11
35 Comments

I've open sourced my AI comics website AMA

Thanks to the great support I've received for the AI-generated comics here, I've decided to open source it.

You can check out the project website at https://dailyaicomics.com and prepare to be wowed by the incredible collaboration between ChatGPT and Stable Diffusion.

What's included:

  • ⚡️ Async jobs for generating content
  • 🤝 Integration of both ChatGPT and Replicate API
  • 🤖 Google authentication
  • 💻 Minimalistic Chakra UI template

The source code for the project is available on GitHub at https://github.com/infomiho/dailyaicomics.

The million-dollar question: how would you monetise this?

  1. 6

    sounds like good idea, it should be fun working with ChatGPT

  2. 2

    The potential is awesome for sure, but the problem I see here is that there is no consequential style. Every piece is completely random without any coherency. That way you cannot use it to create a consistent series of comics.

    1. 1

      You are 100% right, that's the things that kills the magic, is looks poorly executed.

      I've had luck with specifying the style with technique (use black and yellow ink). But now I switched to various famous comic artists "in style of XYZ". But ... that's too broad it seems.

      For tomorrows edition I'll experiment with styles again, tune in and see.

  3. 2

    You are a genius for coming up with this idea.

  4. 2

    That's a great idea! I think it'd be a great idea to just start an Instagram account with them

    You can get a lot of monetization ideas from there and also use the platform and the following for promotion later

    Also could be fun to run an account like Artificial Comics or something similar :))

    1. 1

      Oh, and ideally I would automate that as well, right? Maybe post the best voted comic each day.

      Nice idea!

  5. 2

    looking good! here's an idea: reader's can propose and vote on a prompt for the next day's comic. If you prompt gets accepted you get some karma or sth :D

  6. 2

    I love the fact that you decided to open-source your project! On the other hand, it might hinder your plan to monetize it because of how easy it is to replicate the generation.

    You could partner up with brands to make sponsored content. I also like the idea of making unique merchandise - selling only one of each.

    Good luck with the project!

    1. 2

      I guess there is isn't much that is hard to replicate, why not share it. Maybe later on some of parts become closed source?

      You could partner up with brands to make sponsored content. I also like the idea of making unique merchandise - selling only one of each.

      Oh, that's cool. "One of a Kind T-Shirts". I think some dude is making his business with suggestions from GPT-4, and GPT-4 had this similar idea of selling merch with Mid Journey's images.

      1. 1

        Websites like designbyhumans.com could help validate the idea.

        You just create the design and they print it and ship it when someone orders it.

        They take a big commission of course, but as a start it could be interesting to see if the idea works while getting traffic from e.g. advertising on Instagram

  7. 1

    I am monetising some very similar to this by selling physical books. Drop me a line if you want to chat or ask me here. TLDR use print-on-demand services aimed at photobooks, and get industry pricing not retail if you can.

    My thing is at https://artificial.pub

    The reason I went in to non-fiction was to avoid the visual continuity problem, which I didn’t have the skills/time to overcome. But I’m really enjoying non-fiction!

    In terms of outreach I’m now mostly focusing on Insta ads + bricks and mortar stores in Melbourne. I’ve been surprised at how friendly stores are.

    Hope this helps!

    1. 1

      Wow, that's super good looking.

      If you don't mind me asking. Which models are you using? How are you running them and how are you feeding the initial image to it?

      I'd love to know more and if I can help out in any way, please reach out, seems like a really wholesome project.

      1. 1

        It’s really interesting that you mention initial image. I’m not using one! It’s all original creations “in the style of” great painters. I’ll probably do another book that does use famous init images because of the enthusiasm for that idea.

        I’m using replicate’s dreambooth api to generate 512x512 images and then one of their super cheap upscaler models (I think “esrgan”) to upscale for print. Using an init image with the model that comes out of their dreambooth api is straightforward - publish it somewhere (eg s3) then include the url in inference params.

        I started out using the open sourced code from photoshot.app and went from there… in hindsight I probably should have used an e-com starter, but ce la vie. I’m about to add snipcart and give the site more of a “shop” feel, mostly to move the high-friction “upload 15 photos” step to after payment. I’m happy to refund people if they change their mind (happy? Ok Willing ;) but think it’s better for everyone if that unavoidable friction comes after payment.

