threepp
C++20 기반 크로스플랫폼 3D 라이브러리, 파이썬 바인딩 및 three.js 수준 API 제공
아직 감지된 업데이트가 없습니다. GitHub 릴리스가 나오면 여기에 보입니다.
개인 계정의 공개 저장소 소유자는 GitHub로 로그인해 운영자를 확인하고 소개를 직접 관리할 수 있습니다. 관리자 승인은 필요하지 않습니다.
로그인한 뒤 확인 파일을 만들어 이 저장소의 기본 브랜치에 30분 안에 커밋해 주세요. 다른 브랜치나 포크의 파일은 인정하지 않습니다. 파일을 올린 뒤 아래에서 확인하면 관리자 승인 없이 관리할 수 있습니다. 자격을 유지하려면 파일을 남겨 두세요.
로그인이 필요하면 로그인 화면으로 이동합니다. 로그인 후 이곳으로 돌아와 원하는 확인 버튼을 다시 눌러 주세요.
프로젝트 폴더에서 /nomorevibe verify를 실행하면 소유권을 확인하고 소개를 직접 관리할 수 있습니다.
1분 만에 확인하기메이커가 쓴 소개는 위의 한 줄이 전부입니다. 더 자세한 내용은 저장소 README에서 읽을 수 있습니다. 저장소 README ↗
README 전체를 불러오는 중입니다.
README 전문 보기 ↗The online Faust IDE can be used to edit, compile and run Faust code from any recent Web Browser with webassembly support. It works completely on the client side and it is therefore very convenient for situations with many simultaneous users (workshops, classrooms, etc.). It embeds the latest version of the Faust compiler with an efficient webassembly backend and offers polyphonic MIDI support.
The editor engine is based on Monaco Editor. It provides syntax highlighting, auto completion, code hinting and direct access to the online documentation. The documentation command (Ctrl-D) uses the function name at the cursor position to locate the relevant information.
You can double-click knobs or sliders to reset them, or on the group label to reset all its children.
Enable Save Params State to preserve DSP control values when running or recompiling the program. The restored values are applied to the new DSP and reflected in both the embedded Faust UI and an already-open Faust UI popup, instead of those controls returning to their declared defaults.
The Monaco Editor supports an optional mode for Vim users. To enable it, open the command pane (F1) and search for 'Toggle Vim Mode'. The same command is used to desactivate this functionality. Alternatively, you can right click on the editor and press 'Open Command Palette' to open the contextual menu.
Several DSP files can be added in the top-left Project Files section and edited independently. Any non standard library, like a foo.lib file, can simply be added by drag/drop, then used in the DSP code with import("foo.lib");.
DSP files or libraries can also be loaded or saved with the Upload and Save As buttons in the left column.
The project file list is stored in the browser, so files remain available after a reload. Drag/drop supports several files at once, both on the editor overlay and on the Project Files panel. Structural imports are saved immediately before the import finishes, so a reload right after a multi-file drop keeps the imported files.
On Chrome and other Chromium-based browsers, Faust IDE can also use the browser File System Access API to work with local folders:
.dsp and .lib) opened or dropped from a mounted folder are edited in place: after the project copy is saved, changes are also written back to the original disk file.untitledN file.foo.dsp already exists and you later open/drop a mounted disk file with the same name, Faust IDE keeps the local copy untouched and reports a conflict. Rename or delete the local copy before opening the mounted disk file if you want the disk-backed version.While the option is turned on, the diagram or the DSP UI will automatically be updated from the code. The editor will also try to tell if there is an error in your code.
MIDI Input is available for Chrome and Firefox browsers. You can also use the computer keyboard to input MIDI notes. We are using key map as Ablelon Live: A-line and Q-line for keys, ZX to move octave, CV to change velocity.
The polyphonic mode can be activated by selecting a number of voices in the Poly Voices menu on the left. Standard Polyphony Parameters have to be used in the DSP voice.
A global effect can be added using the effect = foo; convention, possibly adapting the voice and effect inputs/outputs to make them compatible.
You can choose your audio device or use an audio file to simulate the audio input of DSP. Drap and drop your file to the waveform below to substitute the file.
Audio output can be switched on/off using the he Output is On/Output is Off button on the bottom-right.
