Ren'Py [Prototype] A Visual Novel where YOU define the characters, and the game generates the art in real-time (Ren'Py + Local ComfyUI)

[Pukerud]

The Concept
I am working on a Visual Novel framework that integrates Ren'Py with ComfyUI.
Unlike standard AI games where the developer generates 5,000 images and ships them, this project generates the assets locally on your machine while you play.

This means the game is not limited to what I decided the characters look like. You decide.



What works in the prototype right now:
I have a working build using the Qwen-Image-Edit model.

• Full Character Customization: Before the story starts, you configure the MC (e.g., Nerd/Fat/Skinny, Shaved/Hairy) and the LI/Aunt (Voluptuous/Athletic/Fat, etc.).
• Style Selection: You can choose the art style for the whole playthrough (Anime, Comic, Realistic, or 3D).
• Real-Time Generation: The game calls your local ComfyUI install to generate the sprites and backgrounds based on your settings.
• Fast Iteration: Using Lightning LoRAs, generation takes about 10-15s per image on a 4090, 20-30s on a 4070.

Why I’m posting this (The Video)
Since the project relies on a local ComfyUI backend, I can't just upload a simple .exe for everyone to test yet. I wanted to show you a video of the prototype to see if this mechanics interests you.

I have already been testing a lot when working on this, that's why there are already many images and sprites to choose from.
Small Video Demo:

You must be registered to see the links

(Took off here pants in one shot,showing breasts in another. So at least two spicy shot)

Current Barriers vs. Future Plans
Right now, the prototype requires you to have ComfyUI installed with specific models (Qwen, specific LoRAs, etc.).

• My Goal: To eventually create a "Portable ComfyUI" bundle where the game handles the backend automatically (One-click install).
• The Question: Is this worth the effort?

Feedback Needed
I am at a crossroads. I can continue refining this into a standalone game, but it will always require the player to have a decent GPU (8GB).

1. Would you play a game where you have to wait ~20 seconds per image in a scene to generate, if it meant you had total control over how the girls look?
2. Is the setup (running a local LLM/ComfyUI) a dealbreaker for you, or do you already use local AI tools?
3. Dialog and prompts are just for testing so do not get stuck on the story, it will be longer and have more scenes.

This is a passion project, but I want to know if I'm wasting my time trying to make this user-friendly, or if I should just stick to pre-rendering images like everyone else.

Honest/Brutal feedback is welcome.

 

[john1230]

Cool concept but you need a resource heavy computer to run the game, and I don't think most people would have the hardware requirements to play the game.

A 8GB graphics card would mean a minimum requirement of a Nvidia GTX1070, that's a lower high end graphics card.

Most people I would say plays these games on a potato laptop that don't have a dedicated GPU in it, so that would shrink you game's exposure.

I would say that if you want to offer the user a choice of character creation. I would suggest you take a look at other games like Summer Time Saga. The DEV draws character parts (head, eyes, mouth, torso, arms, leg etc... ) separately and stiches them together to show different poses on screen.

You can do something similar by pre-rendering the body parts and let the user choose what parts to use to make their character and then stitching them together in the game code to use in your game.

This approach would cut down the hardware requirement and open your game up to more people but it's more time consuming to code the scenes in the game.

So there's pro's and con's for you to weigh up if you go down this route.

Anyway it's another option for you to consider.

I wish you all the best with your game, bye now.

 

[Pukerud]

Cool concept but you need a resource heavy computer to run the game, and I don't think most people would have the hardware requirements to play the game.

A 8GB graphics card would mean a minimum requirement of a Nvidia GTX1070, that's a lower high end graphics card.

Most people I would say plays these games on a potato laptop that don't have a dedicated GPU in it, so that would shrink you game's exposure.

I would say that if you want to offer the user a choice of character creation. I would suggest you take a look at other games like Summer Time Saga. The DEV draws character parts (head, eyes, mouth, torso, arms, leg etc... ) separately and stiches them together to show different poses on screen.

You can do something similar by pre-rendering the body parts and let the user choose what parts to use to make their character and then stitching them together in the game code to use in your game.

This approach would cut down the hardware requirement and open your game up to more people but it's more time consuming to code the scenes in the game.

So there's pro's and con's for you to weigh up if you go down this route.

Anyway it's another option for you to consider.

I wish you all the best with your game, bye now.

Thanks for the input. I should have a a look at how its done in summer time saga.

 

[Joker5]

I'd like to hear from anyone testing the game about how well it runs on their PC. Running both the game and Comfy UI simultaneously is a headache, and as john1230 said, it requires a higher-spec PC than most people have.