Particle Simulator by Casberry

Particle Simulator by Casberry is a coding tool. A browser-based WebGL simulator for 3D particle swarms of 20,000+ points, where AI-written JavaScript defines motion and color. It adds hand-gesture navigation and exports to JS, React, Three.js, wallpapers, PLY, GLB and OBJ.

Particle Simulator by Casberry is a WebGL playground for building large 3D particle swarms, with the movement and color of each particle defined by a short JavaScript function. The page is built around swarms of 20,000 or more points, and it ships with presets such as a geodesic sphere whose radius can be adjusted. The target audience is creative coders, computational artists and developers who want generative visuals without setting up a Three.js project from scratch. The workflow centers on AI-generated logic. The site provides a ready-made prompt that can be pasted into a chat model (Gemini and Claude are the recommended ones) together with a creative idea, such as a tesseract breathing in four dimensions. The model returns only a function body that writes a position and a color for each particle by index and time. The prompt lays out the available variables, helper calls for sliders, an info panel and floating 3D labels, plus strict rules on performance, finite values and forbidden patterns. A validator checks pasted code and runs a dry run, so unsafe or unstable output is rejected before it renders. Interaction goes beyond the mouse. A gesture mode called Neural Navigation uses hand tracking: one finger to look around, an open palm to zoom, and a peace sign to change speed. Finished work can be exported in several forms: vanilla JavaScript, React, Three.js, wallpapers, and PLY, GLB or OBJ 3D files. PLY can be written as a light point cloud or a detailed solid mesh, and image output offers several aspect ratios, custom dimensions, resolutions up to 4K and PNG, JPEG or WEBP formats. A drawing pad tool is also present. Among alternatives, it sits between general creative-coding environments and shader editors. It trades the depth of a full engine for a narrow, prompt-driven loop focused on particles. Users who want to write the math themselves can do so, while those who prefer natural-language ideas can lean on an LLM. The homepage does not describe pricing, accounts or saved-project limits, though it mentions saving and publishing in its description.