The generated sound can be recorded on the fly then possibly saved using the Record/Save section on the bottom-right.
The currently edited DSP program can be sent to the remote compilation service which allows to export for a large number of targets, like various standalone formats, a Max/MSP external on macOS or Windows, a SuperCollider UGen.etc. Use the Truck button on the left column to access the list of possible Platform and the list of associated Architecture. Select one Platform/Architecture pair and wait a bit until you get the compiled result as a binary zipped file.
Note that some targets produce a source file (like source/cplusplus), or ready to be compiled projects (like juce/plug-in to create a JUCE project).
Tools to display audio signals are available in the left panel:
Mode can be Offline, Continuous, On Event (to synchronize visualisation with control event changes), and Manual (to be triggered with the Plot (Snapshot) button at the bottom)
Samples changes the number of displayed samples (so corresponding to the window width)
Sample Rate displays the currently selected audio device sample rate
FFT Size and FFT Overlap value are used when using Oscilloscope and Spectroscope modes
FFT Window selects Auto, None (Rectangular), Hann (N/4), or Blackman weighting. Auto uses an unscaled rectangular FFT for offline plots, exactly like numpy.fft.rfft: a unit impulse produces a flat 0 dB magnitude and zero phase. Captured-signal modes use Blackman automatically. The optional Hann mode applies the symmetric Hann window and N/4 scaling used by the reference Python plotter.
In the Plot Tab in the middle section, you can switch the visualization between Data, Interleaved, Oscilloscope, Spectroscope, Spectrogram, and Phase. Changing the acquisition mode does not change the active visualization. Note that when used with On Event mode, values in the Data visualisation mode only change at sample 128 (since On Event mode plots from 128 samples before the event).
The Spectroscope view can display magnitude as dBFS or normalized linear amplitude. In dBFS mode, the Bottom and Top number fields set the visible magnitude range directly. The Phase view displays wrapped phase in degrees. Both magnitude and phase views support linear or logarithmic frequency axes with ticks that follow the visible zoom range.
Waveform x-axis ticks are shown in seconds using the active audio sample rate. Drag across an offline waveform to select a range; its length is displayed in samples and seconds. With the plot canvas focused, copy (Ctrl+C/Cmd+C) places a CSV table containing sample index, time in seconds, and every channel value on the clipboard. Use Alt-drag to pan a waveform without changing the selection.
Double-click the x- or y-axis gutter to reset only that axis to its default zoom. In Data mode, sample indices fill each visible column from top to bottom before continuing in the next column.
Both input and output have an audio analyzer in the right panel. In the small audio output window at the bottom, you can switch the visualization between Oscilloscope and Spectroscope, or change buffer size and channel. Three numbers showing at right side are current value, estimated frequency and RMS.
In the display zone, zoom can be changed with a wheel or two-finger vertical gesture. Horizontal trackpad gestures pan the visible window.
The Diagram tab allows displaying the circuit SVG diagram. You can navigate inside it by clicking on the dark blue part to go down in the circuit hierarchy, or clicking on the diagram border to go up in the circuit hierarchy.
The soundfile primitive can be used in the IDE. Note that soundfile access may have to deal with the CORS issue.
The audio files have to be accessed:
by dropping all needed soundfiles in the Project Files section on the left, then directly accessing them in the DSP code
by using a full URL like https://raw.githubusercontent.com/grame-cncm/GameLAN/master/baliphone/Gamelan_1_1_C_gauche.flac
by defining the soundfile base URL folder with the declare soundfiles "https://raw.githubusercontent.com/grame-cncm/GameLAN/master/baliphone"; metadata, then the actual audio file name in the code. See this example. Several base URL can be listed with the declare soundfiles "https://url1;https://url2;https://url3"; kind of syntax
Soundfiles hosted on GitHub can also be accessed through the jsDelivr CDN, which provides CORS-enabled URLs, starting with https://cdn.jsdelivr.net/gh/. Therefore, the previous files can also be delivered using the following base URL: declare soundfiles "https://cdn.jsdelivr.net/gh/grame-cncm/GameLAN/baliphone";.
