How to Improve Old Photo Quality with AI (2026): 7 Tools Compared
7 AI tools to improve the quality of old family photos, compared on what each is built for. Remini, Topaz, jpgHD, VanceAI and more, with pricing and a step-by-step method.
AI & Technology Writer, Incarn
In short
7 AI tools to improve old family photos (1940–1980), by use: Remini specializes in face reconstruction (free in a limited version on mobile). Topaz Photo AI is a desktop all-rounder for upscaling, denoising and sharpening, at $199. jpgHD is free, web-based and needs no sign-up. Golden rule: improving resolution before animating gives the animation AI more detail to work with.
TL;DR: 7 AI tools to improve old family photos, by use: Remini specializes in face reconstruction (free in a limited version). Topaz Photo AI is a desktop all-rounder at $199. jpgHD is free and needs no sign-up. Enhancing a photo before animation gives the animation AI more detail to work with.
A photo from 1965, run through a scanner. It comes out with a bluish blur, lost details, a resolution of 400 pixels wide. It isn't damaged. Not scratched. Just blurry and low resolution.
This is the case for the vast majority of family photos: the problem isn't physical deterioration, it's the image quality itself. AI image enhancement tools are built exactly for this. Not to "restore" in the sense of retouching scratches, but to reconstruct sharpness, upscale resolution, and recover lost detail.
Here are 7 of them, and what each one is built for.
Why Improve the Quality of an Old Photo
Old family photos rarely suffer from just one problem. Usually it's a combination of three:
Low resolution. A 4×6 inch print scanned on a standard desktop scanner produces 300 to 600 pixels wide. That's not enough to print, animate, or display properly on a modern screen.
Blur and grain. Manual focus, slow shutter speeds, high-sensitivity film, film photography from the 1950s–1970s had its constraints. Sixty years later, the details fade.
Flattened dark areas. Black clothing, poorly lit interiors, deep shadows: these areas often turn into a uniform flat tone in old prints.
Upscaling, sharpening, restoration: three different problems
Three operations, three different tools. Picking the wrong one wastes time and sometimes degrades the photo further.
Upscaling increases the pixel dimensions of an image by reconstructing new pixels. A 400-pixel scan becomes 1600 pixels. Algorithms like Topaz Photo AI or Upscayl predict what a higher-resolution version should look like. Result: a larger image, often sharper, without inventing details the original never contained.
Sharpening increases local contrast at edges to give the impression of more detail. It does not add pixels. Adobe Lightroom AI Sharpen works well on slightly blurry photos. On very blurry originals (motion blur, deep focus failure), sharpening alone won't help. Upscaling first, sharpening after, gives better results.
Restoration repairs physical damage: scratches, tears, stains, yellowing. A completely different task from upscaling or sharpening. VanceAI's "Old Photo Restoration" mode and Remini handle some degree of damage repair. For a badly scratched print, restoration comes before upscaling.
Practical rule: small and blurry photo, start with upscaling (jpgHD, Topaz). Damaged photo, start with restoration (VanceAI, Remini). Needs contrast only, Lightroom AI Sharpen is the fastest path.
What Image Enhancement Changes for Animation and Printing
Two main use cases.
Before animating. An AI animation algorithm reconstructs movement from the information present in the source image. The sharper and more detailed the image, the more the model has to work with. A blurry photo leaves it less to read, which can show up as shaky edges or unnatural movement; upscaling first gives it a cleaner face to animate.
Over 12,000 photos have been animated on Incarn, which we make, since launch. The same advice applies to every one of them: start from the sharpest version of the photo you can get, especially on faces.
Before printing. A quality A4 print requires around 2480 × 3508 pixels. Starting from a 400-pixel photo and expecting a sharp print is a lost cause without upscaling first.
7 AI Tools Compared (2026)
Remini (mobile, free + premium)
A widely used tool for portraits. Remini specializes in face reconstruction: it identifies facial features and is designed to restore them, even in heavily degraded photos.
What to expect: Remini focuses on faces, so on a dark, blurry wedding photo the faces are what it works on; the background may stay soft.
Advantages: free in a limited version (1–2 enhancements per day), simple interface, available on iOS and Android.
Limitations: because it focuses on faces, backgrounds may stay blurry, and photos without faces are not its strength.
Pricing: free (limited) / around $10/month or $45/year for the premium version.
Topaz Photo AI (desktop, paid)
A reference for professional upscaling. Topaz Photo AI combines upscaling, denoising, and sharpening in a desktop interface. It can multiply resolution by 4× or 6×.
What to expect: on a grainy 400-pixel group photo, it is designed to raise the resolution (2400 pixels at 6×) while denoising and sharpening in the same pass.
Advantages: the most versatile tool on this list, built to handle all photo types (portraits, landscapes, groups, architecture).
