To generate an image, you need to write a prompt.
The more specific the prompt is, the more detailed and accurate the image created is.
You can use three cues, Who, Action and Where to create a specific image.