Generate assets with Midjourney


Midjourney is an AI-based image generation tool designed to help users create visuals through text prompts. Midjourney can produce high-quality images that cater to a variety of needs, from product visuals and marketing materials to artistic creations and beyond.


Midjourney vs Dall·e 3

Midjourney offers a robust set of customization features, allowing users to adjust styles, add specific details, and refine the output more comprehensively. This can be particularly beneficial for creating consistent and brand-aligned product visuals.
You are way more likely to achieve desired, production-ready output from Midjourney than from Dalle.


Mockup of Boxes


Book Mockup


Digital Space Background


Midjourney vs Dall·e 3 Example - Megaphone calling to Earth


Photo of airplane crew holding a tablet


Photograph of a man walking up the stairs for a treasure


The difference is especially visible when working with photorealistic results, Midjourney is able to provide more believable photos, which is what we want in our visuals. If you are aiming for more cartoon-like aesthetic, Dalle will do just fine.


Getting Started with Midjourney

Quick step setup

1. Sign up for Discord.
2. Sign up for Midjourney, and accept the invitation to the Midjourney Discord channel.
3. Go to any channel, and type /subscribe.
4. Pick your Midjourney plan.
5. Head back to Discord, and use /imagine to start generating images.
6. Edit and upscale your creation.
7. Save your image.


Detailed Setup

1. Sign up for Midjourney

You still need to have a Discord account to access Midjourney, but a new version should be coming in the upcoming months. In the meantime, head to the Midjourney website, and click Join the Beta. This will open an invitation to join the Midjourney Discord channel. If you are not yet logged in to Discord, fill your Display name and Click Continue, and you're in.


2. Subscribe to Midjourney membership plan

Go to channel #getting-started Click Get your Midjourney membership

This will open a link that prompts you to sign up for a paid Midjourney plan.
Midjourney starts at $10/month for the Basic Plan with ~200 image generations per month which is usually enough (it was always enough for us).

Once you've signed up, head back to Discord. It's time to get started.


3. Generate your first image

Click on the Midjourney icon on the left (called Midjourney bot) and click on the Green button to accept TOS.

Follow the tutorial that will guide you through the basics of generating images in Midjourney. It should give you a fairly good overview of Midjourney capabilities and the process of generating images


Prompting Midjourney

Midjourney prompting is an art in itself, however there are certain tools that can help you achieve results quickly and easily. It is also handy to set up some prompt templates, which you will later use to generate your assets that are on brand.



Establish your visuals style

In order to have consistency in your visuals, you should establish a core visual aesthetic.


Example 1

Let’s say you like art direction of Mirrors Edge
And let’s assume you have orange and white as brand colors.


You want to create an aesthetic picture of the office.

In Midjourney, you would write a prompt like this:

3d render in the style of Mirrors Edge, office table with screen, keyboard, mouse and notepad, sleek and clean image, white colors with orange highlights --ar 1:1

3d render in the style of Mirrors Edge

This is your starting base that you will be using all the time to maintain visual consistency

office table with screen, keyboard, mouse and notepad, sleek and clean image,

This is the actual description of the image you want to generate. You can use the GPT for help here
white colors with orange highlights
At the end of the prompt, you will give guidance of colors
--ar 1:1

this parameter at the end will specify the aspect ratio of the image. Usually you will only need images in ratios of 1:1, 4:5: 16:9


Output Results

These are the results Midjourney generated. You can choose to upcale one of them or refine your prompt to achieve the result you want


Maintaining the visual style

Now that you have a base, we can generate new prompts and use our base and colors to maintain the visual consistency

Objective to write to GPT:
Factory filled with robots

GPT output:
A high-tech factory filled with robots, sleek metallic bodies, and mechanical arms, assembling products. The robots are spread throughout the spacious, brightly lit factory. Background includes conveyor belts, machinery, and control panels. Detailed lighting highlights the metallic shine and the futuristic atmosphere.

