2. Topaz Gigapixel - best raw detail recovery

Gigapixel is the tool photographers name first when detail is the whole job, and the reputation is earned. It ships a stack of specialized AI models, Standard, High Fidelity, Low Resolution, Text and Shapes, and more, so you can match the model to the source instead of forcing one algorithm on everything. Per the official product page it upscales from 0.5x up to 6x in a single pass, all rendered locally so files never leave your computer. The catch is the bill. Topaz moved Gigapixel to a subscription-first model in October 2025, so the perpetual license that occasional users relied on is no longer the headline option.

up to 6x

Source

Verdict: Pick Gigapixel when detail recovery is everything and you will use it enough to justify a yearly bill.
My take: The detail is the part that stuck with me. I pushed a soft portrait to 4x and the eyelashes and hair came back with texture that almost nothing else here matches, no waxy skin. Then I hit the wall every casual user hits: it is a rented tool now, and for one poster I did not want a subscription running in the background.

3. Adobe Super Resolution - best for RAW files and existing Adobe users

If you already live in Lightroom or Camera Raw, Super Resolution is the enlarger you already own. You right-click a photo, choose Enhance, tick Super Resolution, and Adobe rebuilds it at twice the linear resolution. As the Adobe help docs spell out, that is two times the width and two times the height, or four times the total pixel count, so a 12-megapixel frame becomes a 48-megapixel one. It is strongest on RAW files, where it has the most data to work with, and weaker on a small JPEG. The limits are the scope and the price: it does a single fixed 2x jump, not the 6x or 8x the dedicated tools reach, and it only exists inside a Creative Cloud subscription.

Verdict: The obvious pick for photographers already in Lightroom who shoot RAW and need a clean 2x.
My take: On a RAW file the result was genuinely clean, four times the pixels with edges that held up. The contrast came when I fed it a small JPEG: the same feature did far less, because there was less real data to rebuild from, and 2x is as far as it goes.

4. Upscayl - best free and open-source enlarger

Upscayl surprised me. It is completely free, open source, and runs entirely on your own machine, and it is genuinely good. It is built on Real-ESRGAN and ships several models (UltraSharp, Remacri, High Fidelity and more), so it is not a toy. Per its GitHub project it upscales up to 4x per pass with images staying local, no account and no cloud, which is a rare combination at zero cost. The one real requirement is a Vulkan-compatible GPU for usable speed, so it is happiest on a machine with a modern graphics card rather than a thin laptop.

4x per pass, 0 cost

Source

Verdict: The best free pick for anyone with a decent GPU who wants private, no-cost upscaling.
My take: I expected a rough hobby app and got a real one. On a photo with a capable GPU it held detail nearly as well as tools that charge monthly. The trade is that it is upscale-only, so when a corner came out soft there was no brush to fix it, just another pass.

5. waifu2x - best for anime and line art

waifu2x is the veteran, and it is still the right answer for one specific thing: anime-style art and clean line drawings. It reduces noise and upscales in one shot, and its open-source project is clear about the ceiling. It supports 2x scaling only, with adjustable denoise from level 0 to 3. Bigger results come from running the 2x model again, not from a native higher factor. For flat colors and hard outlines it is excellent. For a detailed photo of a real face, a modern Real-ESRGAN tool will beat it.

2x

Source

Verdict: The go-to for illustration and line art, not for photographic enlargement.
My take: One job, done beautifully. I ran a scanned line drawing through it and the outlines came out crisp and clean with the paper noise gone, better than any general tool managed on that file. Hand it a real photo and its age shows.

6. BigJPG - easiest web upscaler for a quick job

Would you rather not install anything? BigJPG is the browser answer. It uses a deep convolutional network in the waifu2x lineage, so it leans toward anime and illustration but handles photos too, and you can push it to high scales on the paid tier while the free tier caps dimensions and adds a queue. It is fast and needs zero setup. The trade-offs are the ones every web tool carries: your image uploads to a server, the free tier is limited, and you are trusting a service with whatever you feed it.

Verdict: Fine for a one-off upscale of a non-private image when you do not want to install anything.
My take: I dropped in a small illustration and had a 4x result in under a minute with no signup, which is the appeal. The free tier size cap and the fact that my image left my machine are why it is a convenience tool, not a workhorse.

7. Let's Enhance - cloud batch upscaling

Let's Enhance is the cloud-service take on enlarging. You upload, it processes on its own servers, and it markets upscaling to high resolution with batch support, aimed at e-commerce and marketing teams pushing many product shots at once. The convenience is real: no local hardware needed, and it scales to a queue of images. The cost is the usual cloud math. It runs on credits or a subscription, and every image you enlarge is an image you have uploaded to a third party.

Verdict: Reasonable for a business pushing volume through the cloud, wrong for private or occasional work.
My take: I ran a product shot through the free allowance and the result was clean and usable. Two things kept it off the top for me: the credits tick down fast on a real batch, and the whole point of the tool is that your images live on someone else's server.

8. ON1 Resize - best for very large print enlargements

ON1 Resize (the descendant of Genuine Fractals) is the specialist for going big on paper. It uses a tiling approach built for extreme enlargements, so it is the tool people reach for when the output is a banner or a gallery print rather than a screen. It runs standalone or as a plugin, on Windows and Mac, and it can be owned with a one-time license or rented. It is not an AI-first detail recoverer in the Gigapixel sense, so on a soft small source a modern AI tool often adds more believable texture, but for controlled, very large print scaling it remains a trusted name.

