MidJourney Fundamentals Explained
how does AI image generation work?
The artificial intelligence
The graphic design industry may soon be going out of business, or aleast the most of it. And this is because AI tools like midjourney are rising to the top replacing the need for a graphic designer and putting back thousands of dollars back into your pockets. In today’s episode we will explore how midjourney ai can be your graphic designer friend and help you design websites, illustrations, posters and marketing campaigns.
Training
Midjourney is an AI trained on millions and millions of images and their descriptions. We can think of it like having a fantastic photographic memory, literally! It has seen so many pictures that it can draw new pictures based on the details of similar pictures it has stored in its memory. Think of it like memorizing a giant photo album and the captions for each photo. By doing this, Midjourney knows how to connect images with their relevant words. For example, it would know what a sunset looks like or what we mean by a futuristic city. You may wonder, how does it understand what the user wants?
Understanding Instructions
Using Midjourney is more like playing pictionary with friends. You ask it to draw something and it creates a new picture based on what it understands from your instructions. For example, when you ask it to “draw a dog on a skateboard in space”, it starts by first figuring out what a dog looks like and keeps thinking about more details around what would it look like on a skateboard and then finally the space in the background based on its artificial intelligence. And again, it can do this because it has already seen lots and lots of pictures of dogs, skateboards and space. Now, all it has to do is combine them all together into a meaningful image.
The Drawing Process
Once it has a rough idea of what it needs to draw, it starts by drawing the first subject and then continuing on to the next one. In our example, the AI draws what a dog looks like first and then continues to add a skateboard into the frame. After that, it throws the dog and the skateboard into the space all based on the imagination or simply based on its memories of dogs, stakeboards and space. At this point, one might wonder, how does it know that it has drawn a good enough image to present back to you?
Learning From its Mistakes
Once the first version of the image is ready, which probably is not the best, midjourney tries to put eveything into perspective, and is ready to make further refinements based on further instructions. You can ask it to make the dog’s head bigger or add some coloring on the skateboard. It’s like asking your graphic designer to make changes based on what you would like more to make the image even better until you absolutely love it. The AI will honor your feedback, update its understanding and add the touchups that you ask. What do you think? Is it as perfect as the Picasso or Leonardo Da Vinci? Let’s find out.
Its not perfect
Midjourney is a program and it needs clear instructions to sketch what you need. Sometimes, it may draw a weird-looking dog or with 5 legs. When it does so, it’s simply trying to recall its memory and draw everything from scratch and its not perfect. It does not always understand like us humans do. All we have to do is to kindly ask it to add more corrections to the image and it will deliver without complaints. Rest assured, it will get better over time based on all the new images it generates and continuously updates its intelligence from user feedback.
So, midjourney is a blend of a massive photo album, a fantastic memory and a bit of imagination that’s ready to start drawing for you and its all packed up into an app. All you need to do is describe what you want midjourney to draw for you!
An example prompt
/imagine a pup and his girlfriend on a motorcycle
How to generate Fantastic AI art for Beginners using MidJourney
Please like and subscribe for more content like this. Also, comment how you are using midjourney to help with your graphic design needs.