Topaz Photo AI

About Topaz Photo AI

This is the combined application, where the separate denoise, sharpen and enlargement tools were folded into one pass. The reason that matters is ordering. Applied in the wrong sequence those operations fight each other, sharpening amplifies noise that denoise then has to remove, and the result is worse than either applied alone. Here the pipeline handles the order internally.

On opening an image it runs an assessment. It measures noise, detects the type of blur present, checks resolution against what the subject needs and looks for faces, then enables the relevant modules with starting values already set. On a great many photographs that first suggestion is simply correct, and the whole edit becomes a look, a confirm and an export.

Each module is still fully adjustable when it is not. Denoise offers several models for different sensor and lighting conditions, sharpening distinguishes between motion blur, focus miss and general softness because each requires a different reconstruction, enlargement handles resolution, and the face recovery pass rebuilds facial detail on subjects too small in frame for the general model to resolve convincingly.

The working side matters as much as the results. Batch processing runs a folder unattended with either the per image assessment or a fixed setting applied throughout, camera raw files are read directly so the work happens before any rendering decisions are baked in, and the plugin bridge sends a frame from a host editor and returns it in place with the colour profile and metadata intact.

What is included in this build

• Combined denoise, sharpen, enlarge and face recovery in one correctly ordered pass
• Automatic assessment enabling the right modules with sensible starting values
• Multiple denoise models for different sensor and lighting conditions
• Sharpening that distinguishes motion blur, focus miss and general softness
• Face recovery rebuilding detail on subjects too small in frame
• Camera raw files read directly, before rendering decisions are baked in
• Batch processing across a folder with per image or fixed settings
• Plugin bridge returning a processed frame to a host editor in place

Changed in version 4.1.0

• Improved assessment accuracy on mixed lighting and high contrast scenes
• New denoise model for very high sensitivity frames
• Faster processing on current generation graphics cards
• Face recovery handles profile and partially obscured subjects better
• Reduced memory use during large batch runs

Installation walkthrough

These steps assume a clean machine. If a previous release of the same application is present, deal with that first, since an overlapping install is the most common cause of a setup that stops halfway.

1. Update the graphics driver first, since processing leans heavily on the graphics card.
2. Disable real-time protection until the install has completed.
3. Extract the archive to a short folder path and run setup as administrator.
4. Allow the model files to finish installing before launching, as they are most of the package.
5. Set the output folder, format and default behaviour in preferences before running a batch.

System requirements and file details

Frequently asked about this title