April 12, 2023

Midjourney at the College Level: Visual Representations Created by Artificial Intelligence

This article is a translation of a text first published in Eductive’s French edition.

Do you know Midjourney? This tool uses artificial intelligence (AI) to create images from the user’s written prompts.

Since I discovered Midjourney, I have fallen into an abyss: I am constantly exploring the tool and playing with its different possibilities. I have been thinking about the various ways it could be used at the college level.


AI image generator programs

Midjourney is not the only tool of its kind. DALL-E is another quite similar program that has been drawing much attention. However, I believe the images generated by Midjourney to be prettier. Feel free to explore and compare the 2 tools!

Until March 31, 2023, it was possible to use Midjourney for free for up to 25 generations (Midjourney generates 4 images per prompt). However, Midjourney has temporarily halted free trials of his services. While waiting for its restoration, to make further use of the service, you need to pay for a subscription plan (allowing more than 25 generations).

DALL-E has the benefit of still being available for free. It is, for instance, integrated into Microsoft Designer.

How to use Midjourney?

Midjourney is a robot (bot) available on Discord. Discord is a free instant messaging social platform.

To use Midjourney, you first need to create a Discord account. Then, you can give text prompts to the Midjourney bot to create an image. The Midjourney Quick Start Guide provides detailed instructions.

The paid version gives you full rights to the images generated for you by Midjourney. (It was unclear with the free version.) It also allows you to have a private conversation with the robot.

You can apply any style to any prompt.

4 similar illustrations showing a xenomorph (monster in the movie Alien) seated at a table reading a newspaper it is holding with both hands. In 2 images, a glass that looks like a beer glass is placed on the table. In another image, we see what looks like a portion of a bowl. In the 4th image, an item that cannot be identified (maybe the head of a small dog behind the table?). In 3 images, we can see a pencil, either on the table or in the xenomorph’s hand. One image displays a plate placed on the table. In each picture, the table surface and the image background are decorated with patterns suggesting technical charts or plans, but they are too blurry to identify the details.

4 versions of a xenomorph doing crosswords in a newspaper in H.R. Giger style, as designed by Midjourney

4 illustrations of a woman tenderly kissing one or more cats. The 4 images remind us of the painting The Kiss by Gustav Klimt, each in their own way. Some have a painting style similar to Klimt’s, with circular patterns integrated into the characters, but one of them is more realistic. Nonetheless, this image reminds us of the work of Klimt through the woman’s posture and the warm colours.

4 versions of The Kiss by Gustav Klimt… with cats, designed by Midjourney

We can even provide images to Midjourney (by referring to a URL or by uploading an image on Discord) so it can get its inspiration from them. To explain Midjourney to my mother, I uploaded a picture of her on Midjourney and asked the robot to design her in different artist styles.

4 illustrations of a woman with wavy gray hair, smiling, with closed eyes, who seems peaceful and happy. The background is decorated with circular royal blue and golden patterns in the style of Gustav Klimt.

My mother portrayed in Klimt style

4 illustrations of a woman with grey hair, smiling, seated at a table in a sunlit café with windows. In the 4 images, shades of green prevail (either on the woman’s clothes, or by the colour of the walls and furniture).

My mother portrayed in Edward Hopper style

A new version of Midjourney

Since I started my experimentations with Midjourney, a new version of the tool (the 5th version) has been available online. The realism of the images generated is mind-blowing!

Black and white image with a photorealistic style. A xenomorph is seated in front of a patio table on a wooden terrace attached to a house. We see a neighbouring house in the background. The xenomorph has one hand on its chin, as if it were thinking. Its other hand seems to be holding a pen, but a slight defect in the image makes it seems as if floating, or incomplete. A newspaper is laid out on the table in front of the xenomorph; the xenomorph is looking at it. We do not see any crossword puzzles on the paper. A cup is on the table, next to the newspaper.

I prompted once again the xenomorph doing crosswords, but this time asking it to be outside on a terrace and the result to resemble a photo taken by a Leica 35 mm film camera.

The face of a young woman wearing sunglasses; the sun illuminates her face. The image resembles a photo. The image focus is on the young woman; the background is blurry. The background shows an urban landscape.

An image generated by the 5th version of Midjourney. I asked Midjourney to provide me with an image of a young woman with dark hair in the streets of Montreal on a sunny day, once again asking for the result to resemble a photo taken by a Leica 35 mm film camera.

View of a woman wearing a light coat, with hair slightly wet by the rain. The image resembles a photo. The image focus is on the young woman; the background is blurry. The background shows a street. The pavement is wet. There seems to be small raindrops on the camera lens.

Here, I asked Midjourney to provide me with an image of a young woman with dark hair in the streets of Montreal on a rainy day, once again ensuring the result resembles a photo taken by a Leica 35 mm film camera.

Editor’s note

During the 1st step of the Eductive’s lab on artificial intelligence [in French], which took place on February 2, 2023, 3 students from the Collège de Bois-de-Boulogne did a presentation on (among other things!) AI-powered image generators.

