New features in this preview build
Instancing
New node types:
- Geometry group node: combines all the geometry on its inputs.
- Placement: creates a new instance at a different position, using a transform node.
- Scatter: creates multiple instances, given a list of transform matrices. A matrix is given as 12 numbers, which are the first three rows of
the transformation matrix. They are given in row major order. - Material mapping node: creates an instance of a mesh with some materials replaced.
You can create light sources with instancing. If scaled, the intensity stays the same, the power changes with the square of the scale
factor. IES lights are oriented according to the instance orientation. If a light source is stretched, the direct light calculation is not fully
correct anymore.
Other new features
- Falloff node: a texture which depends on the angle of incidence of the viewing ray.
- Color correction: applies a levels and gamma correction to the input texture.
- Russian roulette on colored surfaces has been improved.
- In the cuda device manager, the checkboxes for GPU’s that are not supported are not shown. You can only enable GPU’s with a supported
compute model. - Smoothing works on a per material input pin basis, rather than per material node. A mix material is smooth if either of its input materials is
smooth (There is still a bug in the updating of mix materials). - If you click on the statistics bar below the rendered image, you get a popup with texture counts and memory usage.
- Octane contains new shading code, so there is a chance that some materials don’t show up correctly.
- There are minor changes to the OCS reader.
for most users yet. Instancing will become much more useful when it is released with the plugins (and later in the standalone version when
we specify an exporter file format).
Download here:
Instancing test scenes:
- Scatter scene
- Sierpinski scene
Windows:
ZIP archives:
- 64-bit — 32-bit
MacOS X:
- Universal (32-bit and 64-bit)
Linux:
- 64-bit
Known issues
- Due to specific code required to support instancing, the rendering performance of this preview build is about 10-12% slower than the 2.58, even in scenes that don't use instancing.
- There is a random bug which causes light patterns and intensity to look incorrect about 1/5th of the time. Reloading the scene works in most cases.
For Linux and Mac OS you also need to install the CUDA Toolkit 4.0 (http://developer.nvidia.com/cuda-toolkit-40).
For Mac OS you also need the latest CUDA 4.0 driver (http://www.nvidia.com/object/mac-driver-archive.html).
All builds support only Tesla and Fermi cards (compute model 1.0 to 2.1).
Kepler GPUs are not supported at the moment but a fully functional build is being developed.
Yours,
The OctaneRender™ Team.
