MiniPx
Add to ChromePricingBlogAccount
No. 07 — AI Tools · Cloud HDPro · on our GPU

Low light, fixed, noise and cast gone.

A dim room never produces one problem. It produces four at once — underexposure, shadow noise, a slow shutter and a colour cast — and they need different tools. Cloud HD takes the noise and the faces; the other two are free fixes you should make before you spend a run.

Pro · Cloud HDDeleted on completionUp to 25 MB · 24 MP
Clean up a low-light photoCloud HD · our GPU

Pro to enhance your photo →

How it works

  1. Pick the dim photo, ideally after brightening it yourself — JPEG, PNG or WebP up to 25 MB and 24 megapixels.
  2. It is uploaded to our server for that run and deleted when the run finishes, and in all cases within 24 hours.
  3. Download it at the same size, with shadow speckle suppressed and faces restored. One general Cloud HD run.

This page is the AI photo enhancer, tuned to what a dim room does to a photograph.

All Image Tools

AI Tools

Pro Tools

Compress

Convert

Resize

Compress to Size

Compress for Platform

Tools & Utilities

By Device

Compare

Solutions by use case

For Developers & Web

For Sellers & Ecommerce

For Job Applicants & Govt Forms

For Creators & Social

For Work & Documents

Fix a Low-Light Photo — Pro, on our server

Every low-light photo is a set of compromises the camera made on your behalf. Faced with too few photons it can open the aperture, which it has already done; hold the shutter open, which invites motion; or raise ISO, which amplifies noise along with the picture. Phones do all three and then guess a white balance from a scene lit by two different kinds of bulb. That is why a night shot arrives dark, speckled, slightly smeared and the wrong colour simultaneously — and why treating it as one problem with one fix leads to a run that disappoints. The Cloud HD enhancer owns two of the four faults squarely and nudges a third.

Order the work so each step gets a fair input. Raise exposure and lift shadows first, in whatever editor you already have — it is free, instant and reversible. That deliberately makes the noise more visible, which is the point: you want the model to see the noise as it will exist in the version you keep, not as it was hiding in a too-dark file. Then crop on your own device to lose the empty black edges. Only then upload, and the clean-up pass has a clear job rather than three tangled ones.

Know which of the remaining faults belongs to which page. If what survives after brightening is mostly speckle, the detail on ISO, shot noise and the denoise-against-texture trade lives on removing noise from a photo. If it is mostly smear, read unblurring a shaky image for what a long handheld exposure does and where recovery stops. And if the faces are the whole reason you care about the photo, the dedicated page on enhancing a face in a photo explains how the face pass and its identity check behave.

The cost, the limits and where the file goes. One enhancement is one general Cloud HD run: Pro Monthly includes 150 a month, Pro Annual 200, resetting on the 1st in UTC with no rollover and a ceiling of 30 runs an hour. Prices exclude tax, and cancellation is arranged by email with a person rather than a button — the pricing page has the detail. Your photo is uploaded to our server for that run and deleted when the run finishes, and in all cases within 24 hours, processed in the United States, encrypted in transit, and never used to train anything. The rest of MiniPx works inside your browser. Each result names its model; RestoreFormer++ is the one running now.

How it works

  1. Lift the exposure yourself first: Any editor can raise brightness and shadows for free. Do it before the run so the model sees the noise you actually want dealt with.
  2. Crop out the dead darkness: Empty black edges add nothing. Trim them with the on-device crop tool so the megapixel budget goes to the part you keep.
  3. Run one Cloud HD pass: JPEG, PNG or WebP up to 25 MB and 24 megapixels. Shadow speckle is suppressed, faces restored, faded colour lifted.
  4. Check the shadows before you keep it: Look at the darkest area at 100%. Cleaner is expected; more detailed than the original is not, and no model can invent what was never captured.

Frequently asked questions

Will it brighten a photo that came out too dark?
Brightening is not what this tool does, and saying so matters more than a bigger promise. The enhancer cleans up blur and noise, restores faces where it finds them, and lifts faded colour. Raising exposure is a slider in any editor and costs nothing — do that first, then run this to deal with the noise your brightening just made visible.
Why did my indoor photo come out orange or green?
Tungsten bulbs are warm, fluorescent tubes are green, and a camera guessing white balance in mixed light usually splits the difference badly. The fade-correction pass lifts a washed-out photograph towards its proper colour, but it is not a white-balance control. A colour-temperature adjustment in an editor is the direct fix and needs no upload.
My phone already ran night mode. Is there anything left to gain?
Sometimes. Night mode stacks several exposures and averages the noise away, which works well for static scenes and badly for people who moved between frames — that is the smeared face in an otherwise clean night shot. The face pass here is trained on exactly that damage, so faces are where the remaining gain usually is.
The photo is dark and blurry at the same time. Which problem do I attack?
They share a cause: to gather light the camera held the shutter open, and your hands moved. Fix exposure yourself, then run one Cloud HD pass and judge the result. If the smear is long and directional the honest ceiling applies, and the page on unblurring a shaky image explains why. A failed or refused run is never charged.
Does it work on concert, candlelight and campfire photos?
Those are the hardest case — extreme contrast, deep noisy shadows, a small bright source and often a moving subject. Expect the noise to improve clearly and any faces to gain the most; expect the deepest shadows to stay empty, because nothing was recorded there to recover. Limits are 25 MB and 24 megapixels either way.