Limitations: expensive, slow on machines without a recent GPU.
Pricing: $199 one-time purchase, or a subscription at around $70/year.
jpgHD (web, free)
A web tool specialized in old photos. No sign-up required: upload, wait 30 seconds, download.
What to expect: aimed at old photos with moderate grain. For very blurry or very low-resolution photos, compare it with a dedicated upscaler.
Advantages: free, no sign-up, fast, specialized for old photos.
Limitations: no control over parameters. Pricing: free.
VanceAI (web, freemium)
An online tool suite covering upscaling, denoising, and photo restoration. The interface lets you chain multiple processes in sequence.
What to expect: the "Old Photo Restoration" mode is designed for old photographs. On heavily damaged prints, check the result closely.
Pricing: 3 free trials, then around $9/month for 200 credits.
YouCam Enhance (mobile, freemium)
A mobile app developed by Perfect Corp. Aimed at portraits, with a simple interface for general consumers.
What to expect: built around individual portraits rather than group photos.
Pricing: free (with ads) / around $4/month for the pro version.
Adobe Lightroom (desktop + mobile, paid suite)
Since 2024, Lightroom has integrated AI features: Super Resolution, AI Denoise, and AI Sharpen. If you already have an Adobe Creative Cloud subscription, this is the natural choice.
What to expect: AI Denoise is the feature to try on heavily grainy photos.
Pricing: included in Creative Cloud Photography plan starting at $10/month.
Upscayl (desktop, free and open source)
The free, open-source alternative for those who don't want to upload their photos to third-party servers. Runs locally on your machine.
What to expect: a good fit for photos in good condition. For difficult cases, a paid all-rounder like Topaz offers more controls.
Pricing: completely free.
Quick Comparison Table
| Tool | Best for | Free? | Platform | Main focus |
|---|---|---|---|---|
| Remini | Portraits, faces | Yes (limited) | Mobile | Face reconstruction |
| Topaz Photo AI | All types | No ($199) | Desktop | Upscaling, denoising, sharpening |
| jpgHD | Old photos | Yes | Web | Upscaling, no sign-up |
| VanceAI | Upscaling + multi-processing | 3 trials | Web | Chained processes, old photo mode |
| YouCam Enhance | Mobile, portraits | Yes (limited) | Mobile | Consumer portrait enhancement |
| Adobe Lightroom | Pros, existing subscribers | No ($10/month) | Desktop + mobile | Super Resolution, AI Denoise, AI Sharpen |
| Upscayl | Desktop without cloud | Yes | Desktop (local) | Open-source upscaling |
How to Improve an Old Photo: Step by Step
Scan at high resolution first. If you're starting from a print, scan at a minimum of 600 dpi. 1200 dpi if the photo is small.
Identify the main problem. Motion blur and portraits: start with Remini. Heavy grain: Topaz or Lightroom. Low resolution on an otherwise good photo: jpgHD or Upscayl. All three combined: Topaz Photo AI.
Start with the free tools. jpgHD or Remini cover most common needs.
Don't chain multiple tools. Each pass introduces new artifacts.
Always keep the original.
If the photo is black-and-white, adding color after upscaling gives the animation AI more information to work with. Our guide to colorizing old photos with AI covers five tools for this step. For photos with physical damage beyond blur, the old damaged photo guide covers what's recoverable and what isn't.
Frequently Asked Questions
What is the difference between AI upscaling and sharpening?
Upscaling increases the number of pixels in an image, typically 4× or 6×, making it larger and often sharper overall. Sharpening increases local contrast at edges, making details look crisper without adding new pixels. For old family photos, upscaling is usually the right starting point: it reconstructs resolution that was never captured at scan time. Sharpening is a finishing pass. For a 400-pixel photo, upscale first (Topaz, jpgHD, Upscayl), then sharpen if edges still look soft.
What is the difference between upscaling and restoration?
Upscaling increases resolution and improves the overall sharpness of an image that is in generally good condition. Restoration repairs physical damage: scratches, stains, tears. For a photo that is both scratched and blurry, you need both: restoration first, then upscaling.
Can these tools invent details that weren't there originally?
Upscaling algorithms reconstruct pixels based on statistical patterns. They extrapolate. On faces, the reconstruction is usually plausible but can subtly change features, so compare with the original. On fine text or logos, artifacts can appear.
Should I enhance quality before animating on Incarn?
Recommended, but not required. Animation works on raw photos, but a sharp, detailed image gives the model more to work with. The free trial (first animation offered at sign-up as a watermarked preview, then 1.99 EUR per animation) lets you see the difference for yourself.
AI & Technology Writer, Incarn
Thomas covers AI and machine learning applications for creative tools. Former research engineer with a focus on computer vision and video generation.