butterchurn is a browser-based, real-time music visualizer that faithfully reimplements the classic MilkDrop visualizer using WebGL2. It consumes audio from the Web Audio API, runs FFT analysis to derive frequency/amplitude data, and drives a shader-based render pipeline from JSON presets that encode MilkDrop-style equations. The core library exposes a minimal `createVisualizer()` entry point, after which consumers load a preset and call `render()` with audio level data each animation frame. The library supports two equation execution paths: a JS-based expression runner and a WebAssembly/AssemblyScript-based EEL runner, with a custom Rollup plugin that compiles AssemblyScript to base64-encoded WASM at build time. It also includes deterministic seeded RNG for reproducible rendering, which is used by its visual regression test suite to compare JS and WASM paths pixel-for-pixel across a curated set of presets.
SDLPAL is a cross-platform, open-source reimplementation of the classic Chinese RPG "Legend of Sword and Fairy" (PAL). The engine loads original game data (provided separately by the user) and reproduces the full game experience including battle systems, map exploration, menu interactions, script-driven events, and cutscenes. It is built as a monolithic C/C++ application using SDL for input, video, and audio, with optional OpenGL rendering and GLSL shader support. The codebase includes a comprehensive platform abstraction layer supporting Windows (desktop and UWP), Linux, macOS, Android, iOS, 3DS, Wii, Emscripten (web), and libretro. Audio is handled through a modular player interface that can use native MIDI, software synthesis (TiMidity, OPL emulation via MAME/DOSBox/Nuked cores), or compressed codecs (MP3 via libmad, Ogg Vorbis, Opus). The project is actively maintained, with recent commits and multiple CI workflows. While the core game logic is tightly coupled to PAL's data formats, many individual subsystems (configuration, audio players, OPL emulation wrappers, binary I/O, resamplers, font loaders, and decompression routines) are cleanly abstracted and could be reused in other projects with minimal adaptation.