This property node causes subsequent SoVolumeRender nodes to be drawn with different rendering effects and/or levels of quality.

NOTE:

• Only one (or none) of the following nodes can be active at a time: SoVolumeShader or SoVolumeRenderingQuality. However since Open Inventor 7.1, both SoVolumeRenderingQuality and SoVolumeIsosurface may applied to the same SoVolumeRender node.

• If the application simply needs to replace one or more of the VolumeViz shader rendering functions with a customized shader, you can use either SoVolumeShader or SoVolumeRenderingQuality. However if the application needs the advanced rendering features of SoVolumeRenderingQuality, e.g. lighting, in addition to the customized behavior, then you should use this node (which is derived from SoVolumeShader). Using the shader rendering framework is explained on the SoVolumeShader page.

• The SoVolumeRenderingQuality node must be before the SoVolumeRender and after the SoTransferFunction.

• This is a shader node! The effect will usually be undesirable if it is applied to standard geometry (polygons, lines, etc). Therefore applications should generally keep the volume visualization nodes and standard geometry nodes separate in the scene graph (i.e. under different SoSeparator nodes).

• Please read the comments for each field. Some options have no effect in ray-casting mode (the default) and some options only apply in ray-casting mode. Similarly, some options only apply to gradient lighting and have no effect on deferred lighting.

  Pre-integrated and lighted rendering

  Various enhanced rendering modes for volume rendering are available:

  • Pre-integrated volume rendering

  • Lighted volume rendering

The pre-integrated mode (preIntegrated field) generally provides higher image quality for the same number of slices. Most applications should enable this field. However, note that pre-integration integrates between color map values. For typical scalar data containing sampled values, this provides a beneficial smoothing. It is especially useful when the color changes sharply between adjacent color map entries. However, when the color changes sharply between adjacent voxels, it can can cause values that are not actually in the original data to be displayed. This is undesireable for some data, for example "label" volumes resulting from segmentation. Pre-integration is not recommended for such data. Also note:

• The preIntegrated field has no effect on SoVolumeIsosurface rendering.

• The preIntegrated field is only considered when SoVolumeRendering.renderMode is set to VOLUME_RENDERING (the default).

When lighting is enabled for volume rendering, VolumeViz applies the same lighting equation used for polygonal geometry, including (optionally) specular highlights. The base voxel color comes from the SoTransferFunction node or (optionally) from a custom shader function provided by the application. This color is modified using the current material (SoMaterial), a vector simulating a "normal vector" and one or more directional light nodes (SoDirectionalLight) found in the scene graph. Voxels can also cast and receive shadows (see SoShadowGroup).

Limitation: Pre-integrated and lighted rendering are not supported if using custom fragment shaders and redefining FRAGMENT_COMPUTE_COLOR slot.

VolumeViz supports two lighting modes using either the lighting field or the deferredLighting field. In both cases lighting is computed on the GPU as part of the rendering process. (Do NOT use the lighting field in the SoVolumeRendering node. This field enables a CPU lighting computation that is slow.)

• Gradient lighting The lighting field enables gradient based lighting, where the "normal vector" at each sample along the ray is a gradient vector computed from the data values in the volume. Gradient based lighting only supports a single light source and only the first light in the scene graph (typically the viewer's "headlight") is used. No other lights affect the volume in this case. The lightingModel field controls the lighting equation to be used. Using the OPENGL model (not the default) is recommended. The gradientQuality field controls the algorithm used to compute gradient vectors. Several other fields affect the gradient computation including gradientThreshold, surfaceScalarExponent and unnormalizedGradientExponent. Setting surfaceScalarExponent to a small value, for example 2.0, is recommended.

• Deferred lighting The deferredLighting field enables screen space lighting, where the "normal vector" is computed from the final image depth buffer. Deferred lighting is faster and supports multiple light sources, but works best when the transfer function makes each data value either opaque or transparent. Deferred lighting is not affected by the lightingModel field or any of the gradient related fields.

