---
title: Occlusion
description: >-
  What occlusion means in computer vision and AR, and why correctly layering
  virtual products behind hands, hair, and accessories is essential for
  realistic try-on.
shortDefinition: >-
  Hiding the right parts of a virtual object behind real-world elements that are
  closer to the viewer.
relatedTerms:
  - image-segmentation
  - augmented-reality
  - garment-transfer
  - image-inpainting
heroImage: /glossary-images/occlusion-comparison.webp
keywords:
  - occlusion
  - occlusion handling
publishedAt: '2026-06-10'
updatedAt: '2026-06-10'
noindex: false
canonicalPath: /glossary/occlusion
---

## What is occlusion?

**Occlusion** is when an object closer to the viewer hides part of an object behind it. In virtual try-on and [augmented reality](/glossary/augmented-reality.md), occlusion handling means rendering virtual products with the correct depth ordering relative to the real scene: a hand in front of a shirt must stay in front of the *virtual* shirt too.

When occlusion fails, the virtual item appears painted *on top of* everything: a ring covering the finger that should hide its back, or a generated jacket overwriting the hair that falls over it. It is one of the most immediate giveaways that an image is fake.

![Correct vs incorrect occlusion handling in a try-on image](/glossary-images/occlusion-comparison.webp)

## Common occluders in try-on

- **Hair** falling over shoulders and collars
- **Hands and arms** crossed in front of the torso
- **Accessories** such as bags, scarves, or jewelry worn over clothing
- **Other garments**, like an open jacket partially covering the top underneath

## How it's handled

In live AR, depth ordering comes from real-time hand and body tracking plus depth estimation of the scene. In image-based pipelines, [image segmentation](/glossary/image-segmentation.md) identifies occluding elements so that [inpainting](/glossary/image-inpainting.md) regenerates the garment *behind* them. The occluders are excluded from the mask and preserved from the original photo.

<Faq>
  <FaqItem question="Why is occlusion harder for apparel than for rigid products?">
    Rigid items like glasses have predictable occluders (ears, nose) at known positions. Clothing covers large, deformable regions where hair, hands, and layered garments can overlap anywhere, so the occlusion mask must be computed per image.
  </FaqItem>
</Faq>