Gabriel Pellan, Éliane Rochefort, and Daniel Delisle presented Dall-E 2 and Midjourney, by focusing mostly on the second one.

Feel free to watch the recording of their presentation for a comparison of the 2 image generators and great ideas on the educational uses of these tools at the college level.

Presentation of 3 students from the Multimedia Integration program at the Collège de Bois-de-Boulogne during an interactive workshop organized by Eductive [in French]

Picturing poetry

When students are introduced to poetry, they tend to picture the text in a straightforward way. When I was teaching poetry, I would often start the semester by showing an abstract painting to the students: a painting by Riopelle for example. I would ask them to read into the painting, which was not easy. Since no one could provide a clear and direct interpretation, I would ask each and every student, while taking attendance, to say one word inspired by the painting. By combining all the ideas, we could formulate a hypothesis on the meaning of the painting. It would allow me to draw a parallel between abstraction in painting and abstraction in poetry.

The advent of Midjourney gave me the idea to go the opposite way. When some verses are too complex or some poetic images are not clear for the students, Midjourney can provide a visual representation. I tested the idea by translating poems by Quebec writers using Google Translate and submitting them to Midjourney. I was pleasantly surprised by the results!

Illustration showing the face of a beautiful young girl with a fair, greenish complexion, with her eyes closed and a peaceful demeanour, emerging from a tangle of stems, branches, and foliage. Her hair is hidden by leaves and flowers.

Illustration from an excerpt of La marche à l’amour by Gaston Miron (in French):
tu viendras tout ensoleillée d’existence
la bouche envahie par la fraîcheur des herbes
le corps mûri par les jardins oubliés
où tes seins sont devenus des envoûtements

English version submitted to Midjourney, translated by Google Translate:
“Your exitence will live all sunny
The month invaded by the freshness of the herbs
The body matured by the forgotten gardens
Where your breasts have become spells”

Illustration showing a blue shell, with pearly beige and golden reflections. The shell has a shape more similar to an ear than to a real shell. The background is completely blue.

Illustration from an except of Mon bel amour by Gaston Miron (in French):
l’oreille comme un coquillage
dans quel pays du son bleu
amour émoi dans l’octave du don

English version submitted to Midjourney, translated by Google Translate:
“the ear like a shell
In which country blue sound
Stirring love in the octave of the gift”

I especially liked the 2nd image, which made me realize that the F clef symbol has a shape similar to that of an ear, which I had never thought about before when reading this poem by Miron.

Picturing abstract concepts

In a Philosophy course, it could be an interesting exercise to ask the students to portray one of the concepts seen in class using an image (stoicism, humanism, or the notion of paradox for example).

Conversely, you could create an image yourself. Present it to the students and ask them to evaluate if the concept is well portrayed or not, and to interpret the different symbols found in the image. Why not organize a debate?

I tried to put it into practice by using Midjourney to illustrate existentialism. After a few tests, I decided to take it further: I asked ChatGPT to give me the prompt to submit to Midjourney.

Screenshot from a conversation between Julien Martineau and ChatGPT. Julien asks: “Peux-tu écrire les mots clés ou les commandes pour qu'une IA crée une représentation visuelle de ce qu'est l'existentiaslime?

I asked ChatGPT keywords and commands that I could later submit to Midjourney to create an illustration on existentialism. I requested them in French and then, had ChatGPT translate them.

4 similar images, all in blue and gray shades (with some black and white as well). All of them show the black outline of a man alone, from behind, standing on a beach with a cloudy sky on the horizon. In each image, geometric shapes are superimposed to the representation of the man on the beach. The first image is vertically divided into 2 parts: on the left half, the sky is grey, and the water is blue, and all is inverted on the right half. A cross is drawn in the middle of the image. The second image is vertically divided into 3 parts: the left part is grey, the one in the middle is whiter, and the one on the right is blue. White broken lines are drawn in the sky on the left and right parts. In the third image, a gigantic triangle is on the beach in front of the man, similar to a pyramid but with a two-dimensional appearance. The left half is blue, and the right half is white. Finally, the fourth image is vertically divided into 3 parts, just like the second one. The middle part is blue, and the 2 others are grey. An undefined shape resembling a lightning bolt is in front of the man, stretching from the horizon to him. In the middle of the image, a huge arrow is starting from the sea, pointing toward the man’s feet.

The 4 images provided by Midjourney from the written description of existentialism by ChatGPT.

4 similar images. All of them show the outline of a character, seen from the back, standing alone on a beach (except for the third image, in which the man seems to be in a desert), with a cloudy sky on the horizon. In the first image, traces in the sand form an arrow pointing toward the character. In the second picture, the character is standing in front of an enormous X that seems to be emerging from the sand in front of him (similar to a stone structure). In the third one, the character is standing in the middle of an X that seems to be traced flat on the glassy ground. In the fourth one, the man is standing inside a circle in which 2 large arrows come together in the middle.

The 4 images provided by Midjourney from the theme descriptions and graphic commands suggested by ChatGPT.

