Topaz Video AI Pro

Overview

Topaz Video AI does one category of thing and does it better than the alternatives: it takes video that is too small, too noisy, too shaky, interlaced or judder ridden, and produces a version that is not. Not by sharpening filters, but by trained models that reconstruct detail from what the source implies rather than merely amplifying what is there.

Enhancement is the headline. Several upscaling models are included, each trained for a different source: clean digital footage, compressed web video, old interlaced broadcast material, film scans and computer generated content. Picking the right model matters more than any slider, and the preview compares models side by side at full resolution on a section of your clip before you commit to an export.

Beyond enlargement the tool set covers the whole restoration problem. Denoise separates grain and compression noise from real detail. Deinterlace reconstructs progressive frames from interlaced fields properly rather than blending them. Stabilisation smooths handheld motion with rolling shutter correction. Frame interpolation generates intermediate frames for slow motion or for converting frame rates without judder.

Processing is GPU driven and this build carries all the models locally, so nothing is fetched at render time. Exports run through a queue, so a batch of clips can be left overnight, and the output goes to the usual delivery formats including ProRes and lossless intermediates for anything heading back into an edit.

Key Features

• Multiple upscaling models trained for different source types, up to 16K output
• Side by side model comparison at full resolution before committing to an export
• Denoise that separates grain and compression artefacts from genuine detail
• Deinterlacing that reconstructs progressive frames rather than blending fields
• Stabilisation with rolling shutter correction and adjustable smoothness
• Frame interpolation for slow motion and frame rate conversion without judder
• Batch queue for processing a set of clips unattended
• GPU accelerated processing on NVIDIA, AMD and Intel hardware
• All models included locally with nothing fetched at render time
• Output to ProRes, H.264, HEVC, image sequences and lossless intermediates

What Changed in This Build

• Improved detail retention on heavily compressed web sources
• Faster processing on current generation GPUs
• Better face handling in upscaled archive footage
• Queue reliability improved on long overnight batches

System Requirements

Processor	Intel or AMD 64-bit, 6 cores or more recommended
Memory	16 GB minimum, 32 GB recommended
Graphics	GPU with 6 GB VRAM minimum, 12 GB or more for 4K and above
Storage	20 GB for install with all models, plus output space
Display	1920 x 1080 or higher

Release Details

Full title	Topaz Video AI Pro
Version string	7.1.6
Publisher	Topaz Labs
Category	Video Editing
Licence	Full version, all models included
Platform	Windows 10 and 11, 64-bit
Interface language	English
Archive size	2.86 GB
Mirrors	4 active, all reporting online
Listed	refreshed 1 month ago, checked again 40 minutes ago

Installation Notes

• Test models on a short representative section before queueing a long clip, model choice affects the result more than any parameter.
• Export to a lossless or ProRes intermediate if the result is going back into an edit, re compressing an enhanced file undoes some of the gain.
• Processing is slow by nature. Plan for real time multiples rather than expecting a fast pass.

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