Why should A.I. have all the fun?

I like the idea of receiving prompts and then coming up with a drawing, spontaneously. It's a good creative exercise for me and it's also fun.

Instead of letting A.I. have all the fun in that area, I asked my 10yr old daughter, Ollie, to feed me some prompts bit by bit and I'd create a comic, panel by panel. I would be the Real Intelligence (R.I.) she'd use to generate a 6-panel comic. We were sitting together on a 30min bus ride, on the way to her art class downtown on a weekend.

1. I asked her for a place. Park. A person or animal. Fox. An expression. Grin. And I drew a quick sketch in red pencil on paper. Just the pictures. No dialogue.
2. I said, I wanted to do a close-up now. Give me an expression. Grin.
3. I said the fox goes to a new place. Give me a place. City.
4. Give me a new character. Girl. Describe her. Headphones, long hair. Expression. Grin.
5. I now want a close-up. Give me an expression. Crying.
6. They move to a new place. Give me a place. Farm.

While my daughter was in class, I drew out the pictures with a dip pen and watercolours and also came up with a dialogue what would tie all the pictures together into one cohesive story. 


Later on, I erased the red pencils and cleaned it up a bit on Photoshop.


I also tried to generate this story using ChatGPT. I'm not great with the prompts yet. But this is what I prompted:

 Make a 6 panel comic. Portrait, A4. 1. A fox walks through a grassy park. He has a big grin on his face. 2. Close up. He looks back. He still has a grin. 3. The fox walks into a deserted city block. He walks in the middle of the road. There are a few cars parked on the side. 4. He walks beside a girl sitting on a car. She has headphones and long hair. She looks at the fox and grins. 5. Close-up for the girl looking sad. She is crying. 6. Wide shot of a farm. We see the girl waving to the fox who is walking to a house.

And this is what the A.I. generated:


It didn't seem to understand what a 6 panel comic was. And the panels still needed revising. Why did it draw a human looking fox?


I tried again, asking it to make it more hand-drawn. 


A.I. can generate pretty good quality drawings, I must admit. With some tweaking, they could be great. Perhaps, I'd get better results if I used Midjourney intead of a free version of ChatGPT. The story still came out very different from mine. I realise I need to be more skilful with my prompts. It's a little frustrating not getting exactly what I wanted even on the 3rd try.

I'm not sold on creating my comics this way as I have more control over how I want my scenes to be, just using my own hands to draw. I still like the act of drawing something myself, rather than having the drawing drawn for me. There is pleasure in the process of creating the drawing myself. I'm curious to see how we artists will have to adapt having A.I. around.

Comments

MiataGrrl said…
Maybe the fox looks human because your prompt referred to "he"? But it's pretty bad overall...no sense of continuity, sequence, and the expressions are terrible, so generic. Whew -- RI is still better. 😅 But the best part is your collab with Ollie. ❤️
Tara Kate said…
I loved your RI comic and your spontaneous collab with Ollie. How fun! What did she think of the comic? More of OllieGPT (G = generated P = prompts T = ToComics).