What to expect when using Ai for Comics. I tackle a very hard comic page.

Рет қаралды 20,135

Жыл бұрын

There is no doubt that Ai assets are making it easier to turn out great looking visual stories, but it seems like they are only good at specific types of angles and shots. Today we test midjourney and dall-e by taking on a single page 13 panel ai comic and try not to compromise on your vision.
In this video adventure with me through the process of ai art generation as I try to populate a comic page filled with fingers, hands, wardrobe changes and even a teeth brushing sequence. All of these images are difficult for ai to give usable results for. We take an honest loot at ai and how it performs in a real world comic setting.
Read the Camp Peculiar ai Webcomic on Instagram
/ camppeculiar
Get my free Instagram Comic Template and more Ai art comic resources at:
camppeculiar.com
Get my free Speech Balloons for Premiere :
bit.ly/3Gf14vM
#midjourney #dalle2 #aiart #aicomics

Пікірлер: 57

@paulligatto8476 2 ай бұрын

What I like about this video is it shows just how far AI art generators are from being able to do good comic books, despite being amazing at single images.

@amdrecording Жыл бұрын

Thanks for doing this... love watching and learning from you!

@reidmoto Жыл бұрын

Very useful information for me, thank you for creating and sharing this video. Though I'm new to the MJ platform, I've already encountered some of the challenges you showed. It was great to see how you attempted to solve them.

@morpheus2573 6 ай бұрын

Great job. Clever work arounds. Thankfully, now one year on, AI is bringing us breakfast in bed, cooked to our specifications and shaving us with a virtual marshmallow. Thank you, The Jetsons!

@kennymonsters Жыл бұрын

I'm watching this 2 months into the future going through the same MJ struggles trying to design page 1 of what I'm now realizing is an all black diamond graphic novel project. I'm going to keep plugging away at it, but I'm very hopeful that MJ's capability progresses further because although I'm a great story teller, my art skills are not to the level needed for the project.

@JoeKawano Жыл бұрын

Whoa…🎉That’s a cool page you put together there! Wow

@fadimantium Жыл бұрын

thanks! This was a brilliant explanation!

@camppeculiar Жыл бұрын

Heya Fadimantium. Thanks for the comment.

@COLORINGBOOK888 9 ай бұрын

Great video

@addict85 Жыл бұрын

Good stuff!!! I've been contemplating creating a comic series using AI art as well. But Midjourney v4 can do an aspect ratio of 2:3 or 3:2. I've been playing with it the last few days.

@chariots8x230 Жыл бұрын

It’s very interesting to watch your approach for using Midjourney to create comics 😊 I find it fascinating to see how this technology evolves when it comes to creating comics. I wish there were more options to teach Midjourney about what we want to achieve. I’d like to use 3D models that I can pose and show from different angles, in order to teach the AI about the type of scene that I want. It would be interesting if we could combine both image references that we create with 3D models, as well as verbal prompts, and use these things to teach Midjourney.

@camppeculiar Жыл бұрын

Thanks for the comments and thoughts Chariots! I agree that 3d input for composition and structure modeling would be interesting. Seems like modeling the ai on generic 3d poses image versus other artists work would be a step in something like the right direction as well.

@chariots8x230 Жыл бұрын

@@camppeculiar 3D models are useful, because not only can you pose the model to get the exact pose that you want, but you also have control over the camera angle. It’s not easy to do this using photo references, since it’s very difficult to find the exact pose you want, and it’s even harder to get that pose from the camera angle that you’re looking for. 3D models offer a lot more flexibility when it comes to this. I would want to use 3D posing apps like ‘Easy Pose’ in order to create references for teaching the AI about the type of pose that I want.

@COLORINGBOOK888 9 ай бұрын

If anyone has any questions let me know.

@MasonIce1999 Жыл бұрын

Thanks for the video and info. I was wondering if you could use negative prompts to help with your bare feet issue. Maybe try --no shoes or something to that effect.

@sofarsogod Жыл бұрын

Interesting, I guess so many people with great ideas will be able to create amazing comicbooks.

@gloriouschoices 9 ай бұрын

Create? Yes. Amazing? No

@NightPhoenixPress Жыл бұрын

My guy say, "After this, it's anyone's guess!" 😂😂😂😂 I finally finished my book. It's on IGG, so fingers crossed.

@lilowhitney8614 Жыл бұрын

I think you made it harder on yourself by using midjourney for the images, especially for stuff like pose and perspective. Img2img is the way to go with those as you can set up pretty complex poses and have them followed very closely.

@camppeculiar Жыл бұрын

Hi Lilo. Thanks for the I used img2img a few weeks ago and the execution times were really long. Is that normal or is their a fast mode? Maybe it’s time to check it out again, appreciate the comment.

@eiyukabe Жыл бұрын

@@camppeculiar Have you considered Stable Diffusion? I am more a tech guy than an art guy so I don't know how it compares results-wise to Midjourney, but if you have a decent GPU you can run it locally and generate images pretty fast (no competing for server load as YOU are the server). In my limited understanding the results of all the big names are pretty equal so I stick with what I can run most easily.

@petercook6570 Жыл бұрын

Great video. Do you use a reference image to try and keep the model the same or are you just hoping MidJourney will not drift too much with each prompt?

