Pop Audio Player is a lightweight ES6 + Three.js (r182) audio player with a GPU-driven visualizer system.
The visualizer is built as a set of GLSL “modes” (fragment shaders) that react to audio analysis (FFT spectrum, peaks, and waveform) and also supports an idle animation path when audio is paused or silent, so the scene doesn’t look dead. Some modes optionally use a ping‑pong feedback/FBO pass to create trails/echo effects.
I am curious how long does it take for you to fully implement a plugin like this? Based on the rate you create these new plugins, your code output is massive, congrats!
f it’s a completely new plugin, it usually takes 2–6 weeks. For this player, it took about 5 days—the JavaScript version—because I already had a base from a different player.
Either way, with AI, things are fast now. For example, documentation used to take me 3–4 days, and I hated it. Now it takes about 2 minutes—I just prompt it, and AI replaces everything. Crazy!
Code-wise, AI helps a lot because I don’t have to search for things the old way anymore. When something doesn’t work, I can just prompt the issue and, most of the time, it gets solved.
I’ve found it very useful for isolated tasks, but giving it full control is a disaster. It once deleted most of my code, and I had to start from scratch. Also, every time you ask the same thing, you get a different version.
I’m as confused about how AI works as AI itself seems to be. It often feels like a gamble, but at this point, it’s clear that we can’t go back.
If I had done this plugin from scratch, I think it would have taken more time because the logic and context are things I have to do; AI can’t do that, and even with AGI, the issue remains: there is just too much context, and it’s impossible to prompt such a thing!
