CentiLeo Renderer由俄罗斯CentiLeo LLC公司开发,目标是快速逼真地渲染超大3D场景。CentiLeo计算完全基于GPU(基于CUDA,暂不支持帕斯卡),使用新颖和高效的缓存系统,场景的几何结构和纹理的量可以远大于GPU存储器容量,然而对于小和中等场景也非常快。CentiLeo渲染器具有灵活的标准材质和阴影系统,提供可以组合的所有基本阴影操作。CentiLeo的光线跟踪基于标准算法但具有一些优化,能智能地将更多的计算能力引导到需要更多计算的图像部分。
CentiLeo现已经集成在Cinema 4D中。虽然目前处于初始Alpha测试阶段,但它的IPR支持流畅的交互式预览渲染,不管是在编辑Cinema 4D中的几何体和实例对象,或者是标签,材质,阴影和照明,都可以得到实时的预览反馈。当场景中的几何体和纹理计算预览时所使用的大小超过显卡的可用显存的时候,CentiLeo依然可以快速的处理这些数据。这项技术是通过计算机的内存来对显卡的显存进行扩展,所以这是一个非常有价值的技术。在渲染工作中,我们已经可以将显卡显存的大小限制抛之于脑后,得益于这种高效的渲染数据流组织。
CentiLeo渲染器不仅仅适用于大型场景,在每日的一些小场景渲染中也具有很强的竞争力。它具有强大的着色能力、照明系统、AOV、置换贴图以及适用于景深效果的自适应光线追踪功能。预计在未来几个月内将会在提高现有功能的基础上添加更多显著提高生产力的功能。
未来更新: Developed Features - Out-of-core geometry (even hundreds million polygons) with GPU acceleration
- Out-of-core and out-of-RAM textures with GPU acceleration
- Adaptive displacement mapping
- Noise aware path tracing
- Specific Metropolis sampler with firefly killer
- Instancing of geometry
- Fast support of dynamic scenes and interactive modeling
- Flexible shader infrastructure and currently implemented shaders of file texture, procedural noise, binary math, falloff, triplanar projection, color correction, blend, const, HDRI (only for environment maps).
- Powerful Standard CentiLeo Material with 2 reflection lobes (potentially anisotropic) and Multi Material container.
- Light sources with assignment to different layers of materials (visible to reflection, diffuse, etc.).
- AOVs/Render passes infrastructure to generate them all in parallel. Currently only light group passes are in publicly available package.
- Fully interactive 3ds Max plugin which reacts to user modeling activity in 3ds Max while the image is rendered.
- Bump mapping can be applied on top of displacement mapping solution and can be combined with bluish normal maps.
- Arbitrary output image resolution (as soon as it fits to CPU RAM).
Limitations
- Up to 30 total ray bounces and up to 10 diffuse ray bounces.
- Up to 16K textures.
- NVIDIA only GPUs with compute capability 2.0 and higher and at least 2 GBs of device memory (however we recommend 3 and more GB of memory).
|