It’s the holidays and I’ve just been going where ever the wind has taken me lately. While I’ve started and restarted writing a few stories, not much is coming out at the moment - whenever I realize I’m just grinding metal, I usually just start drawing a story, then writing the story around those pictures. Then I’ll realize I want the story to go in a particular direction, illustrate then scene, then continue writing again. When I’m in the right frame of mind, I can illustrate an entire story in less than a day. When I’m not, it can become a real chore. It might just be the story, the content or just the way I feel. So I’ve been exploring the different uses of AI to see what it is capable of and whether I could incorporate it alongside my work.

The AI program I used was called Midjourney. Here is how you access it:

You need a program called Discord, which is kinda like a chat program which can be run through browser or downloaded.
Google ‘Midjourney’ and follow the prompts

The process is a bit convoluted, but once you have it all going its easy enough. You can get the program to generate images just from text, which is incredible, but you can also give it an image and it will produce images based of this! I wanted to see if I could use the program to copy my own artwork and reproduce it in different scenarios. Imagine having your character and art style already established and only having to write prompts to create entirely new artwork! Here are the images I used as a prompt to copy from:

Detective Petty breaking into a house

I used this picture because it was a full-body shot without any colouring.

The prompt I initially gave the program was that the character was exploring a dark and scary catacomb. The results are below, but what you’ll notice is that the characters ‘after shadow’ was brought across to AI artwork, so I created a new artwork that was pretty clear.

Here are results of messing around with this much clearer image:

Richos_a_photo_of_this_character_exploring_a_dark_and_scary_cat_286933d4-1009-4539-a380-65eac68d6952.png

Richos_a_picture_of_this_character_investigating_a_crime_black__4caed916-b209-4e98-acc5-6a4d7e34efa2.png

Richos_a_picture_of_this_character_investigating_a_crime_black__97955076-f13b-4d5f-903c-2fc4b19393ca.png

Richos_a_picture_of_this_character_investigating_a_crime_black__a36eab0c-036b-440b-b161-6b3b38cf3926.png

While there is no doubt that these artworks are fantastic, there is absolutely no connection with what I produce to this artwork. You can ‘dial in’ your requests to configure the output, but it didn’t seem to really matter how much I did it, I couldn’t get it to make something similar to my work - perhaps my work is too crap lol. I tried some new commands, namely ‘illustration’ and it produced some variations:

Richos_Illustration_in_this_style_investigating_a_crime_scene_J_20f78bb1-64ec-4fda-bc96-e0ce1cbf0708.png

Richos_Illustration_in_this_style_investigating_a_crime_scene_J_983d8ced-543b-4883-86f1-64bd5466013c (1).png

Richos_Illustration_in_this_style_investigating_a_crime_scene_J_983d8ced-543b-4883-86f1-64bd5466013c.png

Richos_Illustration_in_this_style_investigating_a_crime_scene_J_b05af3f1-d70e-4fe2-ba5c-18842b05a357.png

Richos_Illustration_in_this_style_investigating_a_crime_scene_J_e3969cf0-14fd-4c3a-9f87-79b8dd36b51e.png

Some further configuration yielded some different images again, all based on the one starter image. This time I added in ‘colouring in activity’ and changed the location to a graveyard:

Richos_illustration_in_the_same_style_as_the_picture_colouring__3581e078-9a2b-44b1-a61c-55fb6d7964c6.png

Richos_illustration_in_the_same_style_as_the_picture_colouring__caae120d-0ddf-4fb5-8213-2a740e359a7f.png

Richos_illustration_using_this_exact_character_in_the_same_styl_69e32cf4-7931-4c76-9768-3d036b280338.png

Richos_an_illustration_of_this_character_in_the_style_of_Chika__635be882-f4b7-45e5-b288-ebf02e4b71af.png

By this point I think I went off and did some more research on how to get the program to produce the images you were after. I think I got a little closer to the mark, but by this point it became pretty apparent that I wasn’t going to be able to use this as a shortcut in my own work.

Richos_man_on_a_ladder_in_a_catacomb_full_of_ghosts_scared_phot_dec4bffc-85d2-469a-a976-6a3013539e56.png

Richos_man_reading_a_book_dof_35mm_lens_natural_lighting_upligh_a1e63efe-1569-4bb8-8aba-92bd3fd062d9.png

Richos_man_reading_a_book_dof_35mm_lens_natural_lighting_upligh_baf0a16a-7310-46b6-a495-5a801129d76a.png

Richos_man_resting_in_a_hotel_colouring_in_style_photorealistic_062e9cf3-675b-4527-a45c-438e26f3eaf4.png

Richos_man_resting_in_a_hotel_colouring_in_style_photorealistic_e08f6e4e-2bee-41be-8f1e-b3ed76a8da7e.png

Richos_man_wearing_cardigan_praying_at_a_shrine_illustration_66cfdbb3-1cc3-4114-92b5-3e3e3aede9a6.png

Richos_motel_room_sitting_chair_colouring_in_style_c78e46a8-92d6-4709-9cbd-b3236c9a8861.png

Richos_illustration_man_sitting_at_a_yatai_Wes_Andrerson_style_3555d11c-f7db-4106-a745-5897f99de0fd.png

I also found that as the program churned out more and more images, I found I was ‘desensitised’ to the capabilities of the program. You probably can’t tell from the artwork, but I tried repeatedly to get the character to do various things like climbing, sitting, walking etc. but the program just didn’t cooperate, despite looking up several tutorials. There is a total of I think ONE IMAGE where it actually listen to what I wanted (see if you can find the image where the guy is sitting), all of which led me to the conclusion that I couldn’t really use this technology to produce a series of images for a story. It could make a great standalone image, but getting it to work the way I wanted would probably take longer than it would take to draw it myself.

Personally, I can see some uses from what I found. For a teacher with little time, they could produce a great picture to accompany a bland worksheet. The landscapes and scenes that it created were instant and could provide good inspiration for myself. Were I unscrupulous enough, I could always take those images and use them as a basis for my own work, but this is where the whole ethical debate comes into it. Whose work was it originally? Where did Midjourney get the artwork from to produce this? Could I later find out that the work I copied from the AI was actually someone elses and accidentally produce something breaking copyright? If it were a free program, then I might just use it to help me compose illustrations of my own but for $11USD a month is worth it? Certainly not.

Now don’t get me wrong, each and every image it produced was vastly more detailed than my own and better composed, but there is just something ‘off’ about the artwork. Like, I have no connection with the work and if I showed it to my students I suspect they wouldn’t connect with it either and I think it all comes down to authenticity. Perhaps this is just another facet of uncanny valley, where AI-generated artwork seems created by a human, but somehow obviously aren’t. Also, if I could produce artwork of this quality, I’d be working as an illustrator, not a teacher and my students subconsciously know this. Maybe I’m just salty a computer is running rings around me, perhaps it’s just my personal experience with the program, either way I’m sure it will change with time and experience.

If nothing else, this whole experience has given me a few hours of PD on detecting generated pieces of artwork. Now if I see students produce work where the detail is just a little too good, the character has three arms and is just standing in the middle of the scene repeatedly, I’ll know how they did it.

As a special bonus, I’ll include some images which included the names of various artists to guide the program, see if you can guess which: