At BANANA COMICS, we aspire to deliver as many comics as possible daily, but due to financial constraints, we often find it challenging to update daily, which is disheartening.
Amidst this, I noticed someone on Twitter claiming to be an “AI Manga Artist” and thought, “If AI-generated artwork is interesting, why not give it a try?” So, I decided to experiment with it.
We also decided to fully entrust the story creation to ChatGPT.
The ChatGPT reacted as below.
Title: “Degu the Politician”
- First Panel: Degu protagonist stands at a podium, starting his speech with a serious expression.
- Speech: “Everyone, now is the time to change our country!”
- Second Panel: The audience is intently listening to Degu’s speech. In the background, other degus hold a sign that reads “Change!”
- Speech (Audience): “Oh, what powerful words!”
- Third Panel: The protagonist continues his passionate speech, but a faint noise distracts him.
- Speech: “If we all unite, we can surely… huh? What’s that noise?”
- Fourth Panel: Looking under the podium, he sees a young Degu busily nibbling on cheese.
- Speech: “This is no time to be eating cheese!”
This comic strip concept uses humor and the charm of degus to engage the audience.
t’s subtly realistic and cute… There’s no characteristic tail on the degu, but that’s okay. Let’s continue with the second and third panels..
Don’t suddenly change the art style!
But I’m getting tired of making orders to AI, so let’s move on to the fourth panel.
I will connect the images I made in Canva. I changed the layout a bit because the panel division and dialogue felt a bit cumbersome, and the mouse in the fourth frame was adorably silly.
If you asked me if it’s super funny, I might feel a bit iffy, but if it popped up for free on Twitter, I think I’d give it a read.
Consideration of Problems and Challenges.
The key challenges identified from using AI for drawing manga include:
- Uncertainty about whether desired art styles can be created.
- Inconsistency in art styles.
- Difficulty in composing scenes with multiple characters.
Although there are AI tools that can generate images based on registered character designs, they still struggle with scenes containing multiple characters. Additionally, while AI can automate drawing, controlling the AI’s output to get the desired results can be time-consuming, often requiring several attempts. Moreover, AI-generated images often lack depth and require further editing to emphasize specific features, which might be important in manga. Thus, for those who can draw, manually creating illustrations might be faster, but AI can still serve as a helpful tool for those unable to draw.
Conculusion
While drawing manga with AI illustration tools is not straightforward, it seems feasible to use AI to create certain levels of artwork.