Item — HN Reader

I have a VN that I have wanted to complete for years. Having built and deployed using RenPY, the problem is something I wonder if Rosebud can support:

1) I'm a programmer not an artist. If I give a description of a character can the AI create a 2D version of it that will be consistent?

2) I'm a programmer not an artist, if I give a description of a scene/background can the AI create a 2D version of it that will have proper perspective for the character, and that will look similar later in the game when they return there?

3) I'd like to add flexibility to routes. Can I start a route out and then let the AI and the player interact to take it beyond the starting point?

2 comments

At the moment this process is probably more manual than we'd like. You can definitely make characters and backgrounds.

We're using our own custom AI Agents to generate the games and what we'll do next is expose them for the games to leverage too. At that point you should be able to have the game self-generate based on rules.

So it sounds like I should "keep an eye" on this space and hopefully at some time in the future give this a try. Do you have any VN examples currently just to see where it's at?

We have one game in progress by one of our beta testers that has some VN vibes/elements: https://play.rosebud.ai/games/91d54c96-378d-4c64-b71b-4841fc...

Here you chat with this grandma, make here drink coffee inside and go outside (she'll comment on both of the actions a bit haha).

Would Rosebud help in making consistent characters?

I tested out that example game. I asked her "do you like coffee", and Grandma replied "I despite coffee...". However, when you say "lets drink coffee" she says "Oh, absolutely! Let's drink coffee, my dear user. There's nothing I'd love more ..."

I don't use ChatGPT much myself, but I can see its potential if you can have consistent responses with "personalities"/"characters" for games.

We definitely strive to make super consistent characters possible on Rosebud. I might be wrong but I think this user in particular set this grandma to be a bit quirky like that.

A lot depends on how detailed the user's description for the character is. But eventually there might be some issues with forgetting things deep into the chat as a limit to current LLM possibilities. We want to try and solve this by introducing saves, but still experimenting!

I could see something that bootstraps a renpy project with art + applying the correct framework as saving someone days or weeks of time, overcoming the barrier to get started.

For example, there are different patterns if you want a simulation that has time as a factor, versus a classic choice-based VN.

A lot of people building games struggle with where to start for their specific idea; it's a lot of research and if you make some wrong assumptions, you may have a really hard time executing on your game. Also, a lot of creative people/tinkerers aren't familiar with software design patterns generally.

Would this export the code so someone could edit it manually after the initial creation?

We have a bunch of trending projects that you can already clone from (see https://play.rosebud.ai/home "trending"), and this was in closed beta, so once we open the app to more users there will be a lot of projects to give new creators projects to start from immediately.

You can already kind of export the code because we show most of the game code. However what we're aiming to do is to make it easier to build in platform without worrying about deployment. Also we are aiming to bring an audience to play your games immediately via the Rosebud platform so you don't have to share elsewhere.

I think a lot of indie developers have $0-50 (total; the same as RPGMaker license cost) willingness to pay to start working on a game, but would LOVE help monetizing that game. There's a huge dance on itch.io where a lot of niche games are alpha-quality passion-projects, so they start off for free (demo), and then the creators eventually start asking for money to keep working on the game, typically using Patreon.

Even though there's other creator platforms that charge less % than Patreon, like ko-fi etc., and itch.io has built-in payments, patreon still seems to be the prime destination. I think it's because a lot of the patrons are already on patreon so creators hope it's easy for them to add new creators to their list? Also because game progress can take a while, and something like patreon lets them give devlog updates and preview images and run audience polls to keep patrons engaged in the process.

Anyway, there's a oauth integration between patreon and itch.io so that creators can only distribute to patrons, as an alternative for people paying for a point-in-time game version. I think that is a newer feature.

I'm not sure if itch.io exposes functionality such that you can publish these rosebud games there, but they do already allow you to have hosted in-browser games like with renPY and Twine.

Understood your strategy may be to become the destination for these things, but you may be able to get early distribution faster by making it easy to push these games to existing platforms that have network effects around discoverability already. Though I suppose someone could just create an itch.io page with a link to your hosted game??

To help you get started with visual novel projects, there are already many templates containing AI characters that you can leverage. such as https://play.rosebud.ai/games/52417bb7-6796-4a4b-8a38-38d99e... You can clone these and customize the personalities to suit your storytelling needs. Using Chat, it is straightforward for even creative people without coding experience to add new scenes, interactivity, and dialogue. Simply clone an existing template, edit the characters to your preferences, then build out additional narrative sequences or branching event options through the intuitive chat interface. This makes the creative process smooth and coding-free. Experimenting on top of premade foundations can kickstart your visual novel effectively.

What I didn't see in this (and the couple others I've seen mentioned) is talking with more than one character in a scene, or having multiple scenes with the characters in different positions/poses/clothing.

My particular story has a number of folks with various handicaps (magical realism with a COA vibe), and trying to get an AI to do a reasonable character in a wheelchair is hard but doable, while amputees or folks with hearing aids is nigh on impossible.

To be fair that's probably more a "me" problem than a "you" problem...

Hey! It is potentially possible to prompt the character LLM directly to simulate multiple characters, but having streamlined support is definitely on our Radar! We are also working on a template where you'll be able to to talk to multiple NPCs and also have them talk to each other.

Research for consistent character in different poses is also in the works :)