If you’re like anyone just starting out on this MidJourney…uh…journey, you’ve probably been gobbling up all the prompts you can find and trying them all out. Well, let me give you the one biggest and most important piece of advice you’ll ever get when it comes to prompting for text2img AI: Simple is better!

I’ll bet you’ve seen prompts with references to complicated professional photography equipment, lighting settings, f-stops, lenses, etc. You look at all that and think “damn…I’m not a photographer, I don’t understand any of that. But apparently they do because their image looks fantastic!”

No. Just, no. In all liklihood they’ve only copied from some other prompt thinking the same thing you did. Here’s the truth: MidJourney has no idea what any of those terms mean (with the exception of a couple of isolated items, which I’ll mention below). Ignore them. If you see a prompt for an image you like and it uses all those terms, delete them before you send it through /imagine. Just delete delete delete.

Simple is better

Start out this simple: subject, subject description, environment, aspect ratio (image width and height) – then the rest of your commands.

Ok. Here’s an example:

woman, blonde, middle-aged, shopping bags, mall --ar 2:3 --style raw --s 50

Here are the results I got with this prompt:
Simple prompt - blond woman - grid

These are actually pretty excellent right?

Let’s see how we can modify this simple prompt a little

diverse women, friends, middle-aged, outdoor coffee shop table --ar 3:3 --style raw --s 50

Results?

I’m very happy with these
Simple Prompt - Diverse Women - grid

Ok. I hope this has shown you how simple prompts get excellent results. Start from the simpler and get more complex as necessary. But be wary of technical terms and very specific ideas. Also, ignore things like “i can’t believe how beautiful this is” and similar phrases that might show up when using /describe. They. Are. Noise. Nothing more.

GREEBLES!!!

I promised in my previous post that we’d talk about greebles.

Greebles are a kind of building block philsophy that visual dev artists came up with. Apparently the term originates from model builders for Star Wars, Episode 3 (the first of the Star Wars movies – 1979). The term was used to describe the way the Death Star and the Imperial Star Destroyer had all these little pieces and parts that were bashed together from many different model kits.

Look familiar?
Greebles - Death Star

Ok. Cool. So how does this relate to MidJourney txt2img prompting?

Well, basically, it adds…greebles! (imagine that. Astonishing)

Ok in some cases the greebling is less like building blocks and more like intense 3d details. But let’s look at a few examples

Greebles in a photograph

In photographs like we have above, greebles don’t do a whole lot. So let’s go with something more illustrative/painterly:

Greebles in more artsy imagery

First, the prompt without Greebles
magical mystical forest garden, triptych, gustave moreau and bosch --ar 2:1 --s 1000 --style raw

Here’s the grid I got from that.
Magical Mystical Forest Garden

I’m going to use the top left on (#1) since it’s the only one that’s overtly a triptych.

Here’s the upscaled version of that one from the grid:
Magical Mysictal forest Garden - no Greebles

I click on V1, and add “greebles” to the end of the prompt, like this:

magical mystical forest garden, triptych, gustave moreau and bosch, greebles --ar 2:1 --s 1000 --style raw

And here are the upscaled versions it gave me. Notice the extra tiny details?
Magical Mystical Forest Garden #1 - Greebles

Magical Mystical Forest Garden #2 - Greebles

Magical Mystical Forest Garden #3 - Greebles

Magical Mystical Forest Garden #4 - Greebles

Those are greebles!

Depending on the image, you may get more building-block greebles like in the Star Wars Death Star example above. But basically greebles just add extra details to your image (they don’t do anything really with photographs though, as far as I can tell). Next time you see a prompt that says something like “insane details” or similar, try using “greebles” instea, and see what the result is. I’m betting more often than not it’ll give you those “insane details”. But sometimes you’ll get something completely unexpected too (thus is the nature of txt2img prompting…sometimes you just don’t get anything like what you expect – which can be either a pleasant surprise, or nightmare fuel!)

I would love to know how you use this simplified prompting format, and see your greebles!! Leave me a comment, share the page. I’d appreciate it a lot.

Subscribe and receive my FREE eBook


4 MISTAKES You’re Making on Your Website that are Costing You MONEY!

  • This field is for validation purposes and should be left unchanged.

The Author

Save

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Ready to chat about how ThePixelPixie can help?