About this tool
Breathe new life into vintage, grainy, sepia-toned, or scratched photographs with AI restoration logic.
The Old Photo Restorer runs a scanned or photographed print through a three-step canvas pipeline — contrast and brightness lift, optional sepia removal by luminance conversion, and a 3×3 sharpening convolution — and shows the result against the original in a drag-to-compare slider. It is meant for anyone digitising family albums who wants faded, flat or yellowed prints to read clearly again without learning a photo editor. Images up to 15 MB are processed in the page and downloaded back as PNG.
Open Old Photo Restorer on AltFTool — it loads instantly in your browser.
Press Select Vintage Photo or drag a scan onto the upload panel — JPG, PNG and WebP up to 15 MB.
Tick Enhance Fine Details and/or Remove Sepia Tone, then press Restore Image to run the contrast, de-sepia and sharpening pass.
Drag the Before/After split slider to inspect the change, then Download the result as restored-photo.png or use Copy Image.
Drag the divider across the image to judge exactly how much the restoration changed, instead of guessing from a thumbnail.
The de-sepia option converts using 0.299R + 0.587G + 0.114B rather than averaging channels, so tonal relationships survive the conversion.
An unsharp-style 3×3 kernel recovers edge definition that scanning softens, applied after the contrast lift so it does not amplify a colour cast.
Raise contrast and brightness first, then sharpen — that order matters, because sharpening a flat image just amplifies noise. This tool applies a 1.25× contrast multiplier with a +10 brightness offset when detail enhancement is on, then runs the sharpening pass.
Convert the image to greyscale using luminance weights, which discards the colour cast while keeping the correct brightness for each tone. Turn on the sepia-removal toggle and every pixel is mapped to 0.299R + 0.587G + 0.114B.
Up to 15 MB per image. The sharpening step is skipped automatically on pictures larger than about 4 megapixels (4,000,000 pixels) so the browser tab does not freeze during the convolution pass; tone and sepia correction still apply at any size.
No — this is a tonal and sharpness restoration, not inpainting, so physical damage such as scratches, creases and missing corners stays in the image. For torn or heavily damaged originals you will still need a clone or heal tool in a full editor after running this pass.