How to take photos that make good AI 3D models
The quality of a 3D model depends a lot on the input photo. Following the tips below will give you better results from your very first try. Keep in mind that AI has limits, so the result may still need some editing.
- 01
One main object
Each photo should contain a single clear object. Several objects or a busy scene make it hard for the AI to know what to build.
- 02
Clean, contrasting background
Place the object on a plain background (white, gray, one color) that differs from the object. A messy background tends to stick to the model.
- 03
Bright, even lighting
Shoot in soft light and avoid harsh shadows or backlight that hide details.
- 04
Show the whole object
Do not crop part of it at the edge of the photo and do not let other things cover it. The AI has to guess whatever it cannot see.
- 05
A slight angled view
Shoot from the front, tilted slightly upward (about 3/4 view); this usually shows both the front and the top.
- 06
Sharp image, enough resolution
Avoid blurry, shaky or very small photos. A few hundred pixels on each side is usually enough to start.
- 07
Avoid transparent or very reflective objects
Glass, water and mirror-like metal make it hard for the AI to recognise the real shape. If you must, use diffuse lighting.
- 08
Very thin details are hard
Thin wires, fur, hair and small leaves usually give a rough mesh or missing detail.
Photos that work and photos that do not
- Good: toys, cartoon characters, figurines, household items, shoes, simple products, handmade crafts.
- Harder: objects with many small parts, shiny or transparent surfaces, or photos taken in a cluttered scene.
Once you have a model
You can download several formats depending on your goal. See the comparison of GLB, OBJ, STL and PLY to pick the right one for Blender, Unity or 3D printing.
Have a suitable photo? Try it now
Create a 3D model