Verdict: The pick when your output is a large physical print and you want proven, controlled scaling.
My take: This is a print tool, and it acts like one. I set a target print size and it enlarged with steady, controlled results, no wild artifacts. For a web-size crop it was more machine than I needed.

This is the single feature that should drive your pick, so it earns its own section. When you enlarge a photo, software has to decide what to put in the new pixels it is creating. There are two ways to do it. Interpolation, the old way, averages the neighboring pixels, which is fast and always available and always a little soft, because it is guessing the safe middle value. Detail reconstruction, the AI way, was trained on millions of images to predict what a real edge, pore, or strand of hair should look like at the larger size, and paints that in. That is why a good AI enlarge of a 500 by 500 shot can look sharp at 2000 by 2000 while a bicubic resize of the same file looks like a soft balloon. But reconstruction has a hard limit worth stating plainly, because half the disappointed reviews online come from missing it: the AI can only rebuild detail that is plausible from what is there. Feed it a face that was already a blurry smudge and it will sharpen the shape, not conjure the person's actual pores. On top of that sits artifact control, the second half of the job. The same model that invents detail can overshoot into halos around high-contrast edges, ringing along lines, or that poreless plastic-skin look. The tools we ranked highest add detail and keep those tells down at the same time. If your goal is sharpening a slightly soft shot rather than making it bigger, a focused image sharpening pass is the better first move.

Want to spend nothing? You have better options in 2026 than you did a few years ago, and I put the same small originals through the free routes to see where each one lands. Upscayl is the one to try first: it is free, open source, runs entirely offline, and on a machine with a Vulkan-capable GPU it produced results close to the paid desktop tools, with no watermark and no upload. waifu2x is the free champion for anime and clean line art, capped at 2x but excellent inside that lane. On the web, BigJPG and Fotor's free tier both gave a quick browser upscale, but the pattern is consistent: a size cap, a queue, sometimes a watermark, and your image uploaded to a server. GIMP is the free, fully offline desktop option, but its resize is plain interpolation with no AI reconstruction, so it enlarges without adding real detail. The honest limits of the free route stay steady. You get lower ceilings on scale, weaker batch handling, and, for the web tools, the privacy cost of uploading. For an occasional non-private image, free is a perfectly good answer. For a private batch or a large print, an offline tool you own will save you re-doing the job. When you only need to make a folder smaller rather than larger, a batch picture resizer is the faster route than any AI upscaler.

Print is where enlargement earns its keep, and where the ceiling bites hardest. The rule that has held up across every tool is to start from the largest clean original you can get and to enlarge in moderate steps. A 2x or 4x AI upscale from a solid source usually holds up beautifully at print size. A single 8x jump from a tiny, soft file will show its seams no matter which tool you use, because the AI is being asked to invent most of the picture. Before you send anything to a vendor, zoom to 100 percent and hunt the failure modes: halos ringing around high-contrast edges, a plastic sheen on skin, and mush in fine detail like hair or foliage. If you see them, drop back a step or switch models. Match your final pixel dimensions to the print size and DPI your print shop specifies, and export once at the end. Every extra JPEG save on the way stacks compression artifacts that no enlarger can undo. For a poster from a clean source, Gigapixel or ON1 Resize give the most controlled results; for a private print from a small original on Windows, our own Photo Retoucher keeps the whole job offline.

Offline versus cloud is the cleanest dividing line in this category, just as it is in restoration. Of the eight tools here, the five desktop apps do the work on your own machine, so a client's product shots or a personal photo never leave the building. The web services need to upload. That is not a hypothetical concern for anyone handling images under a client agreement or images of family. Upscayl makes a point of it in its own project notes, that images stay local so you can upscale confidential assets without sending files to the cloud, and that local-first stance is exactly why it and the other desktop tools sit above the web services for private work. For a throwaway image where you just want zero install, a web upscaler is a reasonable trade. For anything you would not email to a stranger, keep it on your PC.

Also considered: enlargers that did not make the cut

Eight picks is a shortlist, not the whole market. These are the tools we tried or read up on and left off, with the reason each one dropped:

• VanceAI. A capable cloud upscaler with a restoration model too, but it runs on a credit system in the cloud, so every batch uploads your images and spends credits. That is the opposite of the private, own-it-once route most people want. We cover its enlarger separately.
• Fotor and Nero AI. Both give a fast, free browser upscale that is fine for a quick social crop, but they watermark or cap the free output and upload your image, so they land as convenience tools rather than finishers.
• GIMP. Free and fully offline, which we like, but its Scale Image is plain interpolation with no AI detail reconstruction, so it enlarges the pixels you have without adding new detail. Good for a mild resize, not for real upscaling.
• Windows Photos and macOS Preview. The resizers already built into your system. Fine for shrinking an image or a tiny bump, but they only interpolate, so a real enlargement comes out soft.

How to choose the right photo enlargement software

Work backward from what you have and where it is going. The two questions that decide most of it are whether the images need to stay off the cloud, and whether you are doing this once or often. Use the scorecard to see the trade-offs at a glance, then read the persona that fits you:

Sources

• Topaz Gigapixel. Official product page; up to 6x single-pass upscale, nine AI models, local rendering, subscription-first pricing.
• Adobe Camera Raw Enhance docs. Super Resolution doubles linear resolution (2x width and height, 4x pixels).
• Upscayl on GitHub. Free and open-source AI upscaler, Real-ESRGAN models, up to 4x per pass, runs locally.
• waifu2x on GitHub. 2x native upscale with denoise levels 0 to 3, tuned for anime and line art.
• ON1 Resize. Dedicated print-enlargement tool with tiled scaling, standalone or plugin.