Settings

In this chapter, we will discuss the various settings that are available in an AI art generator and how they can be used to influence the output of the AI.

By understanding and effectively using these settings, you can create unique, original artwork that is tailored to your specific preferences and needs.

Let’s go through each one of the settings:

Model Version:

This refers to the specific version of the Midjourney model that is being used to generate art. Different versions of the model may have different characteristics or capabilities, and users may be able to select the version they want to use.

For now, let’s just use V4 since it’s the latest and most advanced model.

In general I recommenced always using the latest models and adjusting yourself to them.

Style:

This setting allows users to choose how much the results will be true to the prompt, or how experimental they will be.

Higher style setting will result in art that is more "crazy" or unusual, but less similar to the prompt or input provided by the user.

Conversely, decreasing the value of the style setting may result in art that is more similar to the prompt but less "crazy" or unusual.

Quality:

This setting allows users to adjust the resolution or visual quality of the generated art.

Supposedly, higher quality settings may result in more detailed or smoother images, but may also take longer to generate.
But honestly, just keep this one on Base Quality

Private mode:

As I have mentioned before, private mode allows you to keep your creations secret.

Private mode can be useful for individuals or organizations that want to use Midjourney, but want to keep their work confidential.

By using private mode, users can ensure that their data and output remain private and are not shared with the public.

*This option must be purchased separately, and is also included in the Pro Plan by default.

Remix mode:

This setting allows you to make changes to an existing result.

Just hit the variation button with Remix mode enabled, and change parts of your prompt to adjust it to your needs!

Fast / Relaxed mode:

Fast Mode is like a high-performance sports car - it can zoom through the rendering process and get your images done in record time, but it comes with a cost.

You'll have to pay in terms of rendering hours to get those images at lightning speed.

On the other hand, Relaxed Mode is like a trusty old pickup truck - it may take a bit longer to get the job done, if you don’t mind watching some youtube videos while waiting.

So, if you're in a rush and need to get things done quickly at a price, fast mode is the way to go.

But if you have a lot of images to render and don't mind a longer wait, relaxed mode is the perfect choice. Just be prepared to put in some extra rendering time!

Upscaler:

This setting allows you to choose which kind of upscaler you want to use to enlarge your images.

Regular Upscaler may help "fix" errors or imperfections in the original image.

It basically tries to interpret what should happen in the image, and make everything into a final sensible result.

For example a little spec of dust in the horizon might become a small person - because it tries to make sense of what that dust might be.

This is why sometimes the result won’t be similar to the original image.

Light Upscaler gives you smoother and less detailed result.

It can be useful if you want a smooth looking face, or a smooth looking retro image for example.

Beta Upscaler tries to stay as true as possible to the original image.

This might sound nice, but the Beta Upscaler is so true to the original image, that the amount of imperfections will usually be too noticeable.

I do use Beta Upscaler occasionally, but only when there was something about the original image that didn’t translate well with the regular upscaler.

As an example, let’s take our “two elegant brothers” image and see how it changes when we use Remix mode on it:

*I personally think Remix mode is very hit or miss. It works well on the rare occasion, and may be useful if you really liked a certain image and want to make some changes.

But I find that it usually degrades the quality of the final results - and I find myself preferring to just “start from scratch” and find a better foundational prompt.

Aspect Ratio:

This setting is not included in the main ones, but it’s incredibly important! It allows you to make images with different aspect ratios.

Currently the supported ones for V4 are 3:2 and 2:3.

In order to use them you simple need to add this in the end of your prompt:

an apple —ar 3:2

an apple —ar 2:3

Weights:

I find Weighting prompts to be useful once in a while.

Weighting prompts means you give more weight (or emphasise) to one part of the prompt rather than another.

Let’s use this prompt as an example:

two elegant brothers, stairs by Van Gogh.

Now let’s add some weight:

two elegant brothers, stairs::2 by Van Gogh.

Now the prompt will focus much more on Stairs rather than focusing on the two elegant brothers.

I like using this when I feel like a part of the prompt simply got lost or ignored.

—NO

By using —no we can do an “express” way of exactly the opposite as we did with the weights.

For example when I did this prompt:

Darth Vader on the cover of Vogue

He always appeared as a woman or had very feminine attributes. It was actually quite funny and entertaining but I also wanted some pictures of Darth Vader in his natural form.

In order to do that I simply added:

Darth Vader on the cover of Vogue —no female

Or

Darth Vader on the cover of Vogue —no woman.

That way we can tell the algorithm to avoid a certain concept while it’s creating the image.

Style 4A

Did your image feel like it came out to flat? Too much like a “Midjourney template image?” Or maybe you feel like a lot of your prompt was forced to go in a certain direction that you didn’t have any control over?

Well, Style 4A can solve those problems quite often.

while using —style 4a at the end of our prompt, we are able to use the prototype version of the V4 model which was a bit more crazy and unique.

It has more artefacts and issues, but it looks more unique and “Free”.

Let me show you an example:

Astronaut clocks

Astronaut clocks —style 4a

The default V4 model has the tendency do “dumb things down” and make them look more aesthetic in a forced way.

I find style 4a great when I’m doing a crazier scene or when I feel like everything I wrote in the prompt was “dumbed down” to a very basic aesthetic result.