BACK
As a creative professional I've been curious about the evolution of artificial intelligence for creative purposes and have been exploring ways AI can be used to enhance my design process.
The most impressive use for creative AI currently is its ability to generate artwork.
One of the most exciting AI tools available for public use is Midjourney - an artificial intelligence program that creates stunning images based on user input.
Midjourney has very intelligent comprehension of the user input prompts compared to some of the other Ai tools on the market. This allows you to create some extremely impressive artwork with eerily realistic detail.
Midjourney was even used to create the cover of an issue of The Economist magazine in June 2022. We are living in the future!
To start using Midjourney you need to sign up for the beta. You will need a Discord account to join the Midjourney beta as you will be creating your images using their discord bot.
Once you have joined the beta check out the Quick Start Guide to get up to speed with how to interact with the Midjourney bot.
To generate an image with Ai you enter text prompts using the /imagine command to tell the Midjourney bot exactly what image it should create.
An example of an imagine command: "/imagine jellyfish ufo floating above old western landscape"
This generates four image variations based on that prompt:
The Ai uses photos found on the internet to paint the image you describe in your prompt. For that reason you should include objects in your prompt that would have a lot of photos on the internet to pull from. For example objects like: apple, king, mountain, river, motorbike or fish would work really well because they all have a ton of reference images online.
When your image renders you will see four image variations in a 2x2 grid. Each of these images can be upscaled or re-rolled to further develop that image. This process lets you add much more detail to the image or re-roll it completely to get a new result based on the same conditions.
Here are some examples:
To get the best possible results you can set imagine parameters within your prompt, allowing you to create even more refined results for your output image.
The imagine parameters give you control over:
You can also add image prompts or render into video format!
In my experiments I found providing the Ai with an image prompt as a starting point yields the best final result. You can provide up to x photos as a starting point.
