Today, we will explore the concept of dynamic prompting – a straightforward approach to crafting well-structured prompts. While some refer to it as additive prompting or structured prompting, we will refer to it as dynamic prompting. So, let’s jump right in!
Dynamic prompting involves incorporating an array of tags or sections into your text prompt, which you may have already learned about in the past. To summarize the structure of a prompt in MidJourney, here is a simple visual aid. While it may be referred to as Additive Prompting or Structured Prompting by some, the technique remains the same and allows for better image control during iteration.
This approach has been in use since early 2022 and has gained popularity among individuals such as Linus, Nick, Julie, Lars, and myself. The next generation of prompt-based products will likely include prompt-precursors, which are already available through tools like PromptHunt by Addison.
When creating dynamic prompts, our main focus is on the text prompt section, which we break down into smaller components that can be manipulated. This technique is applicable not just to Midjourney, but to other tools like Dalle2, StableDiffusion, and Leonardo as well.
Here is a basic prompt structure as an example, where the style words are linked to Wikipedia or Google image search to provide visual context.
A straightforward breakdown
To utilize this technique, we require a minimum of two sections in our prompt, although we can add as many as necessary to make it more dynamic. It is essential to separate each section to better understand their roles when making modifications.
/imagine <Type>, <Camera Angle>, <Room>, <Style>
Putting it all together, this is what the prompt will look like. We will throw this into Midjourney to see what we get back. Do this demonstration with me on Midjourney to see what you get and practice your learning.
Photograph, High angle, Industrial Kitchen Kitchen island, Concrete flooring
The Structure of the Prompt
Below is a comprehensive prompt structure breakdown, along with an explanation of each tag. Keep in mind that these tags can be customized as per your preference; this is only a sample structure.
<Type>, <Camera Angle>, <Style>, <House Type>, <Focal Point>, <Textures>, <Detail>, <Color Palette>, <Architect>, <Type of House>, <Lighting>, <Location>, <Time of Day>, <Mood>, <Architecture>
Let’s generate something
Let’s attempt an example by either manually populating the prompt or using the following prompt shared by Nick Floats in ChatGPT to generate a starting point: