# Resolution

URL: https://fulla.dppst.com/docs/generative-fill/resolution/

> How Fulla picks the size of the generated image, up to 4K, so the result is as sharp as your photo.

With **Auto**, the default, Fulla asks the model for an image about twice the size of your selection, so the result looks as sharp as your photo and is never stretched.

## How Auto chooses

- **1K** for a small selection.
- **2K** for a medium one.
- **4K** for a large one, when the model offers it (2K at most for OpenAI models).

## Choose a size yourself

Open the **settings button** next to the model button and pick a **Resolution**. A size you choose is always used. A setting the model doesn't offer isn't shown.


A larger image costs a little more: see [What it costs](https://fulla.dppst.com/docs/generative-fill/cost/).