Skip to main content

Prompt Basic Structure

In Midjourney's official introduction document, the simplest Prompt structure is described as follows:

But I think the above Prompt is too simple ๐Ÿ˜‚, instead Advanced Prompt is more suitable for learning:

Image prompt(optional)โ€‹

Prompt is divided into three parts. The first is Image, which is optional and can be filled in or not.

There are many ways to play with this Image URL, such as merging two images into one, you can see the following example, I uploaded a picture of a marble statue and a flower to Discord:

Then, after giving Midjourney a blend (more on this in the subsequent section on blends), the following image can be generated:

One thing to note:

  • You are entering the URL address of the image and it must be publicly accessible.
  • Image formats are only supported for png, gif and jpg.
  • You can only put a maximum of two image URLs, or one image URL + one paragraph.

If you don't want to upload images to some image bed server, you can also use Discord's image bed, where you can privately chat with Midjourney Bot and send the image to Discord first by uploading it and then right-clicking to copy the image link, like this:

![](/img/AnimatedImage (2).gif)

Text Promptโ€‹

This is the description of the image you want AI to generate. There are many techniques involved, which will be explained in detail later. Here I would like to point out that Midjourney is a bit different from ChatGPT:

Cannot understand sentence structure and grammar: Midjourney cannot understand what you say like ChatGPT does. By analogy, Midjourney is more like imperative programming while ChatGPT is more like declarative programming. You need to give Midjourney complete instructions for it to generate satisfactory results.

Midjourney has prohibited words designed: Some words cannot be inputted, such as ๐Ÿ‘ emoji because this emoji is slang for butts and often leads to the model generating some undesirable images, so it has been banned too.

Parametersโ€‹

Midjourney has about 20 parameters, and I'll cover a few common ones in the next chapter. Once you understand the basic parameters, you should be able to generate good graphs with Midjourney.