# Search part of the screen

> Limit where Optics looks for an element, so repeated text can't be matched in the wrong place and vision runs faster.

An **area of interest** (AOI) is a rectangle on the screen. Give one to a step and Optics only accepts an element inside it. Use it when the same text appears in several places, such as an *Add to cart* button that also appears in a product description, or to make OCR faster by reading less of the screen.

## Set an area

Four values describe the rectangle, each as a **percentage** of the screen, from 0 to 100:

| Parameter | Meaning | Default |
|---|---|---|
| `aoi_x` | Left edge, from the left of the screen | `0` |
| `aoi_y` | Top edge, from the top of the screen | `0` |
| `aoi_width` | Width of the area | `100` |
| `aoi_height` | Height of the area | `100` |

The defaults cover the whole screen, which means "no area". Pass the values by name:

`modules/modules.csv`:

```csv
module_name,module_step,param_1,param_2,param_3
Add To Cart,Press Element,Add to cart,aoi_y=75,aoi_height=25
```

This looks for *Add to cart* only in the bottom quarter: it starts 75% of the way down and covers the remaining 25%.

Some common areas:

| Area | Values |
|---|---|
| Top half | `aoi_height=50` |
| Bottom half | `aoi_y=50,aoi_height=50` |
| Right third | `aoi_x=67,aoi_width=33` |
| Centre | `aoi_x=25,aoi_y=25,aoi_width=50,aoi_height=50` |

The edges must stay on the screen: `aoi_x` plus `aoi_width`, and `aoi_y` plus `aoi_height`, can be at most 100.

## How it works

- **OCR and image matching** crop the screenshot to the area before they search. They read less, so they're faster too.
- **The UI tree** can't be cropped. Optics finds the element as usual, then keeps it only if its centre is inside the area.

Each step that uses an area saves a screenshot with the rectangle drawn on it (`…_with_aoi.jpg` in `execution_output/`), so you can check you chose the right part of the screen.

## Keywords that accept an area

`Press Element`, `Enter Text`, `Enter Number`, `Clear Element Text`, `Get Text`, `Swipe From Element` and `Scroll From Element`. Checks such as `Assert Presence` don't take an area.

In the Python library, only `press_element` accepts area parameters.

## Area or index?

Both pick one match out of several. Use `index` when you know which match you want by order (the second *Add to cart* button). Use an area when you know *where* it is, which keeps working when items are added or reordered. You can combine them.

## See also

- [Find elements by text and image](https://optics-framework.org/docs/elements/text-and-images)
- [Define elements](https://optics-framework.org/docs/elements/define#pick-one-of-several-matches)
- [Action keywords](https://optics-framework.org/docs/reference/keywords/actions)