← All tutorialsTutorial · 2 min read
Taking great closet photos
Lighting, background, framing — what the vision AI needs to identify pieces accurately.
The vision AI identifies your closet pieces from photos. Better photos in, better catalog out. Here's what helps and what hurts.
Setup that works
- Plain background — a bed, a hardwood floor, a wall. White, gray, or any solid color the garment contrasts with.
- Soft, even light — natural daylight from a window beats overhead bulbs. Avoid harsh sun + shadow.
- One piece per photo — even if you wear them together. The AI catalogs items individually.
- Lay flat or hang — flat-lay is easiest. Hanging works if the garment isn't fighting the hanger.
- Fill the frame — the garment should take up at least half the photo.
What to avoid
- Crumpled — fold or hang first; the AI struggles with shape-less garments.
- Worn on body — works, but identification confidence drops because the AI also has to parse you out of the frame.
- Multiple items at once — it will pick the most prominent one and label the rest as background.
- Glamour shots — heavy filters, fashion-magazine angles, dramatic shadows all reduce accuracy.
For shoes specifically
Shoot one shoe from the side (3/4 angle is great) on a plain floor. The AI is good at categorizing — sneaker vs loafer vs boot — and excellent at color, but it can't see the brand unless the logo is clearly visible. Add the brand manually if you want it tagged.
Tip
You can re-upload a photo on any item later if the first attempt labeled it wrong. Tap the item in /closet, edit, and replace the photo.