One-click image enhancer for dull, flat or tinted photos
Many photos are almost good: a little grey and flat, tinted blue by shade or orange by indoor light, slightly noisy, a touch soft. This image enhancer fixes all of that in one click. Drop a photo in and it is corrected automatically — white balance, levels, noise, local contrast, colour and sharpness — with a before/after slider to judge the difference.
Unlike a single “auto” button, every step can be switched off or turned up and down, and a summary shows exactly what was changed: the colour gains, the levels that were stretched, the contrast clip limit. The preview updates as you move a slider, and the download is processed at full resolution.
All processing happens in your browser. Adaptive contrast uses the CLAHE algorithm from OpenCV.js (Apache-2.0); everything else is plain JavaScript. Your photo is never uploaded.
How to use the Image Enhancer
- Drop a photo onto the upload area, click to choose one, or paste with Ctrl+V. It is enhanced immediately with the default settings.
- Drag the slider on the preview, or use “Compare before / after”, to judge the result.
- Fine-tune any step: untick it to skip it, or move its slider to make it stronger or gentler. Auto enhance restores the defaults; Turn all off starts from the original.
- Tick Upscale 2× if you need a larger file — it is applied to the download with Lanczos-3 resampling.
- Choose a format and press Download enhanced photo.
Image Enhancer features
Grey-world white balance
Measures the average red, green and blue of the mid-tones and scales the channels so the scene averages to neutral grey — removing blue, green or orange casts.
Auto levels
Stretches brightness so the darkest 0.5% of pixels become near-black and the brightest 0.5% near-white, without shifting colours.
CLAHE adaptive contrast
Contrast Limited Adaptive Histogram Equalisation brightens shadows and reveals detail region by region, with a clip limit that stops it exaggerating noise.
Gentle noise reduction
An edge-preserving guided filter smooths grain and colour blotches before contrast and sharpening can amplify them.
Vibrance and sharpening
Vibrance lifts muted colours more than already vivid ones; an unsharp mask on brightness adds crispness without colour fringes.
Optional 2× upscale
Lanczos-3 doubles the width and height of the download, up to 16 megapixels.
What each step does and when to change it
| Step | Default | Turn it down or off when… |
|---|---|---|
| White balance | 70 | The colour is intentional — sunsets, candlelight, stage lighting, a deliberately blue night scene |
| Auto levels | 80 | The photo is meant to be soft and hazy, such as fog or a high-key portrait |
| Reduce noise | 25 | The photo is clean or has fine texture such as fabric or grass; raise it for night shots |
| Adaptive contrast | 35 | Faces look harsh or skies show halos; raise it for backlit or hazy scenes |
| Vibrance | 25 | Skin tones look too warm, or for product photos where colour must be exact |
| Sharpen | 40 | You see bright outlines around edges, or you will upscale in another tool afterwards |
How the enhancement works
The grey-world assumption says that, averaged over a whole scene, colours tend towards neutral grey. The tool measures the mean of each colour channel over mid-tone pixels (ignoring near-black, near-white and transparent areas), then multiplies each channel by a gain between 0.67 and 1.5 so the means meet. Auto levels then builds a histogram of luminance and maps the 0.5th and 99.5th percentiles towards 0 and 255.
CLAHE (Zuiderveld, 1994) divides the image into an 8 × 8 grid, equalises the histogram of each tile, clips each histogram at a limit so flat areas are not over-stretched, and blends neighbouring tiles bilinearly. Here it runs on luminance only, via OpenCV.js, so colours are not distorted. If OpenCV.js cannot be downloaded, a global histogram equalisation of luminance is used instead and the summary says so.
Noise reduction uses a guided filter (He, Sun and Tang), which averages pixels only within regions of similar brightness and colour, and sharpening is an unsharp mask: the image minus a blurred copy is added back, ignoring differences of under two levels so smooth areas stay smooth.
Tips
- Very noisy night photos benefit from the dedicated Image Denoiser first, then enhancement with “Reduce noise” turned off.
- For larger prints of small photos, the Image Upscaler offers AI upscaling; leave “Upscale 2×” off here in that case.
- For a stylised look rather than a correction, try the presets in Image Filters.
- Crop first with Crop Image: removing a dark border or a bright sky changes the statistics and usually gives a better automatic result.
Image Enhancer FAQ
How do I enhance a photo online for free?
Drop it into this image enhancer. It is corrected automatically; compare with the slider, adjust any step you like and press “Download enhanced photo”. No sign-up, watermark or upload.
What is CLAHE?
Contrast Limited Adaptive Histogram Equalisation: a method that improves contrast locally, tile by tile, with a limit that prevents noise from being boosted. It is widely used for medical, underwater and backlit images and here runs through OpenCV.js.
Why did my sunset lose its warm colour?
Grey-world white balance assumes the scene should average to neutral, which is not true for sunsets or candlelight. Turn White balance down or off for those photos.
Can it fix a blurry photo?
It can make a slightly soft photo look crisper with sharpening, but it cannot recover detail lost to camera shake or missed focus.
Is the download full resolution?
Yes. The preview uses a smaller copy for speed, and the download is processed at the original size, up to 16 megapixels (4 megapixels before a 2× upscale).
Are my photos uploaded?
No. The enhancement runs in your browser. Only the OpenCV.js library is downloaded from a public CDN, once, for the adaptive contrast step.