Output combined with our base:
3d render in the style of Mirrors Edge, A high-tech factory filled with robots, sleek metallic bodies, and mechanical arms, assembling products. The robots are spread throughout the spacious, brightly lit factory. Background includes conveyor belts, machinery, and control panels. Detailed lighting highlights the metallic shine and the futuristic atmosphere. white colors with orange highlights --ar 1:1

Output Results
Example 2

Let’s try different styles and same prompts

Style:
Dark magical fantasy cartoon

Colors:
Blue and black

Output Results


Creating a book cover with Midjourney

Objective:

Create a cover for an ebook about basketball drills

1. Generate prompts using GPT

Text written into ChatGPT: 

Basketball hoop centered in the scene, clean and uncluttered

Chosen Output

/imagine prompt: A photograph of a basketball hoop, clean and uncluttered, centered in the scene. Background is a clear blue sky with a few wispy clouds. Bright natural sunlight, minimal shadows. Created Using: high-resolution DSLR camera, minimalist photography, outdoor natural light, clear sky emphasis, focus on simplicity, sharp details, vibrant colors, modern sports aesthetics --ar 16:9 --v 6.0

2. Clean up the prompt

Modify and add the prompt template we have created

/imagine prompt: 3d render in the style of Mirrors Edge a basketball hoop, clean and uncluttered, centered in the scene. Background is a clear blue sky with a few wispy clouds. Bright natural sunlight, minimal shadows. Created Using: high-resolution DSLR camera, minimalist photography, outdoor natural light, clear sky emphasis, focus on simplicity, sharp details, vibrant colors, modern sports aesthetics, white colors with orange highlights --ar 16:9 --v 6.0

Tip: Don’t generate covers with text
While image generation models are capable of putting text into compositions, it is still clunky and unreliable. I recommend only focusing on the imagery, and then add text in design editors like Figma or Canva.
You might think you will save time generating it together with cover, but will most probably end up wasting a lot of time until you get it right, as opposed to laying it out yourself in Canva in 3 minutes.


3. Generate the image using Midjourney

Paste the prompt into Midjourney and choose the best variation. If you don’t like any of the results, try adjusting the prompt.

These are the results I have got, but I dont like any of them. I want to have primarily white backgrounds, not blue sky.
I have adjusted the prompts so the sky is not as the main element, and have replaced it with blank white wall.
3d render in the style of Mirrors Edge a basketball hoop, clean and uncluttered, centered in the scene, view from front. Background is a clear white wall. Bright natural sunlight, minimal shadows. outdoor natural light, focus on simplicity, sharp details, vibrant colors, modern sports aesthetics, white colors with orange highlights --ar 1:1 --v 6.0

I like number 3, so I click V3 to generate more similar compositions

Results:


Number 4 looks good enough for me, so I click U4 to upscale the image.

Now we want to have some space around the hoop so we can place the text of the ebook title there. So now we can click on Zoom Out 2x to zoom out the image, which we can later crop as we like. If you use Canva PRO or Photoshop you can use their AI expand feature to achieve the same result.



Number 4 looks good enough for me, so I click U4 to upscale the image.

Now we want to have some space around the hoop so we can place the text of the ebook title there. So now we can click on Zoom Out 2x to zoom out the image, which we can later crop as we like. If you use Canva PRO or Photoshop you can use their AI expand feature to achieve the same result.

4. Export your image

When you find an image with good enough negative space, upscale it once and save it to your computer. You now have an amazing visual as a background for your ebook.

5. Create background texture

If we want to go even more fancy, we can generate some texture for the background. ( Repeat the steps 1-5). Use keywords like closeup, macrophotography

3d render of orange basketball ball. close,up, macro photography focus on simplicity, sharp details, vibrant colors, modern sports aesthetics --ar 16:9 --v 6.0

Bonus Tip 1 - iPhone Photos


If you want to create some authentic photos that look just like they were shot on iPhone, you can use this prompt template

iPhone photo, POV, [object or location]. --ar 3:4 --style raw --s 75

Note
Parameter --style raw uses an alternative model that may work well for users already comfortable with prompting who want more control over their images. Images made with --style raw have less automatic beautification applied, which can result in a more accurate match when prompting for specific styles. - quote from Midjourney documentation

Bonus Tip 2 - 3D Custom icons

You can create your own 3D icons that make your visuals more interesting

3D icon of [object], floating against white background --style raw

Only thing that’s left to do is to use some background remover tool like icons8 and you have your own 3D asset ready for use.

Created with