Topaz Sharpen AI

Overview

Conventional sharpening does one thing: it increases local contrast at edges. That makes an image look crisper at small sizes and produces bright halos and crunchy noise at large ones, and it cannot do anything at all about a frame that was genuinely out of focus or moving. Sharpen AI takes a different approach, using trained models that attempt to reverse the specific blur that occurred rather than exaggerating what survived it.

Three model families cover three different problems. The motion model handles subject or camera movement during exposure, reconstructing the direction and length of the smear. The focus model handles frames where the plane of focus fell slightly wrong, which is the shot that used to be unrecoverable. The general model handles the softness that comes from a lens wide open, a cheap filter, diffraction at small apertures or an aggressive denoise pass earlier in the chain.

Control is deliberate rather than automatic. Model selection can be left on auto detect or forced, strength and detail sliders adjust how far the reconstruction is pushed, and a masking brush restricts the effect to the subject so a background stays as it was. A split view and a synchronised comparison mode show original and result at pixel level, which is the only honest way to judge whether a sharpening pass has helped or invented detail.

It runs as a standalone application and as a plugin inside the common editors, taking a layer or a smart object and returning it processed, which keeps the result non destructive in a larger retouching stack. Batch processing takes an entire folder with one setting, and the processing pass is GPU accelerated where a suitable card is present, which turns a per image wait into something practical across a shoot.

Key Features

• Separate model families for motion blur, focus error and general softness
• Automatic model detection with manual override when the guess is wrong
• Strength and detail control over how far the reconstruction is pushed
• Masking brush to restrict sharpening to the subject and leave the background alone
• Split view and synchronised pixel level comparison against the original
• Standalone application and plugin operation inside the common editors
• Batch processing across a folder with a single setting
• GPU accelerated processing where a suitable card is present
• Raw, TIFF, PNG and JPEG input with full bit depth preserved on output
• Non destructive round trip as a layer or smart object in a host editor

What Changed in This Build

• Models retrained for cleaner results on high ISO frames
• Faster processing on current generation graphics hardware
• Masking brush gains edge awareness for quicker subject isolation
• Batch queue reports per image progress rather than only overall

System Requirements

Processor	Intel or AMD 64-bit with AVX2 support
Memory	16 GB recommended, 8 GB minimum
Graphics	GPU with 4 GB VRAM strongly recommended, 6 GB or more for large files
Storage	4 GB free for install and model data
Display	1920 x 1080 or higher

Release Details

Full title	Topaz Sharpen AI
Version string	4.1.0
Publisher	Topaz Labs
Category	Photo Editing
Licence	Full version, no term limit
Platform	Windows 10 and 11, 64-bit
Interface language	English and six additional interface languages
Archive size	1.42 GB
Mirrors	4 active, all reporting online
Listed	refreshed 1 week ago, checked again 40 minutes ago

Installation Notes

• Sharpen last in the chain, after denoise and after any resize. Sharpening first amplifies everything you were about to remove.
• Judge results at one hundred percent, not fitted to the window. Reconstruction artefacts are invisible at small sizes and obvious in print.
• Model files download on first run of each model family. Open each one once while you still have a connection.

Before You Start

Read the notes above before running setup. Most of the problems people write in about are covered there, and almost all of them come down to an older version of the same software still being installed, a graphics driver that predates the release, or a security suite quarantining part of the package mid install.

There is no archive password on this or any other entry in the library. If something you downloaded elsewhere claims to come from here and asks for one, it did not come from here. Nothing on this site is behind a survey, a shortener or a countdown either.

If the entry does not match what the page describes, say so on the contact page and include the version string from the release details table. Reports naming a specific version get checked first because the checker can reproduce them straight away.

Frequently Asked