This SOP will take curve geometry and convert it to a VDB volume suitable
for rendering. Curves will first be converted to a point cloud by
resampling, which are then rasterized to the resulting volume. The volume
can contain the average hair tangent vector in the
tangentu field which
can be used in instead of
dPdt in hair shaders.
Using VDB volumes to render hair can speed up rendering of background characters with hair, since it avoids the time and memory overhead of rendering curves directly. It can also be used to implement a global illumination system for hair, which would otherwise be very expensive or noisy due the low probability involved in point sampling hair strands.
Size of the voxels in the created volumes.
The filter shape that to use when splatting particles. The gaussian filter will produce smooth blobs and is useful for larger particles. When the particle size is smaller than the voxel size, a box filter will produce a similar result in less time, so is the default as this is usally the case for hair.
Controls the resampling point density when converting curves to points. The default of 1 will create points along the curve with spacing approximately equal to the voxel size. Larger value for the sampling quality will decrease the point spacing, improving accuracy.
Scales the values in the resulting density volume by a constant factor.
Scales the width of the rasterized curves.
Color Shop Path
Specifies a CVEX shader to be applied after resampling to create or
modify attributes on the resampled points. The shader may use the
curveu float parameter to determine the parametric position on
the curve - for example to evaluate a color ramp.
Primitives per Batch
To avoid excessive memory use for the temporary resample points, this parameter limits the number of primitives that are resampled and rasterized at a time. Internally, a ForEach SOP iterates over each batch.
A volume containing the average hair tangent vector will be created when enabled.
Tangent Voxel Scale
Specify a scale on the voxel size for the generated
Tangent Filter Width
Controls the minimum filter width to use when choosing a tangent vector for rendering.
When filtering, the
tangentu field isn’t averaged like
density, color or other attributes. This ensures that the
tangent vectors are not blended together, which could produce
new invalid tangent vectors that didn’t exist in the original
curves. Instead, the rasterization algorithm preserves the
original tangents by randomly selecting a single tangent vector
for each voxel. In mantra, these tangents are randomly sampled
to find a plausible tangent vector in the neighborhood of each
shading point. This parameter controls the search radius used
by mantra to find these candidate tangent vectors.
In general, larger values will reduce noise but increase blurring of the tangent field while smaller values will be sharper but will render with voxel grid noise. The default of 4 will allow 4^3 or 64 candidate samples which will usually render with little noise.
Color Voxel Scale
Specify a scale on the voxel size for the generated Cd field, if
there is a
Cd point attribute on the input geometry
Velocity Blur Type
Controls the way the velocity attribute
v is treated during
No Velocity Blur
Disable velocity blur
Create a VDB vector field for the velocity and rasterize the
v values into this field. This option will generally
result in correct occlusion when rendering the volume with
nearby surfaces, but it can produce incorrect results when
nearby curves have different velocities.
Rasterize Velocity Blur
Bake the velocity blur into the generated volume by averaging along the motion path. This option will produce more consistent motion blur when nearby curves have different velocities but can lead to incorrect occlusion.
Velocity Voxel Scale
Specify a scale on the voxel size for the generated v field, if
there is a
v point attribute on the input geometry.
The shutter time to use for velocity blur rasterization.
The shutter offset to use for velocity blur rasterization.
The number of motion blur samples to use for velocity blur rasterization. Larger numbers of samples will produce smoother and more accurate results but will take more time to compute.