and TTA

This Spanish tutorial covers Upscayl 2.15, an open source image upscaler for Windows, Mac and Linux. The presenter describes processing as local to the computer, with no photo uploads, and says it works offline. The stated hardware requirement is a Vulkan-compatible GPU; support for integrated laptop graphics is uncertain. The metadata identifies the demonstration hardware as an Intel i9-9900K with an NVIDIA RTX 2080.

The Windows walkthrough covers downloading and extracting the application, then choosing startup preferences. Automatic updates and anonymous usage reporting have separate controls. The presenter later disables usage reporting and explains output formats, a saved destination folder and custom output width.

The main example enlarges a 480 × 270 Goku illustration at 4x with the digital art model. The tutorial also explains batch processing and reviews models for natural images, sharper detail and faster processing. It identifies Remacri as unsuitable for commercial use.

For GPU memory management, the presenter recommends automatic tile sizing. Double upscaling carries a warning about processing time and possible performance problems beyond 4x. A second example enables TTA, which the speaker describes as eight varied passes with averaged results. He recommends leaving it off for routine use because of the extra processing time. The image comparisons are the presenter's examples, rather than a measured quality benchmark.