Vision 1: WYSIWYG Comic Site

Eye candy editor into long-tail pipeline; It sucks you in, makes you have fun, and then makes you money

A Bluegem member brought up this idea. We’re both fond of the concept of computer visualization of stories, and he’s exploring how to integrate new ML (Machine Learning) tools into his artistic pipelines.

A short story or comic book which explores using AI, letting you use AI art for the scenes exploring it as a cheat, and then you can also compare the hand-drawn art to the AI art.

And after some discussion, we iterated on this, and this is our vision for this project.

First step instead of building the project, we’re going to put the idea out there and see if people are actually interested in such a tool. If we validate the market before we build, we can manage our development resources and time better. I have a problem with rushing into things I want to develop; only half make them, and no one benefits. So going to try and spec this out here, what our idea is, and maybe you will find it helpful.

The fundamental core of the idea is that this would be a web-based comic book builder. You can be a writer or an artist, and there would be two modes at the top of the tool’s page to select which way you want to work.

We would initially build the artist version of this tool, allowing artists to draw in boxes and add frames to their layout. The borders between each comic frame will be adjustable, so you can change the height or width, and the image will readjust its cropping or zoom in to the new size.

The artist would draw their panels one by one. Be able to save their work in progress to the cloud as it will take quite some time to flush a project out. When finished, we would have an option for them to download a PDF, view it in a comic viewer online and/or sell it in our marketplace. Our marketplace could allow physical and digital versions with or without an NFT certificate of authenticity (and a fixed limit on how many they want to be issued). They, of course, can set a price of nothing in which the platform gets free marketing instead of a cut of the sale. Credit cards and cryptocurrency can both be exchanged without too many weird regulations.

This is all pre-2010s-level technology so far. We might need to build the tool to a specific level to compete with existing art tools. I think many of the samples could be stick-figure comics to emphasize that creators don’t have to compete on a professional level but have something interesting and funny to say with their work. The marketplace could support crowdfunding projects and auctions. Also, targeting low-skill artists will help attract writers who will become the next significant phase of the project.

The writer selection would still have comic frames and options to add rows and columns, just like the artist tool. However, instead of drawing tools to fill in the frame, they would get a multiline textbox to write dialog, narration, or a scene description. At the top, they would select an ink-style, a color-style, an art-style, and an era-styling from a set of preprogrammed options. Eventually, we may allow more of a freeform text box, but we think that:

1. People are not good at generating prompts for ML yet, and it becomes an art form all in itself

1. People do not need to get good at generating prompts. They need to be given a few options to choose from.

1. There are additional engineering benefits from having a fixed set of options. The potential for caching and accelerating common requests will benefit from this.

So the main idea would be like a GitHub copilot tool, that we use ML to be the artist for the writer (optionally, still can have a traditional artist collaborate with you too). And the killer demo or the toy on the land page would be real-time image generation as you type. Instant-immediate feedback.

If you play with good ML, image generation takes roughly 2 minutes per picture. So we reached out to our ML masterminds, who showed us how to speed it up by lowering specific quality settings and believe we can make this real-time without too much hardware. So the idea would be you instantly get a low-quality image (maybe from the cache), but as the text stays static, we would try to upgrade the high quality. Of course, writers might need some locking mechanism, so they don’t lose something they already love in that spot, but that will be something to experiment with when we get to it.

This would provide the writer with a digital collaborator if they didn’t already have one. We can then go back to the artist side and give them an LLM (Large Language Model) ML assistant option too.

I’ve said a little bit about the phases. Let’s talk about why there’s some order to this project. For a new commercial project, the best path is the quickest one to revenue. So a minimum viable product has to strip out everything that’s not needed for the core of the product and needs to immediately represent the revenue model. Since the revenue is likely to come from the marketplace, we could start as an online comic store, but since what makes this project unique and distinct is the copilot option, we decided to focus on the creation tool first.

We will build a simple tool and marketplace and then iterate on it (in conjunction with marketing/sales feedback). The point we add the first ML copilot is where we start being special and unique. Doing the ML at a small scale will be cost-effective, but as the number of concurrent users actively using the ML tool grows, we may need to tweak our revenue model to support the costs.

NFTs and ML have a terrible reputation in the public square right now. NFTs are a valuable tool for artists to create value even though they have a bad reputation, so they will ultimately help the project and encourage the use of cryptocurrency wallets and increase the overall importance of blockchains. ML has an image of having stolen talents and jobs from humans, and while we understand those feelings, we want to make sure we use the ML tools in a way to enable more creations of works of art. So the NFT option would not be put in our marketing materials, and our ML models would be ethically sourced, so we would show we’re respectful of the temperature of the zeitgeist at the moment.

While this is just taking new technologies and applying them to older existing projects, we’ve injected our values and experiences to drive it into a cohesive project with a decent chance of success. We will continue to think and iterate on this idea and may even build some of it. We’re putting this unfinished idea out there like an open-source vision. You are free to attempt to make it on your own. We think the world would be a better place to have this, and if you’re the better person/team to build it, that’s great. Then, we can focus on the next great thing.