[Midjourney Introduction 🔰] Explaining how to write prompts, basics, applications, and tips.

Hello, AI FREAK here.

In the previous note I introduced everything from how to register for Midjourney to how to use it simply.

By looking at the above and trying out Midjourney, I think you will be able to create AI images.

However, when you actually try to create AI images,

I think many people face concerns like these.

So, this time, to resolve those concerns, I would like to explain in detail about 'Midjourney prompt basics and applications'.

If you understand the basics of Midjourney, you can basically create any image.

To reach that point, the basics are really important, aren't they.

I have been using Midjourney almost every day for nearly two years.

What I realized there is,

that is the case.

It is obvious when written down, but I think there are surprisingly many people who neglect the basics.

So, let's take a look at the basic knowledge right away!

Let's understand the structure of a prompt

Well, to create high-quality AI images, the 'prompt' is indeed the most important thing.

And there is a kind of pattern to prompts.

Specifically, it looks like this.

The above is a Japanese translation of information released by the official Midjourney.

As shown above, clearly stating the 'subject,' 'medium,' 'environment,' etc., makes it easier to create the image you are aiming for.

Please take a look at this image as a sample.

This is a cat I created while referring to the template above.

The specific prompt used is,

this one.

Breaking down each part of this prompt looks like this.

1️⃣ cat wizard, 2️⃣ fantasy art, 3️⃣ glowing staff, 4️⃣ cinematic lighting, 5️⃣ 4k, highly detailed

This is like the basic template.

Even though it is just a list of keywords, a surprisingly convincing illustration of a cat wizard appears, doesn't it?

You could write it out in a long sentence like 'a cat is a wizard and the staff is glowing...', but comma-separated is better.

You can create an image like this just by changing 'cat' to 'dog'.

If you understand the structure well, you can give clear instructions even when you want to make changes.

Let's look at another example.

The process here is the same as well.

For example, just changing the subject part from 'retro' to 'future' and 'cyberpunk' to 'steampunk' will change the atmosphere.

As mentioned above, having a certain structure makes it easier to make partial changes.

In that sense, I think it is important to learn the basics first.

Common bad examples of prompts

I receive consultations from various people saying, "I can't create images well with Midjourney...", and many people have

is the impression I get.

Aren't there quite a few people who are convinced that "The more instructions, the better! I should pack as much as I can into the prompt!"?

Of course, it's fine if it works well, but basically, if you pack too many elements into Midjourney, the AI gets confused.

As a result, it doesn't know what to prioritize, so the output image often ends up being different from what you intended.

First, try a style where you narrow it down to about 3 to 5 keywords, such as Subject + Style + Supplement, and then adjust from there.

For example, even with something this simple, you can create an image of the quality mentioned above.

If you add too many detailed elements beyond this, the AI is likely to get confused, wondering "Is it a wolf or a warrior? What kind of accessories go on which part?", so first let's create an initial template.

By adding your favorite parts to the prompt, you can learn while getting closer to the image you are aiming for.

Write in the affirmative, not the negative

This is also an important point.

And I think many people actually don't know this.

Midjourney does not handle negative sentences like "do not draw XX" very well.

In fact, if you write it that way, that element is more likely to be reflected.

Let's look at some specific examples.

Even though I wrote 'no', mountains were drawn prominently.

The AI interprets it as 'Oh, I found the elements of people, text, and mountains in the prompt,' and picks them up on its own, turning them into the image.

On the other hand, please look at the image below.

This image does not contain elements like mountains, does it?

This is because it follows the writing style in the Midjourney rules.

Either don't write unnecessary things in the first place, or exclude them with the '--no parameter'.

If you write it this way, it will properly exclude those elements.

This is a basic syntax rule, so it's worth remembering.

Entering numerical values and specifying composition

By entering specific numerical values, such as 'one protagonist' or 'three buildings in the landscape,' it becomes easier for Midjourney to interpret.

Regarding composition, specifying the viewpoint such as 'close-up,' 'wide shot,' or 'bird’s-eye view' makes it easier to get the screen layout you intend.

Adding movement and action elements

Even in still images, adding verbs like 'running,' 'jumping,' or 'flying' makes it easier to create an anime-like sense of dynamism.

It is easy to create dynamic images with flowing hair.

At first, you tend to create images where the subjects are just standing still, but having movement makes them look great when incorporated into short videos.

Summary: Once you have mastered the basics, it is time for the trial-and-error phase.

In this article, I focused on how to write Midjourney prompts and shared techniques and tips that can also be used for short videos.

If you carefully review the information above, you should be able to cover the basics quite well.

The next note will be on the theme of 'Methods and tips for creating your ideal images'.

Please be sure to check that out as well.

If you found this note helpful, it would be encouraging if you could 'like' it or show your support.

See you in another note!