I did the exercise again with the concept of sophism, then with the philosophy of Nietzsche.

Screenshot of a conversation between Julien Martineau and ChatGPT. Julien asks:

I asked ChatGPT for written indications that I could later submit to Midjourney to create an illustration of the concept of sophism.

4 images show black and white patterns: disturbing and irregular tiled hallways with black and white floors, or black and white mazes. 2 images show many black outlines walking in these abstract spaces.

The 4 images provided by Midjourney from the written description of sophism written by ChatGPT.

Screenshot of a conversation between Julien Martineau and ChatGPT. Julien asks:

I asked ChatGPT for written indications, that I could later submit to Midjourney to create an illustration of the philosophy of Nietzsche.

4 illustrations in an infographic style, all displaying the outline of a person alone at the top of a mountain and showing text (incomprehensible; invented words). Each image highlights a triangle (as a frame around the character on the mountain) reminding the shape of a mountain. 3 of the images also show a sun (one of them shows 2). The images are in shades of orange, yellow, red, and blue.

The 4 images provided by Midjourney from a description of the philosophy of Nietzsche written by ChatGPT.

I believe these results to be extremely interesting! Pedagogically, it seems quite valuable to ask the student for example: “Why does AI associate existentialism with a person alone on a beach?” or “Why is the philosophy of Nietzsche depicted by a man alone on a mountain top?”.

There is certainly food for thought, and learning!

Creating a poster to analyze a novel

In the course Film & New Media that I was giving in French at the Collège Marianopolis, the students were reading a novel that would be adapted or has already been adapted for the cinema (such as Royal from Jean-Philippe Baril Guérard or The Goddess of Fireflies by Geneviève Pettersen). As a task of analysis, I would tell them to create a poster for the film adaptation of the novel. I would ask them to put the emphasis on one component from their reading: what the student would show on their poster had to represent what they would see as important to remember from their reading. There was a whole reflection process on what to show (characters from the novel, the themes discussed) and how to represent them. The substance and structure had a meaning: the text typography of the poster, and the colours; everything had to be carefully selected.

It was a valuable experience, but it was time-consuming. I had to take the students to the computer lab during a class to teach them how to use Photoshop. Then, I would invite a photographer to class to explain the basics of photography (the techniques, the compositions, etc.) so the students could successfully take pictures for their poster. At the end of the project, I would meet each team acting as the film director. I would question the students on their poster to validate their work of analysis and interpretation.

The reflection is quite interesting, but this assignment required some basic knowledge from the students to create an appealing poster. To avoid penalizing the less talented students in that area, I asked them to write a text presenting their intentions, allowing me to evaluate them properly when the poster was less successful on the technical side.

With Midjourney, restrictions related to the mastery of the tools become easier. I could use only one class to ask the students to explore Midjourney and decide on what they what to include on their poster: choosing the components to represent, choosing a style. The students could also present a poster without having to master the tools that were required before.

The use of Midjourney could save class time by implementing the project more quickly. However, my student liked being introduced to photography and learning how to use Photoshop. One option would have been to keep these parts of the course, but still begin by using Midjourney to create a mood board. The students could thus test and specify their film poster concept. They could test different styles, before taking the pictures and editing them in Photoshop.

Editor’s note — Protection of students’ personal information

Since the use of Midjourney required the creation of an account on Discord (although many students certainly already have one!) and the sharing of some information with the Midjourney robot, do not force students to use Midjourney individually. However, you can ask volunteers, making sure these people understand the issues at stake. Then, you can make teams so that there is at least one volunteer on each team.

A general idea: transformation as an analysis tool

In general, I believe that using transformation to make students reflect on a concept is a valuable practice.

When we ask them to turn a poem into an image or a book into a poster, the students have to reflect on what must be retained and how all this must be transferred into a new medium.

When I was teaching literature, I would quite often use this idea to present different concepts, such as figures of speech. For example, I used images to visually show what a metaphor is. However, I had to do some research to find images meeting my needs. Today, Midjourney allows us to easily create perfect images.

Midjourney is therefore an interesting tool for the appropriation of knowledge and initiating reflection.

To go further: a prompter guide to Midjourney

A man designed a spreadsheet to help you generate very specific commands to Midjourney, according to your needs. This prompter guide is available for free, as well as a tutorial.

For my part, as a pedagogical advisor at the Institut de technologie agroalimentaire du Québec (ITAQ), I used Midjourney to create visual documents for our 2023 ped days.

Posters of the pedagogical days in March 2023 at the ITAQ (in French)

What about you? What other pedagogical possibilities do you envision using Midjourney?

About the author

Julien Martineau

After 10 years teaching French as a second language at the Collège Marianopolis, Julien Martineau is now a pedagogical advisor at the Institut de technologie agroalimentaire du Québec, on the Saint-Hyacinthe campus, since 2022.
As a teacher, he loved getting his students to read and write, but even more so teaching them to reflect and use critical thinking skills. He believes humour, creativity, and dialogue deserve to play an important part in education.

Notify of

0 Commentaires
Inline Feedbacks
View all comments