> ## Documentation Index
> Fetch the complete documentation index at: https://docs.glasswarp.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Vision

> How your agent sees the screen — screenshots, observe, and dirty rects.

Glasswarp's eyes are designed for fast agent loops: fresh, native-resolution
frames with the option to fetch only what changed.

## Screenshot

A high-fidelity, native-resolution JPEG — fresh (\~50 ms capture or less), and it
works **with or without** an active WebRTC stream.

```python theme={null}
frame = gw.screenshot(sid)
with open("screen.jpg", "wb") as f:
    f.write(frame.jpeg)

# Optional: downscale and/or crop to a region of interest.
roi = gw.screenshot(sid, max_width=1280, x=100, y=200, w=800, h=600, quality=80)
```

## Observe — one round trip

`observe` returns everything an agent step needs in a single RTT: a JPEG, the
accumulated dirty regions, and grounding targets (Set-of-Mark annotated when
`mark=True`).

```python theme={null}
obs = gw.observe(sid, max_width=1280, mark=True)
# obs.jpeg    — frame (SoM-marked when targets present)
# obs.dirty   — regions changed since last poll
# obs.targets — click targets from host UIA (may be empty)
```

Use `observe` as the default "look" in your loop — it's the leanest way to get
frame + change + grounding together.

## Dirty rects (VP0)

For efficient loops, poll only the regions that changed since your last call:

```python theme={null}
rects = gw.dirty_rects(sid)   # accumulated DXGI dirty rects
if not rects:
    pass  # nothing changed — skip the model call
```

<Tip>
  Skip a model call entirely when `dirty_rects` is empty, and crop screenshots
  to the changed region. This is the cheapest way to run a tight agent loop —
  see [Build an agent loop](/guides/agent-loop).
</Tip>

Continue to [Grounding](/concepts/grounding).