Class SoBufferedShape

  • All Implemented Interfaces:
    SafeDisposable
    Direct Known Subclasses:
    SoVolumeBufferedShape

    public class SoBufferedShape
    extends SoShape
    Node to render geometry stored in SoBufferObject objects. SoBufferedShape is useful to manage the rendering of large geometry, provide application control over where the data is stored (CPU or GPU) and to integrate rendering with the Open Inventor computing framework (through the SoBufferObject classes).

    SoBufferedShape provides fields for:

    • Vertices
    • Indices (optional)
    • Colors (optional)
    • Normals (optional)
    • Texture coordinates (optional)

    In this sense it is similar to the SoVertexProperty node, but SoVertexProperty is just a property node. SoBufferedShape also does the rendering of the shape. Properties that are not specified are taken from the traversal state (e.g. colors) or computed (e.g. normals).

    SoBufferedShape can render many types of geometric primitives including points, lines, quads and triangles. (A single type must be specified per instance of SoBufferedShape.) You specify the type of primitive in the SoSFEnum field shapeType.

    SoBufferedShape can render multiple primitives of the same type. You can specify the number of vertices (or indices if provided) to use for each primitive in the SoMFInt32 field numVertices (similar to SoFaceSet).

    You can also use the primitive restart feature to define multiple indexed strip shapes, for example TRIANGLE_STRIP or LINE_STRIP. The end of each primitive is marked by a special index value in the index buffer and this value can be specified in the primitiveRestartValue field. The behavior is similar to the "-1" value that can be used in Open Inventor indexed shape nodes like SoIndexedFaceSet, but is implemented on the GPU.
    NOTE:

    The geometry and its attributes must be stored in buffer objects (see SoBufferObject). The buffer objects can be SoGpuBufferObjects stored directly on the graphics board or SoCpuBufferObjects stored in system memory. This allows the application to control what data is stored where.

    If lighting is enabled (there is no SoLightModel node or the model field of the SoLightModel is set to PHONG) and the normalBuffer field is not set, then Open Inventor will automatically compute normal vectors, but only in some cases (see Limitations section). Normal generation is affected by the creaseAngle field of the SoShapeHints node, but only if the vertices are NOT indexed (indexBuffer field is not set). If the vertices are indexed the creaseAngle is forced to PI, creating a smooth surface rendering. If the application needs to render sharp edges on a shape, either compute normal vectors and set the normalBuffer field or do not use the indexBuffer field. It is possible to disable normal generation (if for example the normals are generated by a geometry shader) by setting the useNormalsGenerator field to false. If no normal vectors are specified or generated, and lighting is enabled, the primitive may not be rendered correctly.

    SoBufferedShape provides fields to describe the content of each buffer, e.g. the data type and number of components in each buffer, as well as how to access the buffers, e.g. the offset into the buffer and "stride" separating data values in the buffer. The default values for offset and stride assume that the vertices, normals, etc are each in a separate buffer. However setting appropriate offset and stride allows, for example, vertices and normals to be interleaved in a single buffer. In this case the same buffer would be set into both the vertexBuffer and normalBuffer fields.

    To disable computing the bounding box, which can take a long time with very large geometry, use the SoBBox node to specify a pre-computed bounding box.

    Limitations

    • Transparency:
      If there is no color buffer, making the entire shape transparent using an SoMaterial node works as usual. However if there is a color buffer with RGBA values, note that Open Inventor does not currently check the color buffer for transparency (alpha values < 1). So in this case the SoBufferedShape will not be considered transparent geometry (even if there are alpha values < 1) and may not be rendered correctly. You can force Open Inventor to handle the shape as transparent geometry by putting an SoMaterial node with non-zero transparency before it in the scene graph.
    • Normal generation:
      If lighting is enabled and the normalBuffer field is not set, then Open Inventor will automatically compute normal vectors, but only in some cases. Automatic generation of normal vectors is ONLY enabled when:
      • The vertexComponentsCount field is set to 2 or 3,
      • The primitiveRestartEnabled field is set to false (default), and
      • The primitives are TRIANGLES, TRIANGLE_STRIP or QUADS.
        Note: The crease angle (see SoShapeHints) is not used by the normal generator if the vertices are indexed. If the application needs to render sharp edges on a shape, either compute normal vectors and set the normalBuffer field or do not use the indexBuffer field.
    • SoGetPrimitiveCountAction:
      When using the primitive restart feature, the triangle/line count returned by the SoGetPrimitiveCountAction will not be accurate.
    • Concave polygons:
      Unlike (for example) SoFaceSet, SoBufferedShape does NOT automatically tesselate concave or complex polygons. Such primitives may not be rendered correctly.
    • SoWriteAction:
      SoBufferedShape can be saved to and restored from a .iv file just like any other Open Inventor node. However, during the read operation any GPU buffer objects (SoGpuBufferObject) in the file will be created as CPU buffers (SoCpuBufferObject).
    • Material binding (etc):
      SoBufferedShape effectively only supports per-vertex and per-vertex-indexed binding of materials, normals and texture coordinates using the values found in its own buffers.

