Fix blurry, dark, or faded photos — sharpen, adjust, and colorize — 100% in your browser
Drop your old photo to restore
This tool uses browser-based image processing algorithms. Results are approximate — for professional photo restoration, consider specialized services. All processing happens locally in your browser.
Every fix here is a per-pixel pass over an RGBA buffer pulled from a canvas — no server, no upload, no AI model download. Restoration runs as a fixed six-stage pipeline in this order: sepia removal → brightness → contrast → denoise → sharpen → colorize. Order matters: colour casts are neutralised first so the later passes measure the true tonal range, and sharpening runs last so it amplifies cleaned pixels rather than grain. Brightness and contrast are global (one adjustment for the whole image, derived from its histogram), denoise and sharpen are local (each pixel is compared against its neighbours), and colorize is a tonal tint derived from each pixel's own grey level.
Unsharp masking is why sharpening comes last — it subtracts a blurred copy from the original, so any noise left in the image gets amplified along with the detail.
A faded 1940s family portrait, 1200×900 px, measures an average luminance of 62 out of 255 — badly underexposed. You enable Auto Brightness, Auto Contrast, Denoise, and Sharpen at 50%.
Restore old photos: sharpen blurry images, brighten dark shots, colorize black and white. AI-powered. 100% in-browser, free, no upload.
Photo Restore & Colorize is a six-option restoration toolkit that runs entirely on your device through the HTML5 Canvas API. Load a JPG, PNG, or WebP file and toggle any combination of fixes: Sharpen to recover edges, Denoise to smooth grain, Auto Brightness and Auto Contrast to rebalance faded exposures, Remove Sepia to neutralize yellowed antique prints, and Auto Colorize to add tonal warmth to black-and-white photographs. A strength slider lets you dial sharpening from 10% to 100% (50% by default), and the finished image downloads as a lossless PNG named after your original file. A stacked original-and-restored preview shows both versions before you commit to the download.
Family historians digitizing shoeboxes of prints use it to strip the yellow-brown cast from century-old portraits before archiving them. Flea-market and Etsy sellers improve listing photos of old postcards and documents without paying for desktop software. Genealogists colorize monochrome relatives for reunion slideshows, and scrapbookers rescue underexposed snapshots where camera auto-fixes fell short. Because every pixel operation happens locally in JavaScript, it also suits sensitive material: childhood photos, ID documents, or anything you would rather not upload to a freemium restoration site that watermarks results until you subscribe. If the damage is basic (blur, noise, fading, or tint), this tool covers it; heavily torn or missing regions still need a professional retoucher.
(1) Load your photo; it opens straight from disk into an in-memory image element, so nothing is transmitted anywhere. (2) Pick your repairs and press Restore Photo. The pipeline runs in a fixed order: sepia removal pulls the red channel down and the blue channel up toward each pixel's neutral average; auto brightness measures average luminance with Rec. 601 weights and normalizes it toward midtone 128; auto contrast stretches each color channel between its minimum and maximum; denoise averages each 3×3 pixel neighborhood while skipping neighbors more than 30 levels apart, preserving edges; auto colorize steers dark tones warm and light tones cool; finally an unsharp mask subtracts a blurred copy to sharpen at your chosen strength. (3) Compare the stacked before/after previews and download the restored PNG.
No. Every operation here is deterministic math on pixel arrays, computed by the Canvas API in your browser: channel arithmetic for sepia removal and colorization, luminance statistics for brightness, histogram stretching for contrast, neighborhood averaging for denoising, and an unsharp-mask convolution for sharpening. There is no neural network to download, no GPU warm-up, and no inference server. That brings two advantages worth knowing. First, results are reproducible: the same photo and settings always produce the same output. Second, privacy is structural rather than promised, since the file is read via an object URL and never leaves the tab. The honest trade-off: algorithmic colorize applies gentle warm/cool toning rather than guessing true skin tones, so expect tasteful results, not forensic color reconstruction.
It applies AI-powered filters for: sharpening blurry images (motion blur and out-of-focus), brightening underexposed shots, reducing noise/grain in low-light photos, repairing minor scratches and dust spots, and colorizing black-and-white photographs. Each enhancement is adjustable via intensity sliders so you control the restoration level.
It handles moderate damage well: fading, minor tears, dust, and color shift. For severe damage (large missing sections, water damage covering faces, extreme deterioration), professional restoration services ($50-$300 per photo) remain more effective. This tool is ideal for the 80% of old family photos that need sharpening, brightening, and color correction.
This tool is also known by these tasks — each link opens the same tool with a focused guide:
What do you call a crab that plays baseball?
No paywalls, no signups, no data sold. Built by a solo developer who believes useful tools should be accessible to everyone.
☕Support me on Ko-fi— keep tools free100% of proceeds go towards hosting & building more free tools.