Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

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

clevercoloringbook.com

Post reply on HN
Print view View original

161–170 of 203 posts

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

#161

Post by jfengel »

hightrix wrote: post #160
seeEllArr wrote: post #46

Earlier quoted context omitted.

The food you order online was not stolen from the server/bartender without their permission or compensation. Even if the analogy holds, this is whataboutism, and in the U.S. at least tipping is a fucked system too.

You’re right that the food itself wasn’t stolen, but how many restaurants actually come up with their own recipes? And how many use recipes created by master chefs that were ‘stolen’ and used by others? This is how art works and has always worked. Artist should be using the same AI tools that the general public use but create things that the general public cannot. That’s what artists have always done.

Recipes are the least of what goes into a restaurant. It's not a secret. In many restaurants the chef will give you the recipe if you ask nicely. If not, anyone skilled in the art could reproduce it.

Running a restaurant is a trillion other things. Ordering the right amount of ingredients. Hiring, training, and keeping staff. Cleaning the bathrooms. Replacing stolen silverware.

You're not paying for the secret recipe. There isn't one. You're paying for the insane amount of work that goes into putting cooked food on a plate.

Images are much more about the specific process that went into creation. The intellectual part that can be taken is a much higher fraction of the product.

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

#163

Post by jfengel »

richardw wrote: post #23
rafram wrote: post #13

For what it’s worth (and it’s probably not much), it doesn’t cost that much to commission comic book-style art from an actual artist online. When you do that, the proceeds go to an artist, not to an AI company that stole from them and a software developer who wrote a wrapper around their API.

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 o…

I am following a similar mental path. I feel like the AI companies should be paying some sort of tariff on their output, going towards everyone on the planet who contributed anything at all. I don't think you can account for it more finely than that.

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

#164

Post by rahimnathwani »

MaxLeiter wrote: post #38

For those interested in building something similar, I prompted a story book generator using v0 and Gemini’s image generation a few weeks ago: Demo: https://v0-story-maker.vercel.app/ The chat: https://v0.dev/chat/ai-story-book-creator-zw7TrmkN2Eb

I love the method you used to get image consistency: construct a conversation history with the model where it explicitly accepts each image and agrees to use it:

  User: Previous page prompt: [Previous Prompt 1]
  User: [Image 1]
  Assistant: I've received this image and will maintain visual consistency with it.
  User: Previous page prompt: [Previous Prompt 2]
  User: [Image 2]
  Assistant: I've received this image and will maintain visual consistency with it.
  User: Previous page prompt: [Previous Prompt 3]
  User: [Image 3]
  Assistant: I've received this image and will maintain visual consistency with it.
  User: Please maintain character appearance, art style, and color palette consistency with these previous illustrations when creating the next image.
  Assistant: I'll ensure the characters, art style, and colors remain consistent with the previous illustrations.
  User: Generate an illustration for a children's ABC story: [Current Prompt]. Make it colorful, child-friendly, and in a consistent style with any previous images. Include diverse characters with different ethnicities, genders, and abilities. Ensure representation is natural and authentic.
I'm curious: did you settle on this after trial and error? I mean, did having the assistant explicitly agree increase consistency, compared with just asking nicely at the end?

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

#165

Post by thesparks »

bix6 wrote: post #147
ronsor wrote: post #43

Earlier quoted context omitted.

They're dismissive because we've had the same moral panics before with the introduction of photography, then sound recordings, and then digital art tools, and then vector art, and then 3D, and also the Internet to an extent, and... You can see where this is going, right? In the end, humanity and even artists will be fine overall, even if the world changes.

This feels different to me. This isn’t the camera going from film to digital. This is the camera taking over the photo creation process, and developing them, and selling them. What’s the point of the human?

It feels different because you are living through the change. What happened before is something you read about and already know the outcome. Future generations will say the same.

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

#166

Post by MaxLeiter »

rahimnathwani wrote: post #164
MaxLeiter wrote: post #38

For those interested in building something similar, I prompted a story book generator using v0 and Gemini’s image generation a few weeks ago: Demo: https://v0-story-maker.vercel.app/ The chat: https://v0.dev/chat/ai-story-book-creator-zw7TrmkN2Eb

I love the method you used to get image consistency: construct a conversation history with the model where it explicitly accepts each image and agrees to use it: User: Previous page prompt: [Previous Prompt 1] User: [Image 1] Assistant: I've received this image and will maintain visual consistency with it. User: Previous page prompt: [Previous Prompt 2] User: [Image 2] Assistant: I've received this image and will mai…

I didn’t write that - v0 did. You can see it all in the chat

I noticed it wasn’t passing the image back and forth so I asked it to, and it wrote that prompting

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

#167

Post by mmcwilliams »

xdfgh1112 wrote: post #57
mmastrac wrote: post #11

Earlier quoted context omitted.

Hard disagree. Sample #2 is totally the Miyazaki-vibe that is everywhere in OpenAI-generated comics. https://clevercoloringbook.com/samples/2_cartoon.png The cartoon owl at the top has a different vibe and would probably work for the comics as well.

It's nothing like Ghibli, you are overthinking this.

You think that the model is not properly producing the style it was prompted for in the below prompt? I don't see other artists or styles mentioned.

> 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.

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

#168

Post by HideousKojima »

0_____0 wrote: post #97
ronsor wrote: post #4

Earlier quoted context omitted.

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.

OP confirmed that their prompt includes a directive for Ghiblification. Given that Miyazaki is known to hate GenAI I really can't condone... I mean there's nothing anyone can do about it but it's just kind of sad.

>Given that Miyazaki is known to hate GenAI I really can't condone...

Miyazaki also said (in response to the Charlie Hebdo murders): "For me, I think it's a mistake to make caricatures of what different cultures worship. It's a good idea to stop doing that." I still love the man and his work, but he's not some infallible authority on what is and isn't appropriate in art.

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

#169

Post by op00to »

jmathai wrote: post #21
op00to wrote: post #19

Earlier quoted context omitted.

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?

Yes. But they are people, perhaps with families to feed. Not computers. Cool idea. I can see keeping colored pages of these by my kids up on the fridge a lot longer than what’s on there now!

So it’s ok to steal as long as you are feeding a family with it? :) I get what you’re saying, anyway and it’s an important distinction.

I guess what you see as “stealing” I see as inspiration. I also believe that there will be artists who use these new artistic image generation models in ways that are new and interesting just like the first person who used spray paint for graffiti was ripped off by everyone else.

It delighted my kids to see themselves depicted in coloring sheets in situation where they are currently interested. There is no world where I would have paid an artist to make these photos, and we would have just colored on blank paper.

Again, I get that real people’s content was devoured by these big companies, but at the same time I am much more concerned about bigger issues and would rather focus on getting ahead of AI rather than fighting it.

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

#170

Post by rahimnathwani »

MaxLeiter wrote: post #166
rahimnathwani wrote: post #164

Earlier quoted context omitted.

I love the method you used to get image consistency: construct a conversation history with the model where it explicitly accepts each image and agrees to use it: User: Previous page prompt: [Previous Prompt 1] User: [Image 1] Assistant: I've received this image and will maintain visual consistency with it. User: Previous page prompt: [Previous Prompt 2] User: [Image 2] Assistant: I've received this image and will mai…

I didn’t write that - v0 did. You can see it all in the chat I noticed it wasn’t passing the image back and forth so I asked it to, and it wrote that prompting

BTW - thanks for sharing the chat. I'm going to study it for prompting tips. It's good to learn from other people's iterations using coding agents, and not just my own.
Post reply on HN