    Example using CPU buffer:

     // Result should be similar to SoLineSet example in PG-GettingStarted.pdf.
     // This example does not show any of the advantages of using SoBufferedShape,
     // just the simplest possible setup and usage.
     // Coordinate data
       float[] vertices = {
               1.0f, 0.5f,0.0f, 0.0f, 1.0f,0.0f, -1.0f,0.5f,0.0f,
              -1.0f,-1.0f,0.0f, 1.0f,-1.0f,0.0f,  1.0f,0.0f,0.0f, -1.0f,0.0f,0.0f,
              -1.0f,-1.5f,0.0f, 1.0f,-1.5f,0.0f
       };
       int[] numVerts = { 3, 4, 2 };
     
     // Create a CPU buffer object and set its size (allocate memory)
     SoCpuBufferObject cpuBuffer = new SoCpuBufferObject();
     cpuBuffer.setSize( vertices.length * Float.SIZE/8 );
     
     // Copy vertex data into the buffer object
     FloatBuffer vertData = cpuBuffer.map( SoBufferObject.AccessModes.SET ).asFloatBuffer();
         vertData.put(vertices);
     cpuBuffer.unmap();
     
     // Create a buffered shape to render the geometry
     SoBufferedShape shape = new SoBufferedShape();
     shape.shapeType.setValue( "LINE_STRIP" );
     shape.numVertices.setValues( 0, numVerts );
     shape.vertexBuffer.setValue( cpuBuffer );

    Example using GPU buffer:

     // Result should be similar to SoLineSet example in PG-GettingStarted.pdf.
     // This example does not show any of the advantages of using SoBufferedShape,
     // just the simplest possible setup and usage.
     // Coordinate data
       float[] vertices = {
               1.0f, 0.5f,0.0f, 0.0f, 1.0f,0.0f, -1.0f,0.5f,0.0f,
              -1.0f,-1.0f,0.0f, 1.0f,-1.0f,0.0f,  1.0f,0.0f,0.0f, -1.0f,0.0f,0.0f,
              -1.0f,-1.5f,0.0f, 1.0f,-1.5f,0.0f
       };
       int[] numVerts = { 3, 4, 2 };
     
     // Create a GPU (OpenGL) buffer and set its size (allocate memory)
     SoGLContext glContext = new SoGLContext( true );
     glContext.bind();
         SoGpuBufferObject gpuBuffer = new SoGpuBufferObject( SoGpuBufferObject.BufferAccessFrequencies.STATIC, SoGpuBufferObject.BufferAccessNatures.SHARED );
         gpuBuffer.setSize( vertices.length * Float.SIZE/8 );
     
     // Copy vertex data into the GPU buffer object
     FloatBuffer vertData = gpuBuffer.map( SoBufferObject.AccessModes.SET ).asFloatBuffer();
         vertData.put(vertices);
     gpuBuffer.unmap();
     
     glContext.unbind();
     
     // Create a buffered shape to render the geometry
     SoBufferedShape shape = new SoBufferedShape();
     shape.shapeType.setValue( "LINE_STRIP" );
     shape.numVertices.setValues( 0, numVerts );
     shape.vertexBuffer.setValue( gpuBuffer );

    LIMITATIONS: SoBufferedShape needs a graphic card supporting vertex buffer objects, if not available shape won't be rendered.

    File format/default:

    BufferedShape {

      useNormalsGenerator true
      shapeType TRIANGLES
      numVertices 0
      vertexBuffer NULL
      vertexComponentsCount 3
      vertexComponentsType SbDataType.FLOAT
      vertexStride 0
      vertexOffset 0
      normalBuffer NULL
      normalComponentsType SbDataType.FLOAT
      normalStride 3 * sizeof(float)
      normalOffset 0
      indexBuffer NULL
      indexType SbDataType.UNSIGNED_INT32
      indexOffset 0
      colorBuffer NULL
      colorComponentsType SbDataType.FLOAT
      colorStride 0
      colorOffset 0
      colorComponentsCount 3
      texCoordsBuffer NULL
      texCoordsComponentsType SbDataType.FLOAT
      texCoordsStride 0
      texCoordsOffset 0
      texCoordsComponentsCount 2
      primitiveRestartEnabled false
      primitiveRestartValue -1
    }

    Action behavior:

    SoGLRenderAction, SoCallbackAction, SoBoundingBoxAction
    Do the actual rendering / bounding box computation.

