Upscaling Old Photos: A Practical Guide
A guide from the NextG Design team
"Upscaling" gets used loosely, so it's worth being precise about what it actually does: it increases the pixel dimensions of an image and tries to make the new, larger version look plausible — it does not recover detail that was never captured in the first place. Understanding that distinction is the key to getting a result you're happy with.
Why old photos look soft to begin with
A photo scanned at low resolution, a screenshot, or an old low-megapixel camera photo has a fixed amount of real detail baked into it. When you view it larger than its native size, your screen or printer has to invent the in-between pixels somehow — and a basic resize (like most browsers or image viewers do automatically) invents them crudely, which is what produces that soft, blurry look. A dedicated upscaler does a better job of that invention, plus adds back some perceived sharpness, but it's still working from the same limited source detail.
What the Upscaler actually does
The Image Upscaler combines three things: a higher-quality resize than a basic browser scale, edge-aware noise reduction that smooths grain and scan artifacts without also smearing real edges, and unsharp-mask style sharpening that boosts the appearance of detail along genuine edges. Used together and balanced correctly, the result looks noticeably clearer and more print-ready than a naive resize — but it's an enhancement of what's there, not a reconstruction of what's missing.
Balancing noise reduction and sharpening
These two settings pull in opposite directions, and getting the balance right matters more than maxing out either one:
- Too much noise reduction smooths away fine texture along with the actual noise — skin, fabric, and foliage can start to look waxy or plastic.
- Too much sharpening exaggerates the edges that are left, which can look crisp on a screenshot but introduces visible haloing around high-contrast edges, especially on already-noisy scans.
- For a heavily degraded scan, lean toward the Gentle Enhance preset first — noise reduction has more to clean up, so a lighter sharpening pass looks more natural.
- For a clean but simply low-resolution digital photo, Heavy Enhance tends to look better, since there's less noise to worry about amplifying.
Set realistic expectations for the output size
Going from 100% to 200% (doubling each dimension) usually looks convincing. Pushing all the way to 400% works better on some images than others — it depends heavily on how much real detail was in the source. If a heavy upscale starts looking artificial, try a smaller percentage; a slightly smaller but more natural-looking result is usually more useful than a larger, mushier one.
Compare before you commit
Toggle noise reduction and sharpening on and off while previewing, and try both presets, before downloading — the live preview updates instantly, so it costs nothing to compare a few combinations on your specific photo rather than assuming one setting is always best.