To access local audio files, a local server can be started:
use the local server URL http://127.0.0.1:8000 (or http://localhost:8000) with declare soundfiles "http://localhost:8000";). Additional URLs can always be declared if needed.
install the required package with pip install Flask Flask-CORS (or pip3 install Flask Flask-CORS)
create a folder my_faust_server. Inside create a text file server.py with the following code and a folder static_files. Place your audio files inside static_files.
from flask import Flask, send_from_directory, abort
from flask_cors import CORS
import os
app = Flask(__name__)
# Specify trusted origins
trusted_origins = os.getenv(
"TRUSTED_ORIGINS", "http://localhost,http://127.0.0.1,https://faustide.grame.fr"
).split(",")
CORS(app, origins=trusted_origins)
# Serve files from a dedicated directory (e.g., "static_files")
STATIC_FILES_DIR = "static_files"
os.makedirs(STATIC_FILES_DIR, exist_ok=True) # Ensure the directory exists
@app.route("/<path:filename>")
def download_file(filename):
# Simple validation to prevent directory traversal
if ".." in filename or filename.startswith("/"):
abort(404) # Not found for invalid paths
return send_from_directory(STATIC_FILES_DIR, filename)
if __name__ == "__main__":
app.run(debug=False, port=8001) # Disable debug mode in production
python server.py(or python3 server.py).Here are polyphonic samplers examples.
The recommended browsers are the latest versions of Chrome and Firefox for AudioWorklet and MIDI, but it requires an https connection to use the audio inputs.
Install Git and Node.js 20.12 or newer in the Node 20 release line, matching CI. npm is included with Node.js.
Clone a copy of the repo then change to the directory:
git clone https://github.com/grame-cncm/faustide.git
cd faustide
Beware: on Windows, before cloning the repository, and for the libfaust-wasm.data file line ending to be correctly handled, you'll have to do:
git config --global core.autocrlf false
Install the exact dependencies recorded in package-lock.json, build the
development bundle, and start the local server:
npm ci
npm run build
npm run serve
Open http://127.0.0.1:8000/dist/. The server does not rebuild or hot-reload
source files: after a source change, run npm run build again and reload the
browser. Keep the hostname consistent because browser storage is scoped to the
origin.
To exercise the optimized production-equivalent bundle locally:
npm run dist
npm run serve
If the server is already running, only reload the browser after npm run dist.
Do not change the version in package.json merely to run a local build.
Note that the src/static/examples folder is a manually copied subset of the faust/examples that will have to be updated from time to time.
For validation commands, browser/audio caveats, production builds, and troubleshooting, see the local deployment guide.
The TypeScript runtime is organized as a small composition root (src/index.ts)
plus layered modules: model/ (project rules), runtime/ (DOM-free services and
state), ui/ (controllers/views), and scope/ (analyser/plot widgets).
Quality bar for any contribution — keep the code clean, well-tested, and well-documented:
Tests first. Add a test that pins the current behavior before changing it,
and keep the suite green at every commit. Bug fixes ship with a regression
test. The setup is a three-layer pyramid: ESLint/Stylelint, Vitest (jsdom)
unit/integration tests, and Playwright end-to-end tests against the built
dist/.
Run the gate before committing:
npm test # lint (eslint + stylelint)
npm run test:unit # vitest
npm run build
# and, for browser-visible / bundle changes:
npm run dist && npm run test:e2e
CI runs the same gate (with coverage thresholds) on every push/PR; keep it green. Coverage thresholds are a ratchet — raise them as you add tests, never lower them to pass.
Document what you touch. Public modules and methods carry TSDoc headers
describing their role; keep doc/refactor-plan.md (architecture + phase
status) up to date when structure changes.
Small, focused diffs. No drive-by reformatting, no dead code, no scope creep.
This repository is set up to be worked on with AI coding agents (Claude Code,
Cursor, Copilot, …). The rules an agent must follow — architecture boundaries,
the characterize-first workflow, the validation gate, documentation and commit
conventions, and anti-hallucination guardrails — are written in
AGENTS.md, which compatible tools read automatically.
If you use an agent, you remain the author and reviewer: you own everything in the PR. In practice that means
A change produced with an agent is held to exactly the same standard as one
typed by hand — AGENTS.md is the contract for both.
https://faustide.grame.fr/?realtime_compile=0 URL