Automatic colour estimation
Recognises visual context and predicts believable colours for people, sky, landscapes, buildings, clothing and everyday objects.
Convert black and white, grayscale, sepia and faded old photos to natural-looking color in your browser, compare Original vs Colorized, fine-tune colour strength, and download high-quality JPG, PNG or WebP images.
Choose one image, adjust the result, and process it locally.
Predicted colours are creative estimates, not verified historical facts. The output is newly encoded and does not intentionally copy the source photo’s metadata.
Move the slider to inspect the estimated colour result.
Choose a black and white photo, then select Colorize Photo. The vertical comparison divider will appear after processing.
This online AI photo colorizer works as a black and white to color converter for old family photos, historical images, monochrome portraits, scanned prints and faded photographs. It combines estimated colour with the source image’s luminance detail so faces, edges, texture and film grain remain clearer in the downloadable colorized photo.
Recognises visual context and predicts believable colours for people, sky, landscapes, buildings, clothing and everyday objects.
The selected photo remains in the browser. No source image or generated result is uploaded by this tool for processing or storage.
Use a before-and-after slider to inspect faces, objects, edges and uncertain areas before saving the result.
Create subtle documentary colour, a balanced natural look or stronger saturation without re-running the entire model.
Retain the original luminance structure to avoid a soft, low-resolution appearance after the predicted colour is enlarged.
Export PNG, JPEG or WebP with quality and maximum-dimension controls and a safe CorrectionCopy-prefixed filename.
Follow the same workflow on a phone, tablet, laptop or desktop browser.
The browser must be able to decode the selected source and encode the chosen output. Availability can differ by browser, operating system and device.
| Format | Upload | Download | Important notes |
|---|---|---|---|
| JPEG / JPG | Yes | Yes | Widely compatible; output uses adjustable lossy compression. |
| PNG | Yes | Yes | Lossless output; files may be larger and transparency is preserved where applicable. |
| WebP | Yes | When supported | Efficient compression; availability depends on the browser encoder. |
| AVIF | When supported | No direct AVIF option | The browser must decode the input; export as PNG, JPEG or WebP. |
| GIF, SVG, HEIC, TIFF and RAW | No | No | Convert these files to JPEG, PNG, WebP or supported AVIF first. |
Treat the generated image as an interpretation. Keep the original scan, preserve archival records, compare documented references and label the result as colorized when publishing it.
Crop borders, correct rotation, improve exposure and remove severe stains before colorization. Scratches, tears and missing details require separate restoration work.
Your files are protected and safe; we never collect or share your data. The selected image is read from your device, processed in browser memory and exported as a local download. No server-side photo processing or temporary upload directory is used.
The page downloads the browser processing code and a large colour model after you start. Those requests do not contain the selected image, filename or generated photo. Network access is required when the files are not already cached.
Preview URLs, canvases and generated image data exist only in the current browser tab. Resetting, replacing the photo or closing the page releases those temporary resources where supported.
The new image does not intentionally copy embedded location or camera metadata. However, faces, addresses, signs, documents and landmarks visible in the picture can still reveal personal information.
Check the connection, disable network filtering for the page and retry. A hard refresh can clear an incomplete cached download. For the most reliable production deployment, host the required processing files and model on CorrectionCopy’s own domain.
Close other tabs, select a smaller maximum dimension, use a lower-resolution source or retry on a device with more available memory.
Increase Colour Strength. Very faded images and unusual subjects may still receive limited colour when the model is uncertain.
Lower Colour Strength and return Warmth to neutral. Subtle settings are usually better for documentary and archival photographs.
Keep Preserve original detail enabled and avoid enlarging a very small scan far beyond its source dimensions.
Wait for a successful result. If export fails, choose PNG or JPEG, reduce the maximum dimension and process the photo again.
Answers about colour accuracy, privacy, supported images, local processing, performance and downloadable results.
A photo colorizer estimates plausible colours for a black and white image. It analyses visual patterns such as faces, clothing, sky, vegetation and objects, then creates a new colour version while keeping the original photo unchanged.
Choose or drop a supported photo, review the preview, select colour strength and output settings, then choose Colorize Photo. Compare the result with the original and download the new image as PNG, JPG or WebP.
Not necessarily. The tool predicts believable colours from learned visual patterns, but it cannot know the true historical colours of clothes, vehicles, buildings or personal items. Use documented references when historical accuracy matters.
