Family archivists
Scan a faded 1962 print, let restoration rebuild the faces, then finish the color.
topaz photo for final colorTopaz Labs Photo Restoration, rebuilt for the browser
Topaz Labs photo restoration is the dedicated entry point for repairing damaged images: old prints, faded scans, and low-resolution family photos. Where the general Topaz Labs tool enhances anything you throw at it, this restoration workflow is built around the problems archivists actually see — scratches, dust, torn corners, and faces that are barely there. You upload a scan, let the restoration model map the damage, and export a clean copy for printing or archiving. No account, no manual healing brush, no guesswork.
Photo restoration vs the general Topaz Labs entry point
The quickest way to understand the tool is to put it next to the main Topaz Labs app. Topaz Labs built this mode for a narrower job, and the table below shows where the split matters.
| Attribute | Restoration entry point | General Topaz Labs |
|---|---|---|
| Primary job | Fix damaged scans and old prints | Enhance any image |
| Damage profile | Scratches, dust, tears, fading | Noise, blur, compression |
| Face recovery | Rebuilds missing facial structure | Basic face detection only |
| Scratch removal | Dedicated texture model | Generic cleanup filters |
| Workflow | Batch import, repair, export | Single-image edit suite |
| Output presets | Print, archive, display | Creative presets |
The three things only photo restoration does
Most Topaz Labs tools are general-purpose. The photo restoration entry point narrows the job to three passes that show up in almost every damaged scan. Where the full Topaz Labs suite gives you dozens of sliders, the restoration entry point keeps the controls minimal.
-
Map the damage
The model reads scratches, dust, fading, and compression artifacts across the whole scan, so repairs follow the photo’s actual surface instead of a generic filter.
-
Rebuild missing detail
Faces are the priority: eyes, mouth, and skin texture are reconstructed from surrounding structure before the rest of the image is cleaned.
-
Export for the destination
Choose print, archive, or display output; the restoration is baked into a clean file with no layers, no sidecar, and no manual saving steps.
How to start restoring old photos
A photo restoration run takes about ninety seconds per image, so the process below is built around the people who actually do it. Start with the smallest damaged region that contains the subject.
Portrait retouchers
Repair the client scan first, then run the result for final grain and sharpness.
topaz image for final grainVideo archivists
Fix the stills here and take the cleaned frames for motion denoise.
topaz video for motion denoiseBatch upscalers
Restore a low-res family scan before sending it to print.
topaz image upscaler for large printsYouTube restoration channels
Use the restoration pass on keyframes, then normalize the whole clip.
topaz video upscaling for timeline cleanupInside Topaz Labs, restoration is the first pass; sharpening and upscaling come later. If you are already working in another part of the family, these are the routes archivists pair with restoration:
Where photo restoration hits its limits
The Topaz Labs restoration model performs best on scans between 300 and 3000 pixels on the longest side, with at least one recognizable facial feature. It can rebuild a blurry eye or a faded cheek, but it cannot invent a face from nothing. Large missing corners, folded-away sections, or complete color loss need manual reconstruction before the restoration pass. For larger repairs, Topaz Labs recommends cropping to the damaged area before you run the pass. The before-and-after below shows a typical result when the damage is mostly surface-level.
One restoration pass; the divider is roughly where the scratch field ends.
50 minutes saved
vs 1.5 hours of manual healing
Topaz Labs photo restoration: FAQ
Is the restoration tool free?
Yes — Topaz Labs restoration runs in the browser with no account, no upload queue, and no credit system. You process the scan and download the result.
Can it restore a face that is completely missing?
No. It rebuilds faces that are blurry, faded, or partially damaged, but if the face area is entirely gone, it cannot invent a new identity. Topaz Labs focuses on reconstruction, not recreation.
What file types work best?
Topaz Labs works best with JPEG, PNG, TIFF, and WebP scans between 300 and 3000 pixels on the longest side.
Does the tool colorize black-and-white photos?
Not by default. Since Topaz Labs keeps the original color palette, automatic colorization is a separate workflow.
How long does a batch take?
Topaz Labs processes roughly 90 seconds per image, so a 20-photo batch is about 30 minutes of processing; most time is spent queuing, not editing.