[www.reddit.com]

And what if AI art and AI chats get better and had access to tens of thousands of comic book images

could one see a comic book being drawn by dead artists in there style and look like new -original comics??!!!




What will it mean???????

I did a bunch of pictures using OpenAIs DALL-E model. It can generate images in the more prolific artists styles. However, what you'll notice with many of these is it does an atrocious job with drawing motion, like way worse than it does with hands and it does hands badly. Notice that the graphic novel you sent had a cat as the main topic, but the cat was basically always sitting. It also had humans, but most of them had their limbs cut off. Until that happens, I can't see it replacing an artist. However, I could see the situation where an artist works to digitally repair the art, by fixing hands and limbs and recoloring.

It might save a decent amount of time and it's something I considered though I haven't attempted it yet, getting the images all generated is a decently time consuming process. Especially getting something the model doesn't have a lot of in it's data. For example, I tried to do an adventuring party with a gargoyle in it. Didn't work at all

Isn't it also difficult to keep character appearances consistent from image to image?

Yes, but it's manageable to some extent. You can feed the model an image to base your new image off of and in which case, it can get very similar repeat images though it will often duplicate pose in that case, so practically nearly impossible to get the same image in different poses, like you'd need to do for a comic, but honestly, I think, and I've seen other models that could make minor changes to an image after it had been generated, so I think that issue is likely to be resolvable. Hands issue and other such things seem more difficult.

Clarkesworld Magazine closed their open submissions because of ChatGPT influx of submissions!!!

When arte tools used as tools one thing but

tools ( computers say ) used as a crutch another?

I take back my previous comments. I was wrong. This technology is progressing quickly. I'm able to generate much better images already than when I said that hands and teeth were giving issues. Now, it seems like the ai software can generate hands pretty regularly. It seems like there's still some issues since it's hard to get consistent images or specify poses though I'm looking into using a 3d model architecture to sort of use as a wireframe to direct the computer to draw around. It's pretty neat stuff and so much of it is coming out as free and open sourced data. I'll let you know how it works out if I get it working on my computer. The computer demands for these models are pretty high though not incredibly demanding.

I heard more consistent character images could be gotten by making characters look similar to old and new movie stars.

And writing with better and more detailed PROMPTS are working for many would be and already - writers

I've no doubt that an AI could write and draw a graphic novel. The better question is - would it be something anyone would want to read? The art and writing would likely be pretty derivative and not particularly original - at least at this stage in the AI's development.

In a sense - CHAT-GPT is doing the rough draft legwork and a human is improving it like layers being added.

If left UN-Obstructed can run for a touchdown- not the best but getting better.



[www.aidemos.info]

[www.cnet.com]

not perfect but What a Great mini hobby for would be comic fans who Can't draw!!!

Interesting