💬 Some fun with Comic Life 3 and AI images 🖼
I previously wrote of my adventures and amusement with Comic Life 3. I’ve also discussed toying around with AI image generator. So, what happens when you combine the two? Is it possible to make comics using AI image generation and the correct prompts? Simple answer is yes.
The idea came from my kids who brought up Halloween where one wanted to be a zombie character. The other asked why not Spider-Man? Even though there is Venom, the answer was a simple, Spider-Man isn’t scary. Zombies are fun and scary. Spider-Man isn’t a zombie.
But, could he be? Could you stretch the bounds for Halloween?
I went to the AI generator to run a couple prompts through the processor. This set the wheels of imagination turning.
The first result was basically a green Spider-Man, which was kinda cool, but not a zombie by any means. However, it was a bit Green Goblin-esque, maybe Riddler inspired, which sparked a story idea.
After a couple of tweaks and an emphasis on certain keywords it started to come together. The AI image generator, really wants to hold onto the traditional Spider-Man aesthetic. If you want to change him, you have to nudge pretty hard.
The point is, AI is quite good at generating copies of known images, so creating copies of Spider-Man was pretty easy.
I’m in no way drawing Spider-Man, that takes talent. At best, I’m playing Superhero dress up. And it worked.
With that, I took a variety of images and put them together into a comic story where Spider-Man is bitten by a Green Goblin creation and turns into a zombie, with the proper aesthetic. A totally lame story, but my audience is a 5 year old, I don’t need much plot and reason to sell the tale, a few simple links in the chain will do.
This was more of a proof of concept, but it was quite easy to create images of Spider-Man in different places, such as on the side of a building, standing in the street, looking out over the city, etc. It was just as easy to create Green Goblin images in a lab, working on a plague formula, and setting things in motion. The more of a known commodity you work with, the better the results.
Had I planned it better, I would have generated images that better fit the panel types such as landscape and portrait, not just square. Even with that, it worked out quite well, and I was able to assemble a multi-page story to show Spider-Man getting turned into a zombie for Halloween.
Comic Life is for page and panel layout. AI made some pretty decent images in the right kind of style. From that standpoint, making a comic from AI generated images was pretty straightforward. That is not to dismiss real artists and story tellers who make real comics and draft real art. We’re talking AI images that are good enough to entertain a 5 year old.
This was literally to see if I could make a story to amuse one of my kids, and it worked, so I call that a success. This means I can create amusing stories that fit their whims, like what if Spider-Man got trapped on a tropical island? What if Spider-Man were in the dessert? Or taking two ideas that don’t quite belong together and making them work.
I can’t draw beyond stick figures, but if AI can give me images of popular characters in different places and scenarios that fit a story, I think that’s pretty awesome.
Make no mistake, there were some consistency issues. Even Spider-Man had some odd choices thrown in by the AI. In some cases his mask was completely missing, and in more than one cases I got extra limbs. Some times I got the wrong character altogether. It seems AI doesn’t want more than one central character in the image.
Green Goblin was far more problematic, where two images bore little resemblance to each other. That might have been fixed by using a consistent seed value and some other prompt tricks, but it clearly lacks the skill and craft of a real artist. It was the same sort of problem, where one image had a mask, the other didn’t. In one image, the outfit looked one way, and in the other quiet different. Not to mention the shading, coloring, and backgrounds weren’t the same. Ask for two images in a “laboratory” and you’ll get two completely different backgrounds.
Those goofs aside, it was a fantastic little project. It was easy to get the layout in Comic Life 3 and add the lettering, captions, and speech. Like I said my image sizes weren’t optimal, so that’s a lesson for the future. There were continuity errors because I didn’t plan out my story, that’s another lesson.
Still, how cool is it to create a comic from scratch, with quality images, present them to your kids and get a big smile and laugh out of the deal? It wasn’t a quality piece of work, but they enjoyed it, and I enjoyed making it for them. AI and Comic Life 3 work pretty well together.
I’m now wondering what story we can put together next? What action can we create? What creative story can I help bring to life? We’ve got plenty of time before Halloween, so who knows where this will lead.
But, it also makes me wonder, if I get excited and amused over this kind of project, what about someone who wants to pursue art and comics as a career? A person interested in storytelling, comic layout, and design can generate images to fit their narrative rather than random ones grabbed from the Internet. It’s not random Spider-Man, but Spider-Man following your chosen actions.
And someone who wants to hone their skill at drawing, inking, etc, can have AI create a slew of different images, in different scenes, different poses, and different actions. From there, that person can study how it was put together, trace, copy and enhance to their heart’s content. It’s not the same as getting instruction from a real artist, but it seems like valuable schooling.
Or even simpler, it’s some goof like me who wants to put together something nutty for his kids to make them laugh.
Either way, using AI to generate images does have a benefit. It’s a fun exercise and gives non-artists a way to bring visual appeal to a story. It’s even better when your audience is young enough not to notice AI flaws and inconsistencies.