Topaz Video AI 1.7.2 (macOS)

Poor source footage — noisy, soft, or stuck at a low resolution — is the usual reason people reach for Topaz Video AI.

Upscaling — enlarges video, from SD to HD or HD up to 8K, using AI models to keep motion consistent.

Denoising — removes grain and compression artefacts from DVD, DSLR, game, and stock footage while retaining detail.

Sharpening — restores focus and edge definition on soft clips.

Stabilisation — reduces shake in handheld or action footage.

Version 1.7.1 runs on macOS 11 and later, on Intel and Apple Silicon. The interface is in English. What changed in 1.7.1 is nothing documented — the focus is the existing model set and the desktop workflow.

Peg your output to a fast, cool external drive

Topaz Video AI will peg the GPU for long stretches. That is fine — it is what the app is for — but a few bits of housekeeping stop it becoming a nuisance.

• Set an output location on a fast drive. The default is often the same folder as the source. Point it to an external SSD or a secondary internal volume with plenty of free space.
• Trim before processing. Even a ten-second clip is enough for a model comparison. Processing a full hour-long file just to test Proteus versus Artemis will waste an afternoon.
• Turn off unnecessary background tasks. Spotlight indexing a new external drive while Topaz is rendering will slow both. Same for Dropbox or OneDrive sync.
• Check the preview resolution. The preview window defaults to a lower scale. Bump it up for the final check of a short test segment.
• Let the models download once. First time you pick a model, Topaz fetches it. That can take a minute or two depending on your connection. Do a throwaway export with each model you expect to use so the files are local when you need them.

workflow: restoring an old DVD rip to HD

This is one of the jobs Topaz Video AI gets handed most often. The source is an MPEG-2 rip from a DVD — interlaced, soft, maybe a little block noise in dark scenes. The target is a clean 1080p file.

1. Open the clip. Drag the file in and let the preview render a few frames.
2. Set the input type. In the right-hand panel, tell the app the footage is interlaced and pick the field order (Top Field First for most PAL discs). Skip this and you get combing baked into the output.
3. Pick a model. Artemis handles most DVD footage well — it sharpens without ringing and cleans compression noise. If the source is very noisy, try Proteus and dial the denoise slider up a notch.
4. Set output to 1080p. Choose ProRes 422 as the codec for an edit-friendly master, or H.265 for a compressed archive.
5. Run a ten-second test. Grab a section with a mix of motion, a face, and a dark area. Check for smearing, over-sharpening halos, or lost grain that makes skin look plastic.
6. Adjust and process the full clip. Once the model settings look right, queue the whole file and walk away. An hour of DVD footage can take several hours to process, even on an M-series Mac.

scenario: a music video editor with noisy DSLR footage

Someone cutting a low-budget music video shot on an older DSLR at ISO 3200. The footage has chroma noise in the shadows and a slight softness from a kit lens shot wide open. The grade is already done, but the noise pulls attention.

They bring the locked cut into Topaz Video AI as a single ProRes file. Proteus is the model to reach for here — it separates luma and chroma noise well. They set the output to match the timeline resolution, 4K, and let the app run overnight. The next morning, the noise is gone without the waxy skin you get from a heavy temporal denoiser in an NLE. The editor drops the cleaned file back into the timeline, checks sync at a few cut points, and renders the final deliverable. No re-grade needed.

The same person might later search for a topaz video ai mac torrent when a colleague asks where they got the tool — the results speak for themselves once a clip is side-by-side.

Stalled queues from mismatched frame rates

Patch.command throws an error on newer macOS

On recent builds of macOS the Patch.command script can refuse to open or throw a permissions error. The fix is straightforward: open Terminal, drag the Patch.command file into the window, and press Return. If that still fails, run chmod +x on the file first and try again. The script needs to make changes to the app bundle, and macOS Gatekeeper will block that unless you explicitly allow it.

export fails partway through a long clip

This usually means the destination drive ran out of space, or the file system cannot handle the file size. Check that the output volume is APFS or exFAT — FAT32 tops out at 4 GB per file, which a 4K export will smash through in minutes. Switch to ProRes LT or H.265 if space is tight, or point the output to a drive with more headroom.

preview looks good but final export is soft

The preview window rescales. If you judged sharpness at a reduced preview size, you missed the true result. Always check a short test export at full resolution before queuing the whole file. Also confirm that the output resolution is not accidentally set to a lower value than the source — easy to do when switching between presets.

stabilisation introduces warping at the edges

Topaz Video AI stabilises by analysing motion and counter-moving the frame. Heavy shakes can push it past what it can fix without visible warping. Dial back the stabilisation strength or crop in slightly tighter to give the algorithm more room. If the shot is a lost cause, stabilise it in a dedicated tracker first and use Topaz for the upscale and cleanup afterwards.

activation not sticking after a restart

If the app asks for activation again, re-run Patch.command. Some system updates or cache-cleaning utilities can strip the patch. Keep the original download handy — a topaz video ai torrent file stashed on a backup drive saves hunting for it again.

• macOS 11 or newer
• Intel or Apple Silicon Mac