**Free online image brightener — lift shadows in a dark photo**

> Lighten an underexposed or backlit photo online, free, with a gamma curve, a shadow lift or a strong backlit recovery — and see what each curve does first.

Source: https://appvau.lt/tools/image-brightener/ · updated: 2026-09-11

# Brighten a dark photo without washing out the rest of it

Three tone curves, written out rather than hidden behind a slider labelled auto. One lifts the whole picture, one lifts only the shadows, and one is a strong recovery for a subject standing against a window.

DROP · PASTE · CHOOSE

Drop an image here

or choose a file. It is opened in this browser tab and never uploaded — this page has no endpoint to upload it to.

## Underexposed is not the same problem as backlit

Two photographs can both look too dark for opposite reasons, and the fix for one ruins the other. In an underexposed photograph the whole frame is dark: not enough light reached the sensor, and everything from the shadows to the highlights sits lower than it should. In a backlit photograph the frame is not dark at all — the window behind the subject is correctly exposed and bright, and the subject in front of it is a silhouette.

Apply a whole-image brightening curve to the backlit case and the window goes from bright to white, the sky loses its last gradient, and the picture acquires the washed-out, flat look that people associate with "brightened". Apply a shadow-only lift to the underexposed case and the highlights stay too dark, because there was never anything bright about them.

The tool reports the mean luminance of the image when you load it, which separates the two cases in one number. A mean under about 32 of 100 is genuinely underexposed. A mean in the normal range with a dark subject is the backlit case, and the shadow lift or the backlit curve is what it wants.

## The three curves, as functions

Each mode is a single function from input brightness to output brightness, applied through a 256-entry lookup table — the same operation every photo application performs, done once per possible value rather than once per pixel. Here is what each one does to six sample levels at a strength of 60, expressed as percentages of full brightness.

*Input level, and what each curve turns it into at strength 60*

| Curve | 5% | 15% | 30% | 50% | 70% | 90% |
| --- | --- | --- | --- | --- | --- | --- |
| Natural | 16% | 32% | 48% | 66% | 81% | 94% |
| Shadow lift | 34% | 38% | 45% | 57% | 72% | 90% |
| Backlit subject | 51% | 54% | 59% | 67% | 78% | 88% |

Read the right-hand columns. Natural moves the 90% sample as well as the 5% one, which is what "brighten the whole image" means and why it flattens a bright scene. Shadow lift barely touches it: by the time a pixel is at 70% it is getting almost nothing, so a sky stays a sky. Backlit is the aggressive version of the same shape, with a roll-off above 82% so that the brightest areas compress instead of clipping to white.

The curve is applied to each colour channel identically. That preserves hue and reduces saturation slightly as things get lighter, which is what a person expects from a brightness control. Applying it to luminance alone and re-scaling the colour would hold saturation better and would also amplify colour noise in exactly the shadows being lifted, which is the worse trade at this end of the quality scale.

## Noise is the price, and it is not a defect

A sensor records signal and noise together. In bright light the signal is large and the noise is invisible underneath it; in a dark room the signal is small and the noise is a substantial fraction of what was recorded. Brightening multiplies both by the same amount, so a lifted shadow is always grainier than a correctly exposed one.

That is not the tool failing. It is the photograph telling you how much information it actually contains. If the grain is unacceptable, the options are to lift less, to output at a smaller size — where the noise averages away as pixels are combined — or to go back to the camera's RAW file if one exists, because a RAW holds several stops more headroom than the eight-bit JPEG this tool receives.

Keep the output quality high for a brightened image. Noise is high-frequency detail, which is exactly what a lossy encoder spends the most bits on, so a brightened photograph at quality 70 will look markedly worse than the original did at quality 70.

## What cannot be recovered

Clipped highlights are gone. When a pixel reads 255 in all three channels the sensor saturated, and the file records that it saturated and nothing about how far past saturation the scene went. Brightening a photograph with a blown-out window produces a slightly larger blown-out window.

A completely black frame is the same problem at the other end. If the sensor recorded nothing, there is nothing to multiply, and lifting it produces grey noise. The mean-luminance figure the tool reports on load is there so that this is visible before you spend time on it.

And brightening is not denoising, sharpening or white balance. It is one operation, stated plainly, with its curve published — which is more than most sliders offer, and considerably less than a photo editor does. For a photo that is nearly right, this is the whole job; for a photo that needs work, a real editor is the honest answer.

