Show HN: I used OpenAI's new image API for a personalized coloring book service

Thanks! I've added a section at the bottom of the site showing some real photos of an actual coloring book I got in the mail. There are thumbnails of all photos uploaded on the back.

$24 + postage is the lowest I could reasonably charge for this. Printing costs are a bit more than half of that, OpenAI charge a surprising amount for image generation, but there is also a good amount of human effort (and creative choices) in generating the book. It's not a fully automated process and I hope that's evident from the quality of the end product.

It’s not cheap, but my kids treasure coloring books for a long time and probably one like this until it falls apart.

To the author, I have this idea, for each page, put a sheet of transparent plastic or something like that. So the owner will color the plastic which can be erased. But it may increase the cost anc the color may not stick to the plastic.

I suggest that they use the Disney style. It’s similarly popular and well known but no one will stand up to protect it since it’s a multi billion dollar company that we already know has no respect for artists or artistic legacy.

Great point. As per our TOS - users of the site must be over 18 and have the consent of everyone in the image (i.e. their own kids, relations etc).

I put that line about OpenAI's usage policy there for practical reasons. If someone orders something that OpenAI refuses to generate (like a photo of Bart Simpson say), then I can't include it in the printed book. With this project, if someone uploads content that's in any way inappropriate, we'll see it and refuse to fulfill the order (and take other appropriate actions, if needed)

I'm not the OP, but during the recent Studio Ghiblification craze there were a huge number of photos of families and kids passing along in facebook, twitter, and other social media. It was literally everywhere you looked. OpenAI obviously saw all of that. I don't think they actually care unless it's something bordering on illegal.

> "that contain real people"

It seems the loophole on this site, is the examples (by my best guess) are AI.

It doesn't look particularly Miyazaki style to me; it's just a generic cartoon style.

I think the Ghiblipocalypse has gotten people on edge.

This has zero resemblance to Miyazaki’s style. (And I say that as someone who isn’t a fan of this idea at all.)

Close enough! The prompt I use is:

> Make this a page in a colouring book. The drawing is in a simple Studio Ghibli portrait style. Bleed all the way to the edges. Background colour is #ffffff and lines are bold and #000000. There is no shading or crossthatching.

Anyone else feeling this weird vibe in the age of AI? You get a cool new idea, but then you think about how easy it is for someone else to replicate it, and you end up not doing it.

The OP looks like it runs images through a Ghibli filter first

Yeah I've been doing this with image-gen AIs pretty much since they started and it's a lot of fun. Even early Dall-E etc was awesome at doing stuff like "Create a colouring in sheet with some dinosaurs having a party" or generic prompts like that, and more recently giving photos to convert has been loads of fun for the kids.

I’ve done this a bunch with my son. It’s not quite that simple because often times it’ll create images that have too much detail, sometimes it’ll actually include colors. But yeah, it’s not really all that complicated

i've been doing this with my child and picture books. i take pictures of pages, convert it to a coloring book, print it, and then we color her favorite books together.

Thanks so much! Really glad you like it.

You're uploading your family pictures.. nevermind. Go enjoy your coloring book.

In fairness no artists are advertising a personal coloring book. The time, effort and cost would put this out of reach for 99.99 of people.

No artists are losing income because of this and no industry is being upended. This is a new product that's available because of a technology advanced.

Why the focus the artist? Everytime you order in food online you take away a tip from a host, server, bartender and take away a job from a person who answers a phone. Why focus on artists when so many have been affected by technology.

Usually when you commission something you're asking the artist to do art and create something unique with their own artistic flair... not just line-trace an existing photo.

The intention and cost of something like that is not at all comparable to what is being offered here.

I tried to do exactly that once. I was offering between $20-$40 per image to make a few coloring pages as a mother's day gift for my wife. Not complex images either -- just basic coloring pages from photos of my wife and child, without backgrounds, for my kids to color in.

I reached out to multiple artists, and got one image back (from a good friend). I gave up on commissioning actual artists, and traced the images myself on a tablet. I imagine someone with the right knowledge of where to find artists and the willingness to wait on their schedule could have done it faster, but I'd have used this service if it had been around.

I can't imagine how much it would cost to commission an artist to do a whole coloring book and then organize them and send them to print but it's a good point. AI is never going to be as good as a real commissioned artist, but this idea makes having something similar far more accessible to a lot of people.

My opinion isn’t fully formed but I currently think either all content producers have a claim (potentially workable as eg a discount), or only those who contribute should get access to AI’s.

And by all I mean the AI companies owe a huge debt to all humans who wrote or designed or drew anything. The vast majority of the benefit of this technology relies on volume: the billions of pages and lines of code we wrote for other humans, but have now been repurposed. This technology relies on bulk, which was mainly unprofessional or freely given content, by those who intended it for other humans. It was not 100% built only on the output of the few who charge for their exquisite words or designs, even if their output is higher quality.

Alternatively, let the AI companies go for it but everyone who uses any kind of AI should understand that they’re standing on the shoulders of the millions of developers and nonprofessional writers whose work has now been repurposed. Not the few artists and journalists. So those artists and journalists should both refuse to contribute to, and use, AI.