No photo content is uploaded by this tool. The selected image, processing and export remain in the browser. The page downloads the processing components and colour model, but your image is not sent with those requests.
The colour model must be available on the device before local processing can begin. The browser downloads it once for the current session and may reuse its network cache later. The initial download is much larger than an ordinary page asset.
The tool accepts JPEG, PNG, WebP and AVIF images when the browser can decode them. Animated images, SVG files and unsupported camera formats are not accepted. Format support can vary by browser and operating system.
You can download PNG, JPEG or WebP when the browser supports that encoder. PNG is lossless and may be larger. JPEG and WebP provide adjustable quality and generally create smaller files.
No. The original file stays unchanged on your device. The tool creates a separate colourized copy for preview and download.
Yes. The tool converts the source to a neutral luminance image for analysis, so sepia and faded photographs can be processed. Strong stains, colour casts, scratches and severe fading may still affect the prediction.
No. Skin, hair, eyes and clothing are estimates and may be uneven or incorrect, especially in damaged, low-resolution, overexposed or unusual photographs. Adjust colour strength or edit the downloaded result when needed.
The model may be uncertain about small objects, uncommon materials, shadows, damaged regions or ambiguous scenes. Increasing colour strength can make colours more visible, but manual editing may be required for precise local corrections.
No. It adds estimated colour but does not reconstruct missing faces, remove scratches, sharpen severe blur or repair torn areas. Use a dedicated restoration or image editor for those tasks.
This option uses the original image luminance as the fine-detail layer and applies the predicted colour over it. It usually keeps faces, texture, grain and edges sharper than simply enlarging the low-resolution model output.
Yes. Colour Strength controls how much predicted colour is applied. Lower values create a restrained or partially colourized appearance, while higher values increase saturation and visual intensity.
Yes, on current browsers with local file access, canvas export, WebAssembly and enough memory. Performance varies considerably by device. Older phones and memory-limited browsers may take longer or fail with very large photos.
The first run includes downloading and preparing a large on-device colour model. Processing speed then depends on the device processor, browser, available memory and image dimensions. Keep the page active while the browser completes local processing. The model analyses a compact working image, while the selected output can retain the original dimensions.
CorrectionCopy does not impose an artificial number of uses. Practical limits come from browser decoding, canvas dimensions, available memory and processing time. Very large files or extremely high pixel counts may need to be resized first.
The result is newly encoded using the selected format, quality and maximum dimensions. Compression, colour complexity and resizing can make it smaller or larger than the source even when the visible dimensions are similar.
The colourized output is newly created from rendered pixels and does not intentionally copy the original EXIF, GPS, XMP or IPTC blocks. Browsers may still add basic structural information required by the image format.
You can cancel component or model loading and prevent an in-progress result from being displayed. Once device computation has already started, the underlying browser operation may need to finish internally before its resources are released.
Yes. A grayscale, monochrome or black and white image can be processed as long as the browser can decode its JPEG, PNG, WebP or AVIF file. The result is an estimated color version that can be adjusted and downloaded separately.
Clear portraits, family photos, street scenes, landscapes and scanned prints with visible subjects usually work best. Heavy blur, missing detail, extreme fading, scratches and unusual historical objects can reduce color accuracy.
Edit, resize, convert, compress and prepare related image files with active tools from the CorrectionCopy catalog.
Crop, rotate, adjust and edit images with simple controls.
Open ToolConvert images between popular file formats online.
Open ToolInspect photo EXIF and GPS data, remove metadata or edit selected properties.
Open ToolCombine one or more images into a PDF document.
Open ToolCreate ICO, PNG, Apple touch and PWA icons with manifest and ZIP downloads.
Open ToolConvert PDF pages into downloadable JPG images.
Open ToolCreate customizable QR codes for links and text.
Open ToolBundle selected files into a downloadable ZIP archive.
Open ToolAdd custom header and footer text to PDF documents.
Open ToolInsert page numbers into PDF files with flexible placement.
Open ToolApply text or image watermarks to protect PDF pages.
Open ToolCompare two PDF documents and identify page-level differences.
Open ToolCrop or resize PDF pages and image files precisely.
Open ToolRemove selected pages from a PDF and save the result.
Open ToolCombine multiple PDF files into one organized document.
Open ToolReorder, arrange and manage PDF pages visually.
Open Tool