IMG · Images tools

Auto Photo Fixer

Drop a photo and the tool analyzes the image data before applying any transformations. It maps the luma distribution, average saturation, and channel imbalance to build a specific correction profile.

Image-aware correction

The algorithm calculates the corrections the image actually needs:

  • Contrast and exposure: Luma percentiles dictate a bounded S-curve. Bright, flat photos gain contrast, while well-exposed photos receive only a baseline curve to prevent washed-out highlights.
  • Color cast: Minor channel imbalances from indoor lighting are neutralized. Severe deviations from sunsets or colored gels are recognized as intentional lighting and ignored.
  • Local shadow and highlight recovery: The tool calculates a local 32x32 luminance grid to target blocked shadows and blown highlights. It applies localized recovery without flattening the image contrast.
  • Skin-tone protection: HSV color gating detects the specific hue and saturation range of human skin. It masks these areas from heavy vibrance and white-balance shifts so faces do not turn gray.
  • Horizon auto-leveling: A Sobel edge detector feeds into a Hough Transform accumulator to find dominant straight lines in landscapes. If the horizon is crooked up to 15 degrees, the tool rotates the image and calculates the maximum inner bounding box to crop out transparent edges.

The detected issues appear as tags below the controls.

The Intensity slider

The Intensity slider scales every computed correction, including the auto-leveling angle, by the same factor. At 100 percent the full mathematical fix is applied. Reduce the slider to blend the correction linearly with the original pixels.

Engine limits

Automatic correction cannot recover missing data. A sky clipped to pure white has no cloud detail left to restore. Use the preview to verify shadow and highlight retention before downloading.

Frequently Asked Questions

It reads the brightness histogram and local luminance, then corrects low contrast, overexposed midtones, blocked shadows, and muted colors. A Hough Transform detects and fixes crooked horizons.

The algorithm leaves naturally dark midtones alone unless the photo is severely underexposed. It detects when color deviations are extreme, like during golden hour, and skips white-balance neutralization to protect the original lighting.

Intensity blends the computed corrections with the original pixels. At 100 percent the full correction is applied. Lower values give a lighter touch for photos that are already close to final.

After analysis the tool shows what it found, such as 'underexposed' or 'crooked horizon'. Each tag corresponds to a specific algorithmic pass. If no tags appear, the photo required no major correction.

The pixel manipulation pipeline compiles to a WebGL Fragment Shader. Your graphics card processes the entire image in parallel instead of looping through pixels sequentially in JavaScript.

Explore Our Tools

Browse all tools