AI Enhance

Doubles the width and height of a photo using an AI model that runs entirely on your own CPU. No account, no credits, no upload, and nothing to download — the model ships inside the installer.

What it accepts. JPG, PNG, WebP, GIF, TIFF and AVIF.

How to use it

1. Open AI Enhance.
2. Drop photos onto Drop photos here or press Choose photos…. Thumbnails appear; hover one to remove it, or press Clear all.
3. Read the time estimate, then press the Upscale button. Each photo becomes its own job.

What to expect

• Upscale 2× (Real-ESRGAN) is the only operation, and 2× is the only scale. The model upsamples by exactly the factor it was trained for, so there is nothing to choose.
• This takes real time. A small image is a few seconds; a full-size phone photo can take several minutes. PhotoTray shows an estimate for the exact photos you have added, before you queue anything, based on their pixel counts and your CPU's core count.
• If a file cannot be measured in the browser (TIFF is the usual case) the panel says so and warns that the real time may be longer.

Output

Saved to the AI Enhance subfolder of your output folder.

If something goes wrong here

• It is taking far longer than the estimate. The estimate assumes the upscaler has your CPU to itself. Other running jobs compete with it — lower Job concurrency in Settings, or upscale fewer photos at once.
• A job failed mentioning a model. The bundled model file is missing or failed its checksum, which means the installation is damaged. Reinstall PhotoTray.
• I want 4× upscaling, or face restoration. Neither ships today. 2× is the trained factor of the bundled model.