Will any photo work?
No. Blur, lost shadow detail and busy backgrounds can make a pattern worse. Evaluate the source first; generation cannot guarantee recovery of information the photograph does not contain.
Practical guide
PaintGenAI
How to prepare a photo for paint by numbers.
Start with a frame where the subject is clear without tiny details. Simplify distracting backgrounds, inspect important outlines and only then choose the palette. The goal is recognisability, not turning every photo pixel into a separate region.
Request beta access01
The workflow
Prepare and review the pattern.
- 01
Choose a readable photograph
Avoid motion blur, blown highlights and deep shadows over important details. For portraits, inspect eyes, mouth and facial boundaries. If they are already indistinct, palette reduction cannot restore missing information.
- 02
Match the crop to the canvas
Choose the future kit's aspect ratio and leave space around the subject. Remove unnecessary elements or prepare a simpler background in an editor. Keep the original separately so cropping and changes can be compared with it.
- 03
Generate a first pattern
Upload the prepared image to PaintGenAI, select paint by numbers and set the palette, complexity and canvas size. Inspect the source, colour layout and numbered pattern. Gradients will become transitions between a limited number of colours.
- 04
Review the result by subject
Start with the face or other main subject, then inspect the background. If an important outline disappears, try a different source or settings. Simplify a background that breaks into many tiny fragments. Check numbers on a target-size test print before release.
One working context
What stays with the project.
- Original and selected prepared crop, kept distinct from the processed result.
- Target aspect ratio and canvas size.
- A list of details that must remain recognisable.
- Saved variant settings and readability findings.
Practical answers
Project preparation questions.
Should I immediately choose the maximum colour count?
No maximum works for every photo. Start with a considered variant, compare recognisability and paintability, then change one setting at a time.
Are real photo examples available?
Yes. The linked photo gallery shows two credited CC0 photographs processed by the real PaintApp engine, with sources, colour layouts and numbered previews. They are not customer cases or evidence of portrait or physical-print quality. The coffee result requires palette review. The separate drawing demonstrations are not photographs.
Can I print the published demonstration as a finished kit?
It is a reduced-size preview, not a print-ready kit. Your own project needs calculation and review, readable numbers, approved materials and a test print.
PaintGenAI