Colorize Photo: Add Color to Black-and-White Pictures
Add plausible color to an old photo. Free AI: 1/day, or 2 signed in, at 1K. Usually about 30 seconds. Colors and details may differ.
No censored, paywalled, private or adult content.
Try these examples
Saved results · no AI call, no credits used.
Colorize a photo while keeping the original
A black-and-white photograph records light and shape, but it does not tell an AI what every original color was. A dark jacket might have been blue, brown or green. This colorize photo tool adds a plausible interpretation so you can see a familiar scene differently. Keep the original scan alongside the result. The colored copy is an edited version, not a verified record of clothing, skin tone, paint or the weather on that day.
Start with a clear scan or the best photograph of the print that you have permission to use. Avoid glare and include the complete frame. If your main goal is repairing blur, scratches or fading while retaining a monochrome look, use old photo restoration instead. For general clarity problems in an existing color photo, try the photo enhancer. You can return to the image unblur homepage to compare other restoration options.
How to add color to a black-and-white photo
Upload one black-and-white picture in the workbench above. Choose a resolution you can access and press Enhance photo, then review the result when processing finishes. A request usually takes about 30 seconds, although some take longer. Use the comparison slider to look at the same feature in both versions. Inspect the complete image first, then concentrate on faces, hands, lettering, clothing and small objects. Save a copy only after checking the parts that matter to you.
Colorization uses AI after you press Enhance photo. Saved before/after examples are paired model edits; viewing them spends no credits. Your own result may differ because visible source detail affects reconstruction.
Natural-looking color is still a guess
Colorizing old photos can make grass, stone and sky look familiar, but familiar does not mean historically accurate. An AI may choose a believable skin tone or uniform color without evidence that it is correct. If a photograph matters for family history, write down what you know separately and keep the unedited source. You may have a relative’s description or another color photograph that is more reliable than a model’s choice. This tool cannot verify those facts for you.
Look for color spilling across boundaries, unusually bright eyes, patchy skin or a sky that feels unrelated to the lighting. A restrained palette can still contain wrong choices. Do not describe the output as the original colors restored unless you have independent evidence. For publication, label a colorized old photo as an AI-assisted edit when that distinction matters to readers. A useful result can be an interpretation without pretending to establish what the scene really looked like.
Check faces, text and objects before sharing
The instruction asks the model to preserve the people, expressions, objects and framing while adding color. That instruction is not a guarantee. In our sample comparisons, the model also reconstructed fine texture in skin, hair, fabric, grass and masonry. Even a natural-looking result may change a small feature. Check a face against the source rather than judging only whether it looks attractive. Keep the original if a changed expression or feature would be misleading.
Compare signs, captions and dates with the original. AI may redraw lettering rather than only tint pixels. For reading documents, use text deblur and verify important characters. Do not use generated detail as identity, transaction or historical evidence.
Choose a useful resolution and scan
The current resolution settings are 1K and 2K, targeting approximately 1 and 4 megapixels respectively while preserving the source aspect ratio. Actual downloaded dimensions can vary, so inspect the saved file rather than assuming a fixed long edge. A larger output is not proof of additional authentic detail. If a scan is already suitable for your intended screen display, focus on whether the color interpretation helps rather than selecting a larger setting automatically.
For printing, use the print-size calculator to compare the actual pixel dimensions with your paper size and viewing distance. For a separate enlargement task, the image upscaler explains the difference between extra pixels and recovered detail. Scanning the original print again may be more useful than processing a small screenshot. Keep borders and faces intact when capturing the print, and avoid repeated exports that introduce new compression before the model sees it.
What it can and can't do
This tool can add plausible color to a recognizable black-and-white scene. It cannot determine verified historical colors, guarantee unchanged facial detail or rebuild information that was never captured. Very dark, tiny or badly damaged regions can receive poor color choices or reconstructed texture. It does not uncover deliberately censored, private or paywalled content. Use only photographs you own or are authorized to process, and consider whether the people pictured would want an edited version shared.
Free Colorize uses the same daily AI allowance as other tools: once without signing in, or twice when signed in, at 1K. It is not an extra allowance for each page. Credit-based 1K uses 1 credit and 2K uses 2 credits; batch upload is available to members. See plans and credits for access. Keep both versions after downloading so a future viewer can distinguish the source photograph from the color interpretation you chose to keep.
How it works
- 1
Upload the best scan
Choose a black-and-white photo you own or have permission to process.
- 2
Start AI colorization
Review resolution and usage, then request AI colorization. Colors are plausible interpretations, not verified historical facts.
- 3
Compare and keep a copy
Inspect faces, words and small objects. Save the edited result alongside the original.
What it can and can't do
Useful starting points
- Recognizable black-and-white portraits
- Old photos with visible shapes and useful detail
- Landscapes and buildings where plausible color is helpful
Limits to keep in mind
- Colors are inferred, not verified historical facts
- Faces, text and small details may be redrawn
- Severe damage or absent detail may not improve
Frequently asked questions
Can AI colorize a photo accurately?
It can add plausible colors, but a black-and-white source does not establish every original color. Treat the result as an interpretation.
Is it free to colorize old photos?
The shared daily AI allowance is 1/day without signing in or 2/day signed in, at 1K. Other tools use that same allowance.
Does Colorize change faces or text?
It can reconstruct fine details despite the preservation instruction. Compare important features, letters and numbers with the original.
Are the generated colors historically accurate?
Not necessarily. AI adds plausible colors based on visible clues. Compare with reliable references if color accuracy matters.
Can I colorize more than one photo?
Members can use batch upload. Each image follows the normal resolution and credit rules, and each result needs visual review.
Can I print a colorized photo?
Inspect the downloaded dimensions and use the print-size calculator. More pixels or dpi metadata do not verify new detail.
What if the photo is scratched or very blurry?
Consider old photo restoration first and keep the source. Severe damage may produce invented features rather than a reliable repair.