Unlike other primitives (including other VolumeViz primitives), volume lighting is not affected by an SoLightModel node. Also unlike other primitives, if lighting is enabled and there are no light nodes in the scene graph, the voxel color is taken from the transfer function or custom shader function "as is" (other primitives would be black in this case).

Each light node's direction and intensity fields are used, but the color field is not currently used. The current SoMaterial specifies the ambient, specular, diffuse, and emissive color values for the lighting equation. Note that the default diffuse color is "light gray" (0.8), not full white. This allows specular lighting to push the color toward full white (as expected). If specular lighting is not desired, then changing this to full white (1.0) is recommended in order to see the true colors specified in the transfer function.

The following figures show the same volume data:

<TABLE border=1 cellspacing=0 cellpadding=5>
  <TR><TD valign=_top> Default volume rendering  <TD valign=_top> Pre-integrated volume rendering  <TD valign=_top> Lighted pre-integrated volume rendering
  <TR><TD valign=_top> @image html quality_volume.jpg
            <TD valign=_top> @image html quality_preintegrated.jpg
            <TD valign=_top> @image html quality_light.jpg
</TABLE>

Quality enhancement parameters

Jittering: When jittering is set to true, a random offset is added to texture coordinates in order to decrease "ring" artifacts without the cost of drawing a higher number of slices. Note that this creates a "noisy" image. Instead we recommend setting the SoVolumeRender.samplingAlignment field to BOUNDARY_ALIGNED.

  <TABLE border=1 cellspacing=0 cellpadding=5>
    <TR><TD valign=_top> No Jittering  <TD valign=_top> With Jittering
    <TR><TD valign=_top> @image html volume_nojitter.jpg
    <TD valign=_top> @image html volume_jitter.jpg
 </TABLE>

Gradient quality: When gradient lighting (lighting field) is enabled, the gradientQuality field allows you to choose between various gradient computation techniques. The computational cost increases with the quality. Has no effect on deferred lighting.

  <TABLE border=1 cellspacing=0 cellpadding=5>
    <TR><TD valign=_top> Low quality  <TD valign=_top> Medium quality  <TD valign=_top> High quality
    <TR><TD valign=_top> @image html volume_forwarddiff.jpg
    <TD valign=_top> @image html volume_centraldiff.jpg
    <TD valign=_top> @image html volume_sobel.jpg
 </TABLE>

Surface scalar: When gradient lighting (lighting field) is enabled or edgeDetect2DMethod is GRADIENT, the surfaceScalarExponent field disables lighting (or edge detection) on uniform surfaces in order to avoid noise in these area. This field should not be mixed with unnormalizedGradientExponent. The default value is zero, but a small value, for example 2.0, is recommended. Has no effect on deferred lighting.

  <TABLE border=1 cellspacing=0 cellpadding=5>
    <TR><TD valign=_top> Surface Scalar disabled   <TD valign=_top> Surface Scalar enabled
    <TR><TD valign=_top> @image html surfaceScalarOff.jpg
    <TD valign=_top> @image html surfaceScalarOn.jpg
 </TABLE>

Unnormalized gradient : When gradient lighting (lighting field) is enabled, if unnormalizedGradientExponent is not 0, voxels with small gradients will get more contribution from the ambient light than voxels with high gradients. It is similar to surfaceScalarExponent but uses the ambient light instead of the transfer function color for uniform surfaces. Has no effect on deferred lighting.

Lighting model: When gradient lighting (lighting field) is enabled, the lightingModel field allows backward compatibility with the lighting model used in Open Inventor 6 or using the standard OpenGL lighting model. Setting this field to OPENGL is recommended. The following picture shows that in the case of OIV6, the volume is much brighter than the sphere. By setting the lighting model to OPENGL, the brightness of the volume matches the brightness of the sphere. Has no effect on deferred lighting.

  <TABLE border=1 cellspacing=0 cellpadding=5>
    <TR><TD valign=_top> OIV6   <TD valign=_top> OPENGL
    <TR><TD valign=_top> @image html volume_oiv6light.jpg
    <TD valign=_top> @image html volume_oiv7light.jpg
 </TABLE>

Gradient threshold: When gradient lighting is enabled, gradients with a length less than gradientThreshold are ignored during the lighting computation. This avoids doing lighting on noise while still lighting important data. In the following screenshots, setting a threshold of 0.1 removed lighting on the noise surrounding the spheres. Has no effect on deferred lighting.