        For generating all the images, assembling nice pdfs for print and ebook with a couple of customised bits (names, date, dedication), and watermarking still images I’ve got some crufty TS to do most of it. There are 3 bits I still run manually:

        • start the job that generates about 100 portraits (easy to automate)
        • select the best portraits (hard to automate, but a great job for the customer to do)
        • submit pdf to print APIs (easy to automate)

        So mostly what I need to now is marketing. The site is not converting, although I’ve thrown a couple hundred bucks worth of ads at it. People have started to buy when I talk about it offline… even strangers. So online marketing is my squeakiest wheel at this point.

        Anyway, I’m rambling! Thanks for your interest, and for sure let’s talk further here or in DMs/call/wherever. I could use help from grownups in marketing, coding, design, printing, and writing… to name a few 😂

  8. 1

    howd you recruit devs?

  9. 1

    As others just said, i think that it should maintain the same style.... Thats the point.

    1. 1

      I've tried going with more explicit style now: "use ink" or "use pastel colors"

      Check out the recent drawings https://dailyaicomics.com/b9bd2e2e-e179-482b-9b6a-b68d9085fb10

  10. 1

    More reaction options would be great. Just like dev.to has Unicorn and so on. Scary, creepy, confusing, etc. They can even be generated by Stable Diffusion itself :D

    1. 1

      That should be easy to do, hmmm. Something to do tonight 😂

  11. 1

    Thank you making it open source. Must have been fun working on this project.

  12. 1

    Great job, Mihovil! To make it truly open source, did you already decide what license you want to give to it on GitHub?

    Is it MIT, BSD, Apache, or something?

    1. 1

      Didn't really think about it! What would you recommend if I don't really care?

      1. 1

        If you don't care, that's easy: Tell GitHub to use the MIT license for your project. It allows other people to do anything with your software, but it asks people to mention you as the author.

        Here are the necessary mouse clicks: https://docs.github.com/en/communities/setting-up-your-project-for-healthy-contributions/adding-a-license-to-a-repository

        1. 1

          Really cool, I didn't it would be that streamlined to add a LICENCE file.

          Done! Thanks :)

          https://github.com/infomiho/dailyaicomics/commit/7fa122d369979379eb2f6f2e34ee6653084d9a05

          1. 1

            Great, thanks for sharing your fantastic project, Mihovil! 😄

  13. 1

    Great idea. And it's great for you going OSS. For the monetization, you could look into NFTs - ordinals.com

  14. 1

    Nice! How do I access the generation UI? And curious to hear what API you're using for generation.

    Would love to see if you could connect the Evoke API to the image generation as well.

  15. 1

    Great job it looks good...

  16. 1

    Thanks Mihovil for sharing this cool project, it's also a solid example for what you can do with these tools. I just submitted it for you on topAI.tools here

  17. 1

    Great job! Must have been a lot of fun to develop it. Thank you for making it open source!

    1. 1

      It's super cool to see new comics each day. Then I get a little sad since they are not there yet and that gives an incentive to tinker some more.

  18. 1

    First great job 👏

    Second are you willing to continu improving the product (Illustration generation for example)

    IDEA : allow your visitors to create the next chapter of the story, by allowing them to prompt next chapter --> Vote for best next chapter --> then that will build (with some momentum ) nice audience --> collective generated comics --> potentially you can distribute the generated comics and share profit with the collective behind the generation

    Ok let me know if it doesn’t make sense

    1. 1

      Oooh, that's a very interesting idea!

      Community written story where we ask each day, "What should John do next?" if our main character is called like that 👏

      I'd love to make it playful and give people a reason to come and engage.

      1. 1

        ;)
        Also people can join a collective and generate a story together and compete with other collective

  19. 6

    This comment was deleted a year ago.

Trending on Indie Hackers
How I grew a side project to 100k Unique Visitors in 7 days with 0 audience 55 comments Competing with Product Hunt: a month later 33 comments Why do you hate marketing? 29 comments My Top 20 Free Tools That I Use Everyday as an Indie Hacker 18 comments $15k revenues in <4 months as a solopreneur 14 comments Use Your Product 13 comments