    See Also:
    SoCpuBufferObject, SoGpuBufferObject, SoBBox
    • Field Detail

      • shapeUsage

        public final SoSFEnum<SoBufferedShape.Usages> shapeUsage
        Defines the usage of the shape. The graphics driver can make some optimizations if it knows the usage of the shape. Most of the time if we update only a small part of the buffer objects attached to the shape on a regular basis we prefer DYNAMIC, otherwise we prefer STATIC.

        STATIC provides the best performance even if we update the buffers on a per frame basis, as long as we are replacing the contents of the buffer using the SET access mode.

        This field does not have any effect if the specified buffer objects are SoGpuBufferObjects, because the static/dynamic behaviour is set on each buffer object.

        . The default value is STATIC.

        Since:
        Open Inventor 9.2

      • primitiveRestartEnabled

        public final SoSFBool primitiveRestartEnabled
        Enable/disable the primitive restart feature. Default is false. Primitive restart allows you to define multiple indexed strip shapes using only one index buffer. Each time the primitive restart index is reached a new strip or loop of primitives is emitted. This feature is similar to the "-1" that can be used in the OIV indexed shapes This also means that the availability must be checked before being used

        Limitations: Enabling primitive restart disables the normal generator.

        Since:
        Open Inventor 8.5

      • primitiveRestartValue

        public final SoSFInt32 primitiveRestartValue
        Index value for the primitive restart feature. Default is -1.

        Since:
        Open Inventor 8.5

      • useNormalsGenerator

        public final SoSFBool useNormalsGenerator
        Indicates if the node should use the internal normal vector generator if no normals are defined. Default is true.

        This mode is only supported for shapes with float coordinates and 3 components per vertex. It is not supported for the points and the lines.

        Disabling the normal generator can be useful if the normals are computed in a shader or if the shaders don't need any normal at all.

        Normal generation is affected by the creaseAngle field of SoShapeHints.

      • numVertices

        public final SoMFInt32 numVertices
        Total number of vertices/indices or number of vertices/indices to be used per primitive. Specifically:
        • For the shape types POINTS, LINES, TRIANGLES and QUADS
          Only the first value is meaningful and it specifies the number of vertices to be used for rendering. Specifically:
          • For a list of points: numVertices should be the number of points to be drawn.
          • For a list of lines: numVertices should be num_lines * 2 where num_lines is the number of lines to be drawn.
          • For a list of quadrangles: numVertices should be num_quads * 4 where num_quads is the number of quadrangles to be drawn.
          • For a list of triangles: numVertices should be num_tri * 3 where num_tri is the number of triangles to be drawn.
        • For all other types:
          The number of values in this field specifies the number of primitives that will be drawn.
          Each value in the field specifies the number of vertices (or indices if given) to be used for each primitive.
      • vertexComponentsCount

        public final SoSFShort vertexComponentsCount
        Number of components in each vertex. Default is 3 (i.e. X, Y and Z).
      • vertexStride

        public final SoSFShort vertexStride
        Stride in bytes between the first component of two consecutive vertices.
        Default is 0. e.g: If the vertices are composed of 3 float components the stride should be 3 * sizeof(float). If RGB colors are packed in the same buffer the stride should be 3 * sizeof(float) + 3 * sizeof(float), the second part stands for the extra data padding.

        Note: When the values are packed (only vertices in the buffer) the value 0 can be used and OpenGL will compute the stride value.

      • vertexOffset

        public final SoSFInt32 vertexOffset
        Offset in bytes to the first vertex within the buffer. Default is 0.
      • normalStride

        public final SoSFShort normalStride
        Stride in bytes between the first component of two consecutive normals. Default is 0. e.g: If the normals are composed of 3 float components the stride should be 3 * sizeof(float). If RGB colors are packed in the same buffer the stride should be 3 * sizeof(float) + 3 * sizeof(float), the second part stands for the extra data padding.

        Note: When the values are packed (only normals in the buffer) the value 0 can be used and OpenGL will compute the stride value.

      • normalOffset

        public final SoSFInt32 normalOffset
        Offset in bytes to the first normal vector in the buffer. Default is 0.
      • indexOffset

        public final SoSFInt32 indexOffset
        Offset in bytes to the first index in the buffer. Default is 0.
      • colorBuffer

        public final SoSFBufferObject colorBuffer
        Buffer object that contains the (optional) color values. Default is no buffer. Colors are always per-vertex or per-vertex-indexed. Note: This buffer must be an SoCