- Architecture Insights
- Midjourney Basic Prompting (Architecture & Landscape Guide)
Midjourney Basic Prompting (Architecture & Landscape Guide)
Welcome to this week’s edition of Architecture Insights.
Midjourney has been around for over a year now and it’s image-generating capabilities are an asset to every designer, lets see how to setup it up and make use of its basic features…
This past week
1. OpenAI, the company behind ChatGPT and DALL-E fires and then 4 days later rehires their CEO, Sam Altman. (Read)
Summary: Last Friday, November 17th OpenAI fired its CEO after a consensus decision by its own board of directors. A petition of over 700 OpenAI employees was signed that stated they quit unless Sam was rehired. 4 days later he was rehired as CEO. Speculation still remains as to the exact reasoning behind his firing with many different theories circling the internet.
2. ChatGPT releases a free voice feature. (Read)
Summary: This is the equivalent of Apple releasing Siri. Another feature making ChatGPT more accessible.
It’s safe to say that Midjourney is the leader in the AI image-generating space. It is the most advanced platform that provides extensive customization options to create high-quality images. This makes Midjourney an ideal tool for designers to create architectural renderings, gradually reducing the need for non-AI rendering software.
There are many similar applications out there such as DALL-E, another popular tool. This time we will focus solely on Midjourney and how to set it up for use.
If you haven't used midjourney before, you'll need to create a Discord account, this is where midjourney runs from.
4 plans are offered at the time of writing this which go for $10, $30, $60, and $120.
We will explore features that are available in all of the plans including basic.
Midjourney starts working after these few simple steps: Create an account on discord, join the midjourney discord server, join one of the channels (for starters the #newbies channel will be your likely start), and start typing out prompts to tell the AI what you want it to generate.
The most commonly used prompt to generate images starts with /imagine. Proceed by telling midjourney what to generate. Let’s explore this by prompting it to generate “a minimalistic house covered in snow with a dark and modest theme.” A fairly generic prompt.
These are the 4 images it generated:
We used a short prompt to show that when you don’t give a lot of detail to your prompt you will essentially get what you asked for. Regardless, the quality is great and this can still be used as a good starting point.
Let’s try a similar but more detailed prompt: “A photorealistic image of a minimalistic house covered in snow with a dark and modest theme. The home features a chimney on a gable-style roof. The home also has a porch surrounding the entire outside of the home.”
There is clearly some more detail in these images as that is what we asked for.
For more specific results it is important to consider things like environment, colour, lighting, mood, composition, point of view, styles and themes. Making mention of these in your prompt will generate better results.
Under each image, you are given the following options U1, U2, U3, U4, and V1, V2, V3, V4. The U will upscale your image based on the selected number, the V will create similar versions of the selected number. For instance, U3 will upscale our third image, V2 will create 4 more versions similar to our second image.
If we select “U” we are given additional options once our image is upscaled. Vary will slightly change your image, upscale will increase it in size, and zoom with keep the image the same but zoom out and add more parameters.
Along with the /imagine command, here is a list of commonly used commands to get more out of your results.
/describe: Midjourney offers a convenient "describe" command that allows you to upload any image you want. Once uploaded, the platform provides four unique descriptions of the image, which can serve as prompts for generating a new image that matches the style of the original.
/blend: Midjourney offers a useful feature that allows you to blend two images together by typing the “/blend” command. Once you do, two boxes will show up, and you can add your desired images to blend their styles. Keep in mind that Midjourney will use the first image as a base and the second image's colour palette and style to create a new and unique image.
This guide is intended for beginners to understand the basics of using Midjourney, however, one of the next and more intermediate steps to improving your prompts is parameters, without getting into much detail these are essentially ways to be more specific and produce more customizable results with words and inputs, here is a review of parameters.
AI Image of the week
Thank you for reading this week’s issue, check past issues here. Share this newsletter with colleagues, friends, or anyone interested in the combined world of architecture and artificial intelligence.
Feedback is always appreciated, let us know how we did by responding to this email.
Until next Friday,