The prompt for this image was
“Portrait of a British man holding an umbrella, facial asymmetry, striking features, tack sharp, rainy weather, fine-art photography, 180mm f/1.8, by Steve McCurry”
Let’s go over the individual parts of the prompt and look at their effects in detail.
The very first word in the prompt is already one of the most important - especially to get good faces. Experience has shown that DALL-E often gives extra weight to terms at the beginning of the prompt, so it generally makes sense to write there what the overall picture is that we want to have.
In our case, we want to have a portrait, that is, a picture with a single person and especially their face in the foreground.
Next, we describe the object to be depicted, in this case a British gentleman holding an umbrella. How specifically we describe what we want to see can vary. In general, images that are more specific are often of better quality, but too detailed descriptions frequently lead to very poor images.
Next we add modifiers that change the object in the image. Here we write things that should be included in the overall impression of the image, so we separate them from the actual object description with a comma.
For prompts in general, you can often be quite loose in what you actually write. If you want to focus on something you just have to bring it into the prompt. So you might write "serious expression", not because you actually want the person to have a serious expression, but because you want your result to be influenced by the kind of photos that mention it. It will have a different quality.
Same for mentioning skin. Photos that mention the skin will most likely have more visible structure in the skin.
Similar to previous modifiers, we add a few more expressions that should affect the overall look of the image.
<aside> 💡 For a harmonious image, try to use modifiers that fit the theme of the image. For example, if our person is holding an umbrella, we make the weather rainy.
</aside>