## The related case: keeping the highlights instead

There is a second way to make a photograph look brighter that does not touch the shadows at all: give the highlights more headroom to work with on a screen that has it. That is what a gain map does, and it is a different tool — [the Ultra HDR converter](https://appvau.lt/tools/hdr-image-converter/) builds one, so the same photograph is unchanged on an ordinary screen and noticeably more three-dimensional on an HDR one.

The two are complementary rather than alternatives. Brightening changes the pixels everyone sees; a gain map changes what a capable display does with the pixels that are already there. A dark photograph usually wants the first. A correctly exposed photograph with a bright window or a specular reflection usually wants the second.

Whichever you use, the output here is encoded from pixels, so the location, the camera and every other metadata field are dropped along the way. [The metadata tool](https://appvau.lt/tools/photo-metadata/) is where to look if that is the point rather than a side effect, and [the compressor](https://appvau.lt/tools/image-compressor/) is the page for getting the brightened result under a size limit.

WHAT IT DOES NOT DO

## The limits, stated here rather than discovered later

- Brightening amplifies whatever the sensor recorded, including its noise. A photo taken at high ISO in a dark room gets brighter and grainier at the same rate.
- Clipped highlights cannot come back. A blown-out window is 255 in all three channels and there is nothing behind it.
- This works on the 8-bit image. A camera RAW file holds far more headroom, and a RAW editor will always beat this on the same photograph.

## Questions people ask about this

1. 01 How do I brighten a dark photo? Decide first whether the whole picture is dark or only the subject. A whole-image gamma curve fixes underexposure; a shadow lift fixes a subject standing in front of a window, because it leaves the bright part where it is. Applying the first to the second problem is what produces the washed-out look people associate with brightening.
2. 02 What does the shadow lift actually do? It adds an amount proportional to the square of how dark a pixel is, so black gets the full lift, a midtone gets a quarter of it and a highlight gets none. The curve is written out on this page rather than hidden, because the difference between the three modes is the only decision the tool asks you to make.
3. 03 Why does my brightened photo look grainy? Because the grain was always there and was too dark to see. Brightening multiplies signal and noise equally, and shadow noise is the largest part of what a sensor records in low light. A gentler setting, or a smaller output size, both hide it.
4. 04 Is my photo uploaded? No. The curve is applied to pixels in this tab through a 256-entry lookup table, and nothing is sent.
5. 05 Can this recover a completely black photo? No. If the sensor recorded nothing, there is nothing to amplify — you get grey noise. The tool reports the mean luminance when you load a file so the answer is visible before you try.
6. 06 Will it change the colours? It applies the same curve to each channel, so hue is preserved and saturation drops slightly as things get lighter, which is what every photo application calls brightness. It is not a white-balance tool and does not attempt to be.
7. 07 Does brightening remove metadata? Yes, because the output is re-encoded from pixels. If that is the goal rather than a side effect, use the metadata tool, which does it without touching the picture.

## Other tools on this site

[All 10 tools](https://appvau.lt/tools/)

- [Image files Image compressor](https://appvau.lt/tools/image-compressor/)
- [Image files Image resizer](https://appvau.lt/tools/image-resizer/)
- [Image files Ultra HDR converter](https://appvau.lt/tools/hdr-image-converter/)

These tools exist because the same problem keeps turning up underneath the app: a photograph says more than the person sending it meant to. App-Vault is the iPhone side of that — an encrypted vault for the files that should not be in the camera roll at all. [What the vault does](https://appvau.lt/photo-vault/), [the cryptography behind it](https://appvau.lt/encryption/) are documented rather than asserted.

[Home](https://appvau.lt/) [Browser tools](https://appvau.lt/tools/) Image brightener

GET STARTED

## Seal the vault.

Free to download. The free tier holds one file. Premium removes the limit — Weekly from $3.99 or Yearly from $29.99 with a 7-day free trial.

[Open App-Vault on iPhone](https://appvau.lt/download/)

---

HTML version: https://appvau.lt/tools/image-brightener/
Structured data for this site: https://appvau.lt/api/v1/openapi.json · https://appvau.lt/llms.txt
Free to quote and reuse with a link back to the source URL above (CC BY 4.0).