Topaz Gigapixel AI — Mac download

Photographers and print designers use Gigapixel AI to salvage low-resolution shots without the fuzzy, over-processed look of old-school upscaling. The desktop app runs AI models locally to enlarge images while adding believable texture. Before you spin it up, confirm it'll run on your machine.

• Supports Apple Silicon and Intel 64-bit Macs
• Requires macOS 12 or newer
• Removes pixelation and compression artifacts
• Restores detail in old, degraded photos
• Cleans up low-quality line art and PNG files
• Rebuilds low-resolution portraits with natural skin texture

The current build provides separate installers for ARM and Intel architectures, so you pick the package that matches your Mac. Version 1.3.5 marks a clean break from the legacy Gigapixel AI line, which ended at 8.4.4.

Dialing Back the Face Recovery to Stop Plastic Skin

Even a straightforward upscaler can trip you up. Here are the issues I see most often.

• Portrait mode over-smooths skin. Gigapixel AI can get overenthusiastic with the face recovery model on heavily compressed JPEGs. Dial the Suppress Noise slider down to 20–30% and switch from High Fidelity to Art & CG if you're working with a studio shot that needs to preserve texture.
• Batch queue stalls on an image. A corrupt or oddly formatted PNG can freeze the whole queue. Open the offending file in Preview, export it as a fresh PNG, and re-add it. If the filename contains special characters like em dashes or colons, simplify it.
• Output file size balloons. Upscaling a 2K JPEG to 6× and saving as PNG can produce a 300 MB file. Set output to JPEG with quality around 90% or resize in steps—4× then another 2×—to keep things manageable.
• Gigapixel AI won't launch after an OS update. macOS security sometimes quarantines helper binaries. Run the Patch.command again and grant Full Disk Access in System Settings > Privacy & Security if prompted.
• Colour shift in processed files. This can happen when working in a non-sRGB colour space like Adobe RGB. Convert to sRGB before importing, or check that your display profile matches your export intent.

Eleven Phone JPEGs to 3200 px Hero Banners in One Drop

You get an email from a client. Eleven product photos shot on a phone, JPEGs at 800 × 600, slightly noisy. They need them at 3200 px wide for a web store hero banner. You drag all eleven onto the Gigapixel AI window.

The app auto-detects the content type and picks a model. You switch the first image from Standard to Lines because it's a close-up of packaging with sharp text. You set Scale to 4×, tick Remove Noise, and place a face recovery box over a tiny logo in the corner. Processing takes eight seconds on the Apple Silicon build.

You copy settings to the remaining ten shots, hit Start All, and walk away. Two minutes later, the queue is done. You open the output folder, spot-check three files for colour accuracy, and package them for the client. Nothing crashed, nothing hung.

Workflow: Rescuing a Low-Resolution Portrait for Print

You're handed a 1200 × 1800 portrait scan from a 2006 family album page. It needs to become a 10-inch print at 300 DPI—that's 3000 × 4500 pixels.

Open the file in Gigapixel AI. The app suggests the Face Recovery model automatically because it detects a human subject. You zoom into the eyes and notice JPEG blocking from the original scan compression. Switch from Auto to Manual mode and set Remove Compression Artifacts to around 60 on the slider.

Set Scale to 2.5× to hit the target pixel dimensions exactly. Tick Suppress Noise at 25% to clean the background texture without erasing the fabric weave on a shirt collar. Place a face selection box over the subject and a second one over a small dog visible in the corner—Gigapixel AI handles faces in separate regions independently.

Preview the result at 100%. The iris gains structure instead of blurring; the skin keeps fine pores rather than turning to plastic. Save as TIFF, 16-bit, Adobe RGB. The file lands in InDesign without further adjustment.

• macOS 13 or later
• Apple Silicon & Intel 64-bit