* I’ve written very little of this useful content, but would be happy to pay my share to those that have built what we have. I also turn off training on my content, but I pay a lot for models. Feel free to help me think through this with comments of your own.

If it does not cost that much, that is obviously because the artist is too cheap. If you find that to be a preferable equilibrium, that's a choice I guess, but I find it fairly ironic in light of the purported motivation.

I can get ChatGPT to do this for literally free. Even in the free tier, I can get a couple images per day.

This is a cool technological feat but what is the cost to humanity and its artists?

Some of these replies seem rather dismissive to the artists’ plight.

If this person’s service was to pay human artists $24 for a 23 page custom coloring book you’d be crying on here about them not paying human artists enough.

Almost nobody is paying $100 or more for a custom 5-page coloring book.

This service isn’t taking work from human artists.

Those transactions never would have happened, and never will happen.

Maybe, but then I have to negotiate with the artist, handle their refusal to draw art of my choosing, and wait for their (possibly unpredictable) schedule. AIs mostly avoid these problems.

[dead]

Didn’t the artist “steal” from artists that came before them by looking at and taking inspiration from their photos? Especially ones that would do such artistic genres as commercial coloring book art?

Thank you!

OpenAI costs are surprisingly expensive. It's about $7 to generate a whole book (24 pages). There are 8-24 images allowed in a book, with a cover too. So there'll be 48 max pages in a book (incl blank pages).

Where do you get that from?

I don't know much about it, feel free to send an email to dara@clevercoloringbook.com

Why would you do that to Sapiens, it's an enjoyable read

Thanks!

I'll implement this today.

Edit: I've implemented this! It's a lot of code changes so I hope I didn't break anything.

Thanks! Hasn't been a problem yet, but if that happens I'd just email the customer directly and ask for a replacement photo/if it's okay to leave that one out.

Oh, lol. Nice catch - I'll fix that.

I was thinking of doing different sizes alright! They might not be too much cheaper, that's the only problem I can think of there.

Thank you, I agree and I'll update the example photos.

I didn't add a PDF but I added some photos of the real end product to the bottom of the landing page now.

Can you show me an example? I think before 2-3 days ago this wasn't really possible without the output images looking really bad.

Yes! https://openai.com/index/image-generation-api/

I was very excited when it came out. Google have Imagen 3 (is that the same as Gemini Flash?), but you need special access to be able to edit images. I haven't tested it yet but I think it's a lot cheaper than OpenAI

Gemini has released new models, maybe they are better.

Also, I request you to expand further, why Gemini is not better?

Presumably you aren’t the parent of a 5-7 year-old child. I might try this manually and save some money but my kids will absolutely cherish coloring themselves, their friends, and their parents. We’re on vacation now and this is gonna be big when we get back.

Thanks! I priced it as low as I could given the costs of printing the book, Sora's API costs, and the human effort that goes into it (there are some creative choices to be made too!). The 100 page books you see on Amazon aren't personalized and probably not the best quality. I'm also hoping a completed page from one of these books will be a nice keepsake for the parents as well as being more of an incentive for a child to exercise some creativity.

Ask GPT to give you a prompt that would create a color book :-)

I always use LLMs for meta-prompting. They know themselves better than others :>

Thank you!

The printing aspect of this wasn't too easy... I wish I could charge less but as it stands (especially with surprising API costs) I'm barely making a profit on this.

Wow, yes you're right. I did in fact take that image from the Sora homepage because I thought it was cute.

I pretty much just assumed they're all in the public domain. I can't find the image anymore so I've decided to remove it for now. I generated the other three myself.

Thank you! I was happy with the discount thing alright but it's not useful for tracking referrals unfortunately.

Thanks! There is a good bit of manual effort involved on our side - we generate the images, regen any that don't look great, choose the best cover, and then send the PDF to Lulu. It's dropshipped, we never see the physical book.

Yes I use the Lulu API. It's great!

No, it's not a real-time process. We have an admin panel to generate the images, cover and PDF. There is a good bit of human input involved.

I mean it's pretty close. The handmade element comes from coloring it in.

There is a good bit of manual effort involved on our side before we send the book to print - we generate the images, regen any that don't look great, choose the best cover, and then send the PDF to Lulu. Yes it's dropshipped, we never see the physical book.

Canny filter: https://i.imgur.com/Ju7Ggac.png

Sora: https://clevercoloringbook.com/samples/3_cartoon.png

I've had this idea years ago and searched extensively for a way to turn images into nice line art, but it turns out there needs to be a good bit of creativity (AI) to do so. Old school computer science techniques don't cut it.

Do you not think the AI output looks far more polished and print-ready? Canny edges have a lot of noise and don't look at all clean for coloring book purposes.

Oh nice catch! Fixed, thanks.

No, but perhaps I could add this as a paid option.

Thank you!

You can simply open up Chatgpt and generate the image yourself, faster than it’d take to transact with this third party. The cool thing is that they are printing a physical book for you.

Presumably, it’s more profitable to sell a physical book, which makes sense.

wow!!!!

Any idea how much it costs to create a book?

The quality of the images is incredible! Great job

How much does the api cost to run? Do you have any safe guards in place in case bots try to build 1000 stories?