Skip to main content
  1. Posts/

Midjourney - an AI based image generator

·534 words·3 mins· loading · loading · ·
English AI Midjourney
rOger Eisenecher
Author ::..
rOger Eisenecher
> 12 years leading and building a SOC for MSSP • > 20 years working in security • > 40 years working with IT • 100% tech nerd.
Table of Contents

Currently there is a hype ongoing regarding AI based tools, like text based systems like ChatGPT or image generators like Midjourney. Here I will show you some pictures and hints regarding Midjourney.

Introduction

To start with Midjourney just go to their website https://midjourney.com and join the beta by signing up. You will get a base credit to play with Midjourney.

Here I will present you some examples from Midjourney.

How Midjourney work?

To start it is very easy. You have just to enter a so called prompt. Based on this prompt Midjourney will do his magic and creates four images as a preview. Based on this preview you could then generate variations (Midjourney re-interprets your prompt) or get one of the four images in better quality. Of course there are more options for the promt. For detailed instructions regarding possibiliets of the promt you should have a look into the documentiation .

First Example

Please note that following examples are from other users from Midjourney and provided here as examples.

Starting prompt

The first example I want to present is based on the following prompt: Marianne, light medium skin tone, beautiful, top model, look real, portrait photo by Annie Leibovitz, film, studio lighting, detailed skin, ultra realistic, bokeh, sharp features, front view, illustration, vector


Each four images are a variation of the previous one.

Slightly changed prompt - variant 1

The next step was to slightly change the prompt: marianne holding a flag by one hand high, medium skin tone, top model, look real, close up portrait photo by Annie Leibovitz, film, studio lighting, detailed skin, ultra realistic, bokeh, sharp features, front view


Each four images are a variation of the previous one. The last image is a so called upscaled image (from a previous image).

Slightly changed prompt - variant 2

Another adaption of the prompt: marianne holding a flag by one hand high, light skin, top model, look real, close up portrait photo by Annie Leibovitz, film, studio lighting, detailed skin, ultra realistic, bokeh, sharp features, front view


Each four images are a variation of the previous one.

Another slightly modified prompt - variant 3

Another iteration of the prompt looks like: marianne holding a flag by one hand high, light medium skin tone, top model, look real, close up portrait photo by Annie Leibovitz, film, studio lighting, detailed skin, ultra realistic, bokeh, sharp features, front view, illustration


Each four images are a variation of the previous one.

Other examples

Do you know Walter White?

walter white as a drag queen
Prompt: walter white as a drag queen

walter white in a dress
Prompt: walter white in a dress

What you can see here is that Midjourney respects the context. So it generated images related to the previous prompt - that’s why even wiht the second prompt Walter White is shown as a drag queen.

How Midjourney imagine how an angel looks like

Also an intersting prompt, right?

whats-an-angel-look-like
Prompt: whats an angel look like

Bottom line

It is really interesting to play around with Midjourney and to see what others are trying to create. This is only the starting point and I showed you only some simple prompts. Of course you could create more advanced prompts - only you fantasy are the limits.