How I used artificial intelligence to make a graphic novel (despite the fact I can’t draw)
I’ve always been an avid reader of graphic novels (or comic books, as you prefer), in particular those that are grounded in reality, relating a true story or an epic journey.
Until recently though, because I simply can’t seem to draw anything good, I never thought I could actually make one. I’ve therefore always stuck to writing and photography to tell my own travel stories.
But during my last trip to India with my fiancee Anna, I started using the smart filters from Prisma on my pictures before sharing them to my friends and family.
Prisma’s deep learning algorithms take you pictures and turn them into artworks based on a specific graphical style you select from their gallery. The artificial intelligence behind Prisma’s filter can be applied to any picture and render an artwork that follow consistently the logic of the selected style, in the same way an artist would paint different landscapes or settings with the same underlying style.
I really liked that by using the same filter on all the pictures I was shooting in Delhi, whether it was a colorful street, an imposing government building or an old architectural landmark, I was able to create a very cohesive and aesthetically pleasant account of our city explorations.
While reviewing all the artworks I had generated during our first few days of tourism in the Indian capital, it stroke me that by adding a few bubbles of text to my “Prisma album”, I could actually easily turn it into a graphic novel.
It turned out it took quite a bit more work than I had expected to achieve the above result, but my intuition was right.
Exploring a new style
I immediately started taking pictures with a different mindset, already thinking about the way it would look like when transformed into artwork.
I also needed to take into account how each picture could serve my story, acting more like a reporter of our own travel, documenting events that would have their place in a graphic novel but not in a holiday photo album.
Some aspects of photography like colors and depth of fields were suddenly less important, while the composition of my pictures was now my main concern.
For example, below are three pictures I shot from the moving jeep that was taking us up to Darjeeling, that I usually wouldn’t have taken. And while the pictures themselves are pretty bad, once transformed via Prisma, they turned out great to illustrate our ride.
Obviously, a big determinant of how my graphic novel would look like was the Prisma filters I would use.
To maintain a sense of harmony and coherence, I couldn’t use a different filter for every new image. But on the other hand, I couldn’t find one Prisma filter that I liked and that gave good results for all the different ambiances of our travel.
In the end, I selected one filter for the pictures I took in the dusty urban settings of Delhi and Agra, and another filter for the hills of Darjeeling and the Himalayans mountains of Sikkim. In the end, switching between the two styles help avoid too much visual fatigue for the reader and give a specific atmosphere to each section of the story.
Creating with the help of AI, the beginning of new era for artists?
Just as I don’t consider myself a writer for writting some stories about earlier travels, I of course don’t consider myself a graphic novel author after this experiment.
But by doing this exercice of transforming what would have normally been a simple holiday photo album (and some hand-written notes) into this new kind of artbook, I started reflecting about the broader use of artificial intelligence in art.
From a certain perspective, Prisma’s artificial intelligence is a co-author of the book. Of course, I took the pictures, I arranged them in a specific order to tell the story I wrote; but to some extent, Prisma is in large part responsible for the general look and feel of the graphic novel I created. I guess if someone wanted to take this technique further and write and sell an actual comic book using artificial intelligence, ownership would be a serious issue.
One could also argue that the humans behind the creation of the original artworks that were used to teach Prisma’s algorithms to reproduce a certain style, should be listed as the illustrators of the book.
Taking this possibility into account, you can even imagine that an illustrator could first feed a deep-learning algorithm with his own artworks in order to create his very own “filter”, before using it for his own creations. The natural next step would be for these illustrators to have the possibility to sell their styles, so that people like myself who can’t draw, could use them to illustrate their work.
As in almost every area where artificial intelligence enters, it’s fascinating to imagine the impact it will have, the jobs it will replace and the one it will create.
I’ve learned a lot through this creation process, and I’m quite happy with the end result which makes for a very nice souvenir of our trip to India. You can download the 44-page book here (it’s all in French, sorry) if you want to have a look at it, and I would love to hear what you think of the idea of using Prisma to make a comic-book-like graphic novel.
P.S: If you liked this post, you can give it a small ❤ or share it ;-)
Thanks Antonin, Flore and Emilie for your feedbacks and corrections, thanks Anna for traveling with me and forcing me to finish this ;-)