How to upscale an image with AI

Sharpen and enlarge any image with AI upscaling. A step-by-step guide to raising resolution, recovering detail, and exporting print-ready images in Morphic.

Try on Morphic

To upscale an image with AI, upload it to Morphic, run the upscale (a paid feature, with enhancement if you want it), and export the larger, sharper version. AI upscaling reconstructs real detail rather than stretching pixels, so edges and textures stay clean at the new size.

Drag the handle to compare the same image before and after AI upscaling, blurry to crisp.

The same portrait sharp and high-resolution with fine detailUpscaled
A blurry, low-resolution close-up portrait lacking detailLow-res

Steps at a glance

  1. Upload the image to upscale
  2. Choose upscale, and enhance if needed
  3. Set the target resolution
  4. Review the detail
  5. Export the upscaled image

Upscale your image step by step

1.

Upload the image to upscale

Start with the sharpest version of the image you have, because upscaling amplifies whatever detail is already there. A crisp original upscales cleanly; a small, blurry, or heavily compressed one has less for the model to reconstruct and will hit its limits sooner. Whether the image was generated in Morphic or came from a camera or another tool, upload the best-quality copy rather than a screenshot or a re-saved, degraded version.

2.

Choose upscale, and enhance if needed

Run the upscale operation. This is a paid feature, and the free tier keeps upscale off, so you will need a paid plan for this step. If the image also needs help beyond size, add enhancement to denoise, sharpen, and fix lighting at the same time. Keep the two goals clear in your mind: upscaling makes it bigger and crisper, while enhancement cleans up quality, and you can do both together when a photo needs it.

3.

Set the target resolution

Decide how large you need the result. Match it to the use: a web banner and a large print need very different resolutions, and upscaling past what the source can support wastes effort without adding real detail. Aim for the resolution the final use actually requires rather than the maximum available, since a sensible target gives the cleanest result and the model is not asked to invent detail the original never had. As a rough guide, doubling or tripling the dimensions of a decent source is a safe, clean jump, while a huge multiple from a tiny thumbnail is where results start to look soft.

4.

Review the detail

View the upscaled image at full size and check the parts that matter: faces, text, fine textures, and edges. A good upscale looks faithful, just larger and sharper, with no blocky artefacts or over-smoothed, plasticky areas. If something looks off, try a smaller target or add enhancement, since pushing a weak source too far is what produces artefacts rather than the upscale itself being at fault.

5.

Export the upscaled image

When the detail holds at full size, export at the resolution you set. For print, confirm the dimensions suit the physical size and DPI you need; for screen, a clean full-resolution export is enough. Because upscaling is a paid feature, exporting from a paid plan also gives you a clean, watermark-free result, which matters for anything published, sold, or printed rather than tested.

Upscale versus enhance versus restore

These three overlap but do different jobs. Here is which to reach for.

UpscaleEnhanceRestore
Main jobMake it bigger and sharperDenoise, sharpen, fix lightingRepair damage in old photos
Best forLow-resolution imagesSoft or noisy shotsScratched, faded originals
Changes the look?No, just largerCleans up qualityRepairs and can colourise
PlanPaidPaidPaid

To generate a high-resolution image from scratch, see how to make an AI image; to repair an old or damaged photo specifically, restoring old and damaged photos is the focused guide.

Put it together

Upscaling with AI is a paid step, and the trick is starting from the sharpest source you have and setting a sensible target. Upload the best copy, run the upscale (adding enhancement if the image needs cleaning too), and check faces and edges at full size before exporting. Aim for the resolution the final use needs, and a small or soft image becomes a clean, print-ready one. This is the step that lets AI images and older photos alike hold up at full screen or in print, so it is worth doing last, once the composition and any edits are locked, rather than upscaling something you will only change again. Match the size to how much real detail the source holds, and done right the upscale is invisible: the image simply looks like it was always that sharp and that large.

FAQs

Is AI upscaling free?
No. Upscaling and enhancement are paid features in Morphic, and the free tier keeps upscale off. You can generate and edit images on the free plan, but to raise resolution and recover detail you will need a paid plan.
How much bigger can I make an image?
AI upscaling raises an image to a larger, sharper version rather than just stretching pixels, so the result stays clean at the new size. The practical limit depends on the source: a sharp original upscales further and cleaner than a small, blurry one.
Does upscaling actually add detail?
It reconstructs plausible detail rather than inventing it from nothing, so edges, textures, and small features come back sharper instead of blocky. It works best on images that were reasonably sharp to begin with; it cannot recover detail that was never captured, so a heavily blurred photo has limits.
Will it change how the image looks?
A good upscale keeps the image faithful, just larger and crisper, so faces, colours, and composition stay the same. Enhancement can also fix lighting and denoise if you ask for it, but pure upscaling aims to preserve the original look while raising resolution rather than restyling anything.
What formats can I upscale?
Standard image formats work, whether the image was generated in Morphic or uploaded from elsewhere. Upload the sharpest version you have, since upscaling amplifies whatever detail is present, then export the upscaled result at the resolution you need for print or full-screen use.