Make Comics in Stable Diffusion!

In 2022, I made a video about making comics in Midjourney, and now I wanted to revisit the topic using the latest technology in Stable Diffusion and Photoshop's new Generative Fill tool.
0:00 - Some historical information
2:16 - Why I am switching from Midjourney to Stable Diffusion
3:51 - Stable Diffusion
5:05 - My first SD comic
6:08 - Consistent Characters
7:40 - SD Interface and prompting
9:14 - CivitAI
11:43 - The script
12:43 - Prompt Stealing
16:40 - Generating Characters
28:28 - Prompting a background
31:46 - Photoshop Beta
37:25 - Generative Fill
44:33 - Wrap-up
masterpiece, best quality, female, soldier, white tank top, sleeveless, green pants, brown belt, brown boots, curly brown hair, action, standing, arms crossed, open mouth

[@jabelardo]

An optimization of your workflow is to generate a batch of images and only perform high res fix for the ones you want to keep. In that way you are avoiding the extra time required for the high res fix for images you are going to discard anyway. The only extra thing you need to do is to copy the seeds of the images you want to keep and redo them a second time with the high res fix enabled (and a low denoise level for the fix to avoid the fix to change the image composition)

[@hyperbolicfilms]

Thank you! My workflow has changed a bit since switching over to SDXL, but that is a tip I can carry over.

[@boscoe5334]

That is the reason I use Stable Diffusion. I don't have to pay Midjourney and with so many different styles to choose from and now the SDXL model. Midjourney has a sameness to the art that makes me able to pick it out of a crowd. SD has so much going for it absolutely NO censorship. Which is awesome. I could have some super sexy steamy scenes. (Add in Lora's and the ability to use Controlnet to control the pose... It's near perfect for what I like to use it for.)

[@WalidDingsdale]

thanks for sharing this amazing walkthrough of combination of SD and PS. This is my first comic art class, which interests many. keep going with more such prefessional comic insight and ideas.

[@rishiisrani7783]

This was a great tutorial. Finally was able to install Stable Diffusion. Now to figure out how to control the prompts to get what i want. I went to school with you in the 3D Design class. Ai would have made life so much easier then. -Rishi

[@hyperbolicfilms]

Glad you liked it. Getting repeatable results or what you actually want is the hardest part. I tried to redo this page using a different checkpoint that has more of a 3D look, and it changes the details so much that it's incredibly hard to get something consistent. It is a huge jump from what we had back then, but harder to control.

[@itanrandel4552]

Excuse me, how do I add the speech bubbles?

[@TheSoleProprietor]

Sometimes, the advanced tools does not give me the option of a batch count. just everything else like choosing the model, adjusting the random seed number, etc.

[@Elaneor]

When I found this video, I hoped the authoк would use IP adapter of Control net extension ...

[@godofdream9112]

Sir please make more video on this topic... I want to make comic.

[@alibaba-wy1iv]

How to make every generation, is the same character? I heard some people using own lora training. Can you make a tutorial for it? Thanks

[@TheSoleProprietor]

I made the mistake of trying to prompt a scene where I square off against Bruce Lee, as a 12-year old challenger... and it banned me! Sometimes these stupid ai engines get their "minds in the gutter" and misinterpret under-age prompts as being child porn. While that is understandable, it was not what I intended ... to be deliberately offensive. Rather than having to bug the administrator and get unbanned, I don't log in to my account, anymore and use the SD XL 1.0, anonymously. I still would like to learn some of the AI tweaking tools, such as random seed number and Base Guidance and negative prompts and how to use them more effectively.

[@CessarVillarreal]

burns my eyes see people using chatgpt or SD white theme 😂😂

[@user-pc7ef5sb6x]

controlnet can fix all the consistency problems.

[@hyperbolicfilms]

I've started using it in the past two weeks, and it definitely can help a lot. Canny in particular is giving me great results for replicating the clothing of the person in my source image.

[@honzo1078]

Nice video, but in the simple style you're using, it would probably be quicker, easier and more controllable to just draw it.

[@hyperbolicfilms]

For someone who is a better artist, perhaps. I just spent a year drawing 24 pages that were mediocre at best. This methodology would also work for many other art styles that can be loaded as checkpoints into Stable Diffusion.

[@dwainmorris7854]

Yes, you're absolutely correct, it has too much censorship. That's why I'm gonna go ahead and pay that $2000 to buy a gaming computer. So I can run Stable de fusion on it directly. I'm tired of being toe note go ahead. M***********, it won't let you do fight scenes. It won't let you let the figures hold weapons. And that's a plus if you're gonna make comic books. If you're gonna make superhero comic books you're gonna have to show some violence and midjourney as well as others like leonardo I won't let you do it

[@Xandercorp]

My dude, =you don't have to wait 5 minutes to load one of those... Why is your install so slooooow?

[@arielmorandy8189]

I thing you really need to practice stable diffusion before making images. You are not using SD properly. You need to use img2img , not text2img. You cannot refine an image, by trying to have one single prompt to generate your image. Try to look at Sebastian Kampf tutorials.

[@crimson_fire_Dragon1]

darn so you need to pay for mid journey and cant do what you want