Realistic Midjourney Images
On 28 August 2022 Midjourney released some new options for improving image quality that allow you to create much more realistic images.
Two new modes were released
- A general purpose artistic mode you can use by typing --test
- A photo-realism mode you can use by typing --testp
I wanted to share some of them, and also compare the same prompt showing the differences between the standard Midjourney image generator and the test versions.
Most of these images were upgraded using the --upbeta command.
Prompt: A photograph of the Planet Jupiter
Prompt: Dante's Inferno
Prompt: A human astronaut walking towards the towering gates of a mysterious alien city designed by HR Giger
Prompt: an astronaut exploring a red jungle on an alien planet
As you can see from these examples, compared with the standard Midjourney images the --test and --testp images are more realistic. But they also seem to lose a certain mysterious, artistic charm that the standard images have.
Nevertheless you get some beautiful results. Below are some of my favourite images so far created using the --test and --testp commands.
--Test and --Testp Midjourney images
My Other Blog Posts
More Midjourney Images
More images I created using Midjourney.
For these images I added --ar 16:9 to to the end of each prompt so that the images would be generated in a 16:9 aspect ratio, the standard ratio for widescreen televisions and most films and TV shows.
For some images I also used many different descriptive words. For example "highly detailed, realistic, 8K, cinematic" to try and influence the result. And for others I used nothing but a simple phrase or word to see what Midjourney would default to.
I don't have much else to say, except that I enjoyed creating these images and hope you check out Midjourney too!
My Other Blog Posts
Making Midjourney Monsters
Since I was a teenager I've been fascinated by the artwork of H.R. Giger, the Swiss artist who, among many other works, designed the creature (later dubbed "xenomorph") in the 1979 film Alien.
Giger's unique style featured biomechanical creatures in nightmarish environments.
I recently tried out Midjourney, a new tool for creating beautiful (and as we shall soon see, sometimes ugly) images generated in a few seconds.
I tried using the following prompt to create some AI-generated pictures based on Giger's unique style. For the following pictures I used this prompt:
"an HR giger xenomorph monster with sharp teeth looking directly at you, alien covenant, making eye contact, Annie Leibovitz style photography, full HD"
The images below are all based on this prompt, and I hope you find them as unsettling but fascinating as I do!
Below are all the initial images created by Midjourney, before being upscaled.
My Other Blog Posts
Testing out Midjourney
I recently started creating images with Midjourney, a new online tool that allows you to create images created by Artificial Intelligence.
According to their website, Midjourney "is an independent research lab exploring new mediums of thought and expanding the imaginative powers of the human species. We are a small self-funded team focused on design, human infrastructure, and AI. We have 11 full-time staff and an incredible set of advisors."
So far I've been using the Midjourney server on Discord to create images.
In one of the Newcomer Rooms, you can type in /imagine followed by the prompt for your text.
For example let's try something like "a happy kitten sitting on a chair"
The Midjourney bot then starts to create the image.
We then have our finished result, which is four images. We then have the choice to upgra
We then have the option to Upscale any of the images. For example if I click U4 then we will upscale the fourth image that was created.
Once finished we can take a closer look at the upscaled image.
The full resolution version is below.
You can also adjust the aspect ratio of the image produced. For example, the four images below were created with the prompt "desert planet, far future, ancient city --ar 16:9".
"--ar 16:9" indicates that you'd like the image to be in the aspect ratio of 16:9, the most common widescreen format.
Examples of Images created by Midjourney
Below are some of my favourite images I've created so far using Midjourney. It is amazing at imitating fantasy and science fiction art so I've leaned towards those types of prompts.
My Other Blog Posts
Comparing Craiyon, DALL-E 2, and Midjourney
I thought I'd add photos from three AI image generators; Craiyon, DALL-E 2 and Midjourney.
How they describe themselves;
- "Craiyon, formerly DALL·E mini, is an AI model that can draw images from any text prompt!"
- "DALL·E 2 is a new AI system that can create realistic images and art from a description in natural language.
- "Midjourney is an independent research lab exploring new mediums of thought and expanding the imaginative powers of the human species."
My relatively uninformed opinions of them;
- Craiyon is good for funny and topical images, but images lack detail.
- DALL-E 2 can create some almost photo-realistic images, but can lack artistic imagination.
- Midjourney creates stunning images that lean towards fantasy art rather than the real world.
In this post I just wanted to share some examples of the images produced by these three.
A Cat looking up at a South African flag, digital art
A Galaxy inside a Snowglobe, cinematic
Two Astronauts Exploring a Tomb on Mars
A Photograph of the Toronto skyline
A Hideous Monster staring you in the face
A Beautiful, Cinematic Rainbow
A Window into Another World
Two Astronauts explore the interior of a Colossal, Derelict Alien Spacecraft. Cinematic
Star Wars characters depicted in a medieval tapestry
Alien Covenant
Cape Town, South Africa
The Gates to heaven opening and revealing the view of a Gigantic Wonderous City
Other Blog Posts
Multiple DALL-E 2 images created from the same prompt
For those that haven't tried out DALL-E 2, it's a new Artificial Intelligence (AI) system that can create realistic images and art from a description in natural language.
You can take a look at some previous posts on the topic
- Testing out DALL-E 2
- More Examples of AI-Generated DALL-E 2 Images
- Comparing different photo styles in DALL-E 2
In those previous blog posts I tended to only show what I considered the best result from each of the prompts. What I wanted to show in this post is just how many unique images DALL-E 2 is able to conjure up from a single descriptive prompt.
In this case the prompt I used was:
A Soviet Propaganda poster of a Robot in Love
Below are the results of this same, exact prompt used again and again to generate new images in a few seconds. Each time you type in the descriptive text and click 'generate' four unique images are created. Each one beautiful and unique in its own way.
Hope you enjoyed reading about multiple DALL-E 2 images created from the same prompt.
Here are some more blog posts about DALL-E 2:
- Testing out DALL-E 2
- More Examples of AI-Generated DALL-E 2 Images
- Comparing different photo styles in DALL-E 2
Full Size Versions of the Robots in Love
Here are four of my favourite images in full resolution.
