Get in contact on Discord! https://discord.gg/wuYPxUF
Check out the documentation (WIP) here: https://gscept.github.io/nebula-doc/
- OS: Windows and Linux(WIP)
- Compiler with support for C++17.
- GPU and drivers supporting Vulkan 1.2+
- CMake 3.21+
- Python 3.5+
- Python requirements (Windows):
- Matching architecture (64-bit if you're building for 64-bit systems)
- Installed for all users
- Added to PATH
- Installed with debugging symbols and binaries
- Python requirements (Windows):
./fips set config vulkan-win64-vs2022-debug
in your project directory, alternatively vs2019 or vs2017, depending on your version
Run fips nebula
verb to set work and toolkit directory registry variables:
fips nebula set work {PATH}
fips nebula set toolkit {PATH}
In your project directory:
fips fetch
fips physx build vc17 debug
(if you are running VS 2022, usevc16
orvc15
for vs 2019/2017 instead)fips anyfx setup
fips ultralight
fips build
fips physx deploy vc17 debug
Nebula is being developed continuously, which means that features keep getting added all the time. Currently, we support this:
- Completely data-driven design from bottom to top.
- Data structure suite, from containers to OS wrappers, everything is designed for performance and minimal call stacks.
- Multithreading.
- SSE-accelerated and intuitive maths library.
- Full python supported scripting layer.
- Advanced rendering framework and shaders.
- Test-benches and benchmarking.
- Profiling tools.
A lot of effort has been made to the Nebula rendering subsystem, where we currently support:
- Unified clustering system - fog volumes, decals and lights all go into the same structure.
- Screen-space reflections - working condition, but still work in progress.
- SSAO - Horizon-based ambient occlusion done in compute.
- Physically based materials and rendering.
- Multi-threaded subpass recording.
- Shadow mapping for local lights and CSM for global/directional/sun light.
- Volumetric fog and lighting.
- Geometric decals.
- CPU-GPU hybrid particle system.
- Skinning and animation.
- Scripted rendering path.
- Vulkan.
- Tonemapping.
- Asynchronous compute.
- Virtual texturing using sparse binding.
- Fast and conservative GPU memory allocation.
Nebula has historically had a database-centric approach to entities. With the newest iteration of Nebula, we've decided to keep improving by adopting an ECS approach, still keeping it database-centric.
- Data-oriented
- Data-driven
- Minimal memory overhead per entity.
- High performance without compromising usability or simplicity
- Blueprint and template system for easily instantiating and categorizing entity types.
- Automatic serialization and deserialization
Deferred Lighting using 3D clustering and GPU culling.
Geometric decals, culled on GPU and rendered in screen-space.
Volumetric fog lighting.
Local fog volumes.
Profiling tools.