@camppeculiar Жыл бұрын

Hey Peter. Great question. I think there is a lot to be said for reference images as both a visual reminder and as a prompt to try and get MJ to stay close to a particular style. To be honest I should use image prompting more, but I just have some odd reservations about it. Maybe it’s the time it takes to find the image, all of 1 or 2 minutes, copy the url, hope it’s url with a structure MJ can accept. It works well now and I should have used it on this project but my style is to generate a lot of grids and work on multiple panels in my head at the same time. So, I’ll type in prompts that don’t specifically match any panel in hopes of finding bits and pieces of gold. I should go back and use image prompts for the harder panels but I’m just wired to generate hundred of images and then make it work with what I was able to get, feels more like dumping a bucket of legos onto floor and I kinda like that. I’ll do an image prompts video to force myself back into that prompting practice.

@bink84 Жыл бұрын

Thanks for the video. For now it looks complicate when we just want(or can) write script and have a comics draw by AI. But maybe with time and update it's gonna be easier ^^.

@jannarkiewicz633 6 ай бұрын

I'm in the same boat

@35matinee Жыл бұрын

It would have been nice if you had included the prompts, but hey, great work (it's obvious that you have drawing skills).

@rud200 Жыл бұрын

Ты крутой 😎 дядька , респект ✊. Мощный контент, спасибо 🙏

@NightPhoenixPress Жыл бұрын

I did my comic in Wombo AI. It’s a lot easier for consistency. I use Midjourney for covers.

@TripleSixxxRonin Жыл бұрын

Which a.i generator could I use to color comic pages as I dont have a colorist rn

@ThatGuyNorm Жыл бұрын

How do you get the AI to use the same characters you create throughout the comic? Currently every new scenario created geerates a totally different looking character.

@benitofranklyn4237 4 ай бұрын

I'm having a hard time having my charter walk away in the distance, like the end of Rambo or Hulk, any tips?

@madmania5080 Жыл бұрын

hey oooo cool stuff

@camppeculiar Жыл бұрын

Whats up Mad, hop your productions/issues are going well.

@madmania5080 Жыл бұрын

@@camppeculiar going great! Just published issue #4!

@FutureTechPilot Жыл бұрын

Hey man, great video. As far as getting a consistent model, I think you can do it through image prompting ... at least, that's what I would try. Image prompt a face or person you want, write your prompt, and adjust the weights if necessary -- I think you could get a consistent character that way

@camppeculiar Жыл бұрын

Thanks for the suggestions Future Tech Pilot!

@MarcusHiggs Жыл бұрын

Great job Aaron. A similar idea regarding the 'getting out of bed' image. When my text prompting skills are not meeting my needs, I've found Image prompts are work exceedingly well in version 4.

@NoCodeFilmmaker Жыл бұрын

Actually you can now use -an aspect ratio of 2:3 to get a more vertical rectangle image. I use it to create images for movie posters.Great work, I always look forward to your uploads. You really inspired me to get heavy into Midjourney. Super props on the video showing the neural filters in Photoshop :)

@camppeculiar Жыл бұрын

Hey NoCode. Thanks for the update about the aspect ratio support for v4, I’m a 33:1 guy ;) but 2:3 and I assume 3:2 will be a nice addition to the endless squares I’ve been generating. Thanks for the comment!

@NoCodeFilmmaker Жыл бұрын

@@camppeculiar They also added the ability to remix in v4

@troglodytezero1697 10 ай бұрын

Looks like writing a story with AI assistance is easier than ai drawing help.

@shootlukumar 6 ай бұрын

What to use to generate panel

@awr3684 Жыл бұрын

How many time it took? To create this. Imean

@digitalsketchguy7844 Жыл бұрын

Just don't ask it to draw hands.

@joytekb Жыл бұрын

Thats amazing but i am waiting for language

@bradshipp5746 8 ай бұрын

I would just draw it myself. I use AI here and there for inspiration, but nothing beats actually having the right skills

@radcanadian7365 Жыл бұрын

no no no please don't give the big companies ideas

@camppeculiar Жыл бұрын

Hey Canadian, i appreciate the comment and perspective, truly I do and would tend to agree with your position, there have already been ai comic projects taken on by non indie studios. Check out Campfire entertainment’s Bestiary Tales. We are creating in interesting times.

@radcanadian7365 Жыл бұрын

@@camppeculiar For sure dude, and I'll have to take a look at that Campfire Entertainment channel. Hope you didn't think I was being rude or anything, that was the last thing I meant. Have a great day!

@camppeculiar Жыл бұрын

Not at all. Totally appreciate the comment, it’s a weird time to be a creator and I never assume to know how someone else feels about ai, I don’t even know how I feel about it day to day. Not rude at all, I get what you were saying, totally. Have a great weekend.

@radcanadian7365 Жыл бұрын

@@camppeculiar You too! Glad I didn't come off as rude.

@gloriouschoices 9 ай бұрын

Why so many panels? If you acknowledge the pacing needs to be three pages... at print size, you've created a claustrophobic environment . Plus the details will become a muddy mess at print. your panel layouts break a comfortable flow for the reader. Z the flow. Left to right before up down. Youre using middle columns to be read up down before moving to the right column. You should go from the tall p01 to top middle column to top right column. Then bottom middle column to bottom right column. And that flow should be followed in the lowest third of the page. You're not "leading the reader in the proper story flow. You have to many layouts where your dude is leading the reader to the previous panel. I would guess you could have drawn this page by hand way quicker than fumbling with ai

@jerrygraves6531 Жыл бұрын

You look like the type of person who would try to start a KZfaq channel based off a scam like AI art.

@learningcurve7308 Жыл бұрын

real

@PolytronikDigital 6 ай бұрын

dude....wtf....thanks ,,, this video certainly proves that "A.i" " ART" is just far more hassle than its worth the time to fuck around with. From the moment you started youre compromising choices to suit the software. we in the actual human art industry have nothing to worry about you with you "prompt-pissers" thinking you can draw using a robot. hilarious.