Getting Started with Image Generation in Midjourney

Jeremy Osborn: Midjourney is a generative AI program that creates images based on natural language descriptions called prompts. People use it for all sorts of projects, from replacement of stock photography and illustrations, to rapid prototyping of artistic concepts, and everything in between.

Hey there, I’m Jeremy Osborn. And in this tutorial, I’m going to show you the best way to get started with image generation in Midjourney.

This tutorial assumes:

• You know how to create a Midjourney account.
• And use the Discord interface.

If you need help with that, go to docs.midjourney.com to learn more. Now let’s dive in.

In the Message field at the bottom of the Discord interface, type the / (slash) key. From the commands that appear, choose Imagine. Everything you type after this will be the description of what you want to see.

So let’s say you’re working on a project that needs the classic recycling logo, but you want it to be 3D and rendered as green grass. Well, you need to write a text prompt to make that happen.

Big Idea: The more detailed and articulate your prompt is, the better the results.

Many people use short, generic prompts and then get inevitably disappointed with the outcome. So the more specific the details of your prompt, the better.

Type the following:

A detailed and photorealistic rendering of the universal recycling symbol made entirely out of lush, green grass. The lighting is soft and natural, and the background is a neutral gradient to emphasize the symbol.

Then press Return. After a few moments, you’ll get four variations back.

Click anywhere in the grid of images to enlarge, and you’ll see all four generally did a pretty good job. So it’s possible you might be satisfied with these first results. But often, you’re going to want to refine them to get closer to your vision or just to experiment.

Click anywhere outside the grid to return to the interface. One way to refine an image is to choose the variation you like best and then create four more variations. That’s what this row at the bottom of your image grid is for. These buttons correspond to the grid of images. So V1 is top left, V2 is top right, V3 is bottom left, and V4 is bottom right.

I’m going to click V2, which corresponds to the top right image. And a remix pop-up window appears with the original prompt. Now, if you did nothing here except click Submit, you would get four more variations of that image, all slightly different. And sometimes, that’s fine.

But what’s way more powerful is modifying that prompt slightly with more specifics.

So go ahead and type:

The border of the arrows are a brighter green to improve visual contrast.

And then press Submit.

The image will regenerate. And when it’s done, you will have four updated images based on your revised prompt. And I think V2 of this new batch is the best. Let’s compare that with the original one. And yeah, I think the added contrast on the edges of the arrows is definitely an improvement.

Now, you can always return to your original images and click the Variation buttons again and change your prompt language. So as an exercise, try pausing the video and giving that a try. So make the border red or glowing, or change the background entirely to sand or water.

Finally, let’s talk about getting a higher resolution of your images. The default resolution for square images is 1024 by 1024 pixels. But you can double that using the Upscale feature. That’s what the U in this top row stands for. So click on U2 or the one in your grid you’re most happy with. And it’s going to take a moment, but you’ll get a single image, as well as a few rows of new buttons.

Ignore all that other stuff for now, except for these two, Upscale subtle and Upscale creative.

As a Best Practice: Upscale subtle is what you’re going to be using most of the time.

This will increase the resolution while maintaining the original details of the first image. Upscale creative will add new details to the image when upscaling. And that’s something that can come in handy in certain situations, but not this one.

And let’s look at 100% magnification. And you can see, yeah, in this upscaled image, the amount of detail is pretty significant. To download this image locally from your browser, Control + click or right-click the image, and then choose Save Image As… to the location of your choice.

That’s it for now. Be sure to go to the Resources section of this tutorial to find more information on how to use Midjourney. And thanks for watching. And please check out our other Take 5 tutorials, as well as the entire course catalog here at Gymnasium.