MiniPx
Add to ChromePricingBlogAccount
AI Tools — Cloud HD

Add color to a black-and-white photo

Upload a monochrome photo and the model adds colour region by region — skin, sky, fabric, foliage — while the light-and-dark structure of your photograph passes through untouched. The file is uploaded for this run only, deleted on completion, never used for training.

Add color to a photoPro · runs on our GPU
Black-and-white photograph of a long-haired tabby cat, before color was added
Sample — processed with ddcolor-modelscope 8565d1f7c686

How it works

  1. Pick a black-and-white photo — JPEG, PNG or WebP, up to 25 MB and 24 megapixels. Colour is chosen automatically, region by region.
  2. It goes to our GPU for this one run, and nothing is kept afterwards.
  3. Download the colour version at full resolution. One run of your monthly cloud allowance.

Pro to run it on your photo →

This is the same engine as the photo colorizer; this page explains how it decides.

All Image Tools

AI Tools

Pro Tools

Compress

Convert

Resize

Compress to Size

Compress for Platform

Tools & Utilities

By Device

Compare

Solutions by use case

For Developers & Web

For Sellers & Ecommerce

For Job Applicants & Govt Forms

For Creators & Social

For Work & Documents

Add Color to a Photo — Free, Fast & Private

A black-and-white photo is not missing data so much as ambiguous about it. Every grey pixel in your photo was once a colour that lost its hue on the way to film. Many different colours map to the same grey, so “adding colour back” is really a choice among the candidates — and DDColor makes that choice the way an experienced restorer would, by knowing what the world usually looks like. Grass is overwhelmingly green; a 1960s kitchen wall is a toss-up. The demo above shows the honest consequence: the cat’s colours are convincing because cats have statistics; its blue eyes came back amber-green because pale eyes photograph to a grey the model reads as the more common colour.

Luminance is sacred. The tool changes chroma only. Composition, focus, grain, the exact play of light that made the photograph worth keeping — all of it passes through byte-for-byte in structure, returned as a full-resolution PNG. That is also why a soft or damaged photo comes back soft or damaged, in colour: fixing those is a different job, and the old-photos page covers how to sequence it.

When a colour must be exact, finish by hand. Regimental uniforms, team kits, flags, a bridesmaid dress the family still argues about — the model cannot know these, and a confident wrong answer is worse than grey. The practical workflow: let the AI do the whole frame, then repaint the one region whose colour you actually know in any editor. The AI base layer turns an hour of manual colorization into minutes of correction.

What it costs and where it runs. This tool runs on our GPU server — the model is too heavy for a browser — and is part of MiniPx Pro’s metered cloud runs (150 a month on Monthly, 200 on Annual, meter shown before every run). Your photo is encrypted in transit, deleted on completion and in all cases within 24 hours, and never used to train anything. Everything else on MiniPx runs on your own device.

How it works

  1. Open the tool: Go to minipx.com/add-color-to-black-and-white-photo. The sample slider shows the real pipeline output.
  2. Upload your photo: A black-and-white JPEG, PNG or WebP up to 25 MB. Pro accounts see the upload button.
  3. The model adds color: DDColor predicts per-region colour on our GPU server, leaving the light-and-dark structure untouched.
  4. Download and adjust if needed: The PNG comes back at full resolution. If one object needs a specific colour, correct just that region in any editor afterwards.

Frequently asked questions

How does the AI know which colors to add?
It doesn’t know — it infers. DDColor was trained on millions of colour photographs, so it has strong statistical expectations: foliage lands on green, sky on blue, skin on warm tones, weathered wood on brown-grey. For each region of your photo it predicts the colours that would most plausibly produce those exact grey values.
Can I choose the colors myself?
Not in this tool — it is fully automatic. If a specific colour matters (a uniform, a flag, a wedding sari), colorize first and then adjust that one region in any photo editor; the AI output makes a good base layer for manual correction.
Why did it get a specific object wrong?
Because grey carries no colour information. A medium-grey dress could have been red, blue or green in life — they can all photograph to the same grey. The model picks the statistically likeliest option, which is exactly why unusual objects are where it misses.
Does it change anything besides color?
No. The luminance — the light-and-dark structure that IS your photograph — passes through untouched, and the result is a PNG at the same resolution as your input. Sharpness, grain and damage are all preserved exactly.
Is my photo stored?
No. It is uploaded over an encrypted connection for this run only, deleted on completion — in all cases within 24 hours — and never used to train anything.