ChenlizheMe/Infernux
ChenlizheMe/Infernux has 2,495 GitHub stars and 81 forks. GitHot measured +30 stars in 24 hours and +190 in 7 days. Last checked 2026-11-11 10:53 UTC. Last push 2026-10-28.
Comment ces chiffres sont mesurés
Développement d’un moteur natif pour les réseaux neuronaux (3N). La création se fait en Python et l’exécution repose sur un moteur d’exécution en C++. Compatible avec Vulkan/WebGPU, Windows, Linux, Android et le Web.
What you can do
- Python first. Components with Inspector fields, coroutines and prefabs. Project scripts and plugins hot-reload transactionally, including the behaviour of instances that already exist.
- A renderer you can author. Describe a RenderStack in Python and the native RenderGraph runs it: Forward, Forward+ and Deferred paths, PBR, shadows, post-processing, GPU particle graphs, skinned animation, screen and world-space UI.
- A complete game loop. Multi-scene editing, FBX and Blender import with live source sync, Jolt rigid bodies and scene queries, audio, animation state machines and Timeline.
- Compute in the game. Batch NumPy access to world data, CPU/GPU JIT and Taichi integration for numerical gameplay. This is the data path models will use.
Progression mesurée des étoiles
- in 24h
- +30
- in 7 days
- +190
- in 15 days
- Historique en cours
- in 1 month
- Historique en cours
Variations nettes sur des périodes qui se chevauchent, pas des totaux quotidiens. Les mesures absentes restent vides. Les étoiles indiquent l’intérêt, pas la qualité.
README.md
Original0.4.3 hotfix 8 brings reversed-Z rendering, isolated camera settings, repaired skies and shadows, and fixes for Hub window flashes and startup focus. See the changelog.
Infernux · 熔炉
Build worlds. Give them intelligence.
An open-source game engine you write in Python, on its way to 3N: a Neural Network-Native engine.
Website · Download · Tutorials · API · Roadmap · Community · 简体中文
What is Infernux?
Infernux is a game engine where Python is where you work and C++ is where it runs. You write gameplay, editor tools, render pipelines and numerical code in Python; a C++ runtime, a Vulkan renderer and Jolt physics do the heavy lifting. The same project ships to Windows, Linux, Android and the browser.
It is also built toward a specific goal: 3N, a Neural Network-Native engine. In most engines, AI is a plugin bolted on after the game is finished. In a 3N engine the world is data a model can read, the simulation is a place where it can act, and the shipped game is somewhere it can run.
import infernux as inx
class Spin(inx.InxComponent):
speed: float = inx.serialized_field(default=90.0, range=(0.0, 360.0))
def update(self, delta_time: float) -> None:
self.transform.rotate(inx.Vector3(0.0, self.speed * delta_time, 0.0))
An ordinary Python class. speed appears in the Inspector and can be tuned while the game runs.
[!WARNING] Infernux is alpha software under active development. APIs still change between releases, and the neural-network features below are roadmap items rather than shipped features. It is ready for experiments, tools and small games, not yet for a production schedule.
Highlights
- Python first. Components with Inspector fields, coroutines and prefabs. Project scripts and plugins hot-reload transactionally, including the behaviour of instances that already exist.
- A renderer you can author. Describe a RenderStack in Python and the native RenderGraph runs it: Forward, Forward+ and Deferred paths, PBR, shadows, post-processing, GPU particle graphs, skinned animation, screen and world-space UI.
- A complete game loop. Multi-scene editing, FBX and Blender import with live source sync, Jolt rigid bodies and scene queries, audio, animation state machines and Timeline.
- Compute in the game. Batch NumPy access to world data, CPU/GPU JIT and Taichi integration for numerical gameplay. This is the data path models will use.
- An editor you can extend. Dockable panels and menus written in Python. InxPackage plugins carry components, tools, assets and platform exporters. An optional MCP plugin lets agents drive the editor through the same command and undo paths you use.
- Built to be inspected. Assets keep a GUID identity from import to cooked Player. Players load a packed asset index, never your project folder.
Platforms
| Target | Editor | Player | Graphics |
|---|---|---|---|
| Windows x64 | ✅ | ✅ | Vulkan |
| Linux x86_64 | ✅ | ✅ | Vulkan |
| Android arm64 / x86_64 | — | APK / AAB | Vulkan |
| Web | — | HTML + WASM | WebGPU |
Requirements and known limits per platform are in SUPPORT.md.
Get started
Use the engine. Install InfernuxHub, pick an engine version and create a project. The Hub manages Python environments and Android tools for you; no native compiler is needed. Then follow the five-step start to a component spinning a cube, and continue with the courses.
Build from source. Windows needs a Visual Studio installation with the Desktop C++ workload, CMake 3.25+, the Vulkan SDK and Python 3.13. The Windows preset lets CMake select the newest compatible Visual Studio generator installed on the machine.
git clone --recurse-submodules https://github.com/ChenlizheMe/Infernux.git
cd Infernux
./scripts/setup/configure_development.ps1
conda activate infernux
cmake --preset windows-msvc-release
cmake --build --preset windows-msvc-release
cmake --build --preset windows-msvc-install-wheel
python packaging/launcher.py
If configuration cannot create the selected generator, install a CMake version
that lists a compatible Visual Studio generator in cmake --help, or set
CMAKE_GENERATOR to one of the listed generators and configure again.
On Linux, run scripts/setup/install_linux_dependencies.sh and bash scripts/setup/configure_development.sh, activate infernux, then use the linux-clang-release and linux-clang-install-wheel presets. CONTRIBUTING.md covers the full workflow.
The road to 3N
| Stage | Status | What it unlocks |
|---|---|---|
| Python authoring, batch world APIs, CPU/GPU compute | ✅ Available | Models and gameplay share one data path into the world |
| Deterministic stepping, snapshots and replay | 🔜 Planned | Reproducible runs for debugging and learning |
| Batched worlds and a tensor data plane | 🔜 Planned | Many environments stepping side by side, zero-glue data exchange |
| Portable inference in the Player | 🔜 Planned | Trained models ship inside the game on every target |
Ten interactive maps chart every system on the roadmap. Release notes live in UpdateLog.md.
Community
- 💬 Questions, demos and ideas: community forum
- 🐛 Reproducible bugs: GitHub issues
- 🧩 Plugins: authoring guide and template
- 🛠️ Contributing: start with CONTRIBUTING.md
License
Infernux is released under the MIT license: free to use, no royalties. Third-party notices are indexed in licenses. Windows builds are signed free of charge by SignPath.io with a certificate from the SignPath Foundation (policy).
Citing Infernux
@software{chen2026infernux,
author = {Chen, Lizhe},
title = {Infernux},
year = {2026},
version = {0.4.3},
url = {https://github.com/ChenlizheMe/Infernux}
}
Explorer des projets similaires
codecrafters-io/build-your-own-x
Maîtrisez la programmation en recréant de zéro vos technologies préférées.
561,844 étoilespublic-apis/public-apis
Une liste collective d'API gratuites
498,199 étoilesmorluto/rea
Rétro-ingénierie de n'importe quoi avec des agents, du comportement des applications jusqu'aux binaires natifs.
452,426 étoilesopenclaw/openclaw
L'IA qui fait vraiment les choses. Tout système d'exploitation, toute plateforme. À la manière du homard. 🦞
394,362 étoiles