Midjourney is an American company and at the same time the name of a service that uses artificial intelligence generates an image based on text input. This is a similar service to the better known DALL-E, but with comparable results.

The advantage of Midjourney in the current beta is that you don’t have to wait for an invite. In no time you can start thinking about what the AI ​​should draw for you.

Commands are entered in English. Here I want: Hairy guinea pig portrait, redhead, studio ghibli, cinematic lighting, ultra quality, 4k, octane render, ultra detail –ar 16:9

Or this example, other examples can be found below in the article: Scary future house, ufo flying around, dark scary style, ultra realistic, detailed, arial view –ar 2:1

Instructions for using Midjourney

On page midjourney.com click button Join the beta.

The discussion service page will open Discord.com. It’s not his fault, you’ll be communicating with Midjourney via Discord chat. If you already have an account, log in.

If you don’t have a Discord account, create one: Enter a nickname; pass the test if you are not a robot; enter your date of birth, email address and create a password. You will receive a message by e-mail, click on the confirmation link in it.

Register on Discord

You’ve opened your Midjourney page on your Discord page. if not You can find it here.

Midway through on Discord

In the left column are the individual channels. You can see the parts image gallery or Gene themed image for interesting creations. For example in daily-theme is an image on the currently entered topic. On the channel announcement You will find news, for example author here announcing the availability of new parameters for build. For example, –test and –testp are tested for more realistic output.

We generate images

Scroll down in the left column to one of the starter channels. This is a place for newbies to try, where you can immediately see live traffic from other users. Keep in mind that your communications with the Midjourney bot are also visible to others.

Most of the time you will see some starter channels

Write in the line at the bottom of the window /imagine, press tab and add your command in English in the field after the prompt. Press enter.

I want mystical underwater animals with big eyes, bright color images, and good depth of field

You will gradually see how the images of the four previews are created. It will be created in less than a minute, the finished image will be moved to the end of the conversation. With more live traffic on channels, it’s sometimes harder to find your own work. You can click on the rectangle and see it in a slightly bigger version, save it. Or continue with the button below.

Use the buttons to choose what you want next when a thumbnail with four previews isn’t enough

The button on the right with an arrow starts the generation again according to the same input. Press the “U” button to calculate the larger selected image, in the case of a standard grid with a resolution of 1024 × 1024 pixels. Press the “V” button to use the selected image as the basis for setting up four more variations. Previews are numbered from left to right, for example three on the bottom left.

I chose the first option from the preview and waited for the larger image to build. The percentages are ticking away, Midjourney is adding more details

Once the image is ready, click on it to view it in full version. You can save it from here. Also pay attention to the buttons High end up to Maxwhich in case of square will make the largest image size 1664 × 1664 px.

I’m going back to the “V” button. I chose the fourth variant from the original preview and made four more based on it. I can now work with them as I did with the first draft.

Additional variants are generated based on the selected preview

Get inspired, be creative

Midjourney is the strongest while generating various worlds and fantastical creatures. It has better results with them than trying to get a photo like image. In addition to a description of what should be in the resulting scene, focus on defining the artistic style (you could also use the artist’s name), trying to describe how it would look.

It’s helpful to browse other people’s images on Discord and get inspired by their input. Then try elements that have worked elsewhere in your command and incorporate whatever comes to mind. Otherwise, even just watching the beginner’s channel, where new pictures are constantly being added, is very interesting.

Midjourney’s free beta has limitations. This lets you generate dozens of images with each Discord account, with 25 minutes of CPU time. It will then ask you to sign up for a paid program. Basic version costs $10 per month (200 minutes), standard version costs $30 per month (15 hours). Paid programs offer additional benefits, see website.

In preparation, I created a number of images that I selected for the article introduction. I’ll show the more successful one here in the end, each time with a specific command I used in Midjourney.

Ancient world with computers, colorful background –ar 2:1 (Computer doesn’t show up, Midjourney sometimes ignores some tasks)

Hyperrealistic female alien portrait in front of futuristic computer, octane render, colorful background –ar 2:1

Future world, mc escher style, night, highly realistic, detailed –ar 2:1

Scary house with vines, colorful birds flying, peacock on the ground, scary dark style, ultra realistic, detailed, arial scene –ar 2:1

Indian fantasy steampunk, city with huge thin bridge between houses, steampunk airship in the sky, highly realistic, detailed –ar 2:1

Unused illustration image for today’s article with SLS rocket launch broadcast