  <TABLE border=1 cellspacing=0 cellpadding=5>
    <TR><TD valign=_top> With #gradientThreshold set to 0  <TD valign=_top> With #gradientThreshold set to 0.1
    <TR><TD valign=_top> @image html volume_gradientThr0.jpg
    <TD valign=_top> @image html volume_gradientThr1.jpg
 </TABLE>

Image enhancement parameters

Various image enhancement techniques are available in this node and in SoTransferFunction.

• Ambient occlusion The ambientOcclusion field enables a rendering technique that simulates self-shadowing of the volume. In other words, the amount of ambient (global) light in the scene reaching each sample is reduced by neighboring voxels. This effect makes it much easier to see the relative depth of features in the volume. Generally we recommend using this effect rather than the following effects. Both lighting and/or shadow casting may also be enabled, but neither is required to use ambient occlusion. This effect works best when the volume data contains surfaces (region boundaries with relatively sharp gradients) or in voxelized rendering mode. In both cases it works best when voxels are either transparent or nearly opaque. If you use ambientOcclusion, you should set the SoVolumeRender.samplingAlignment field to BOUNDARY_ALIGNED to reduce "slicing" artifacts.

• Boundary opacityboundaryOpacity increases opacity depending on the length of the gradient vector. Areas with large gradient changes will have their opacity increased according to the boundaryOpacityIntensity. Note that this option has a significant performance penalty.

• Edge coloring When edgeColoring is on, the color of each voxel may be mixed with the edgeColor. Areas where the normal (computed from the gradient) is facing the camera will have an unmodified color, whereas areas where the normal is more perpendicular to the view direction will tend towards edgeColor.

• Edge detection When edgeDetect2D is enabled, an image space filter is applied on the volume rendering image in order to detect edges, which will be highlighted. The results are affected by the edgeDetect2DInnerThreshold and edgeDetect2DOuterThreshold fields. The edgeDetect2DMethod bitmask allows to apply the edge detection on the image's luminance, depth and/or gradient. The gradient method may give noisy result, the surfaceScalarExponent may help to improve the result in this case. The gradient method has a significant performance penalty.

The following table shows the available edge detection techniques (explanation of faux shading is in SoTransferFunction):

No Edges	Boundary Opacity	Edge 2D
Edge Coloring	Faux Shading (see SoTransferFunction)

Because this node is derived from SoVolumeShader, IVVR_FIRST_RESERVED_TEXTURE_UNIT applies to it. See SoVolumeShader for more information.

Volume projection

Volume projection (SoProjection or derived classes) is incompatible with some options enabled by this node. Do not enable the preIntegrated, jittering or edgeDetect2D fields when using projection.

FILE FORMAT/DEFAULT

VolumeRenderingQuality {

lighting	false
preIntegrated	false
jittering	false
gradientThreshold	0.0001
edgeColoring	false
edgeColor	(0, 0, 0)
edgeThreshold	0.2
boundaryOpacity	false
boundaryOpacityIntensity	1.5
boundaryOpacityThreshold	1.5
edgeDetect2D	false
edgeDetect2DInnerThreshold	0.1
edgeDetect2DOuterThreshold	0.1
edgeDetect2DMethod	LUMINANCE
gradientQuality	MEDIUM
lightingModel	OIV6
colorInterpolation	true
unnormalizedGradientExponent	0
surfaceScalarExponent	0
segmentedInterpolation	false
segmentedInterpolationThreshold	0.5
voxelizedRendering	false
voxelOutline	false
voxelOutlineThreshold	2.
voxelOutlineWidth	2.
voxelOutlineColor	(0, 0, 0)
ambientOcclusion	false
deferredLighting	false
interpolateOnMove	false

}

ACTION BEHAVIOR

SoGLRenderAction Sets volume rendering quality parameters in the traversal state.

Reference