2025
AI, Three.js, ComfyUI
An AI space generator based on panoramas designed to provide an infinite exploration experience with just one-time prompt input, using Three.js and ComfyUI. By moving away from the goal-oriented nature of typical AI tools, this project shifts the focus of AIGC from providing answers to facilitating an infinite, imaginative journey.
As users, they can enter prompts in the first space—the void—to generate the initial panorama space, presented in the form of a door. After entering this door, they can take a screenshot to create a series of named panoramas, with each name derived from the captured image.
In the end, through a series of choices, they may arrive at a scene completely different from the original prompt, getting a different answer.
I named it AI Panorama Wander, as I believe that when we use AI, we often receive answers that align too closely with our expectations. When outcomes are always anticipated in advance, it can be boring. So my initial thought was: what if generative AI were no longer just a tool for fulfilling desires and responding to needs, but instead became a medium that disorients us, which lets users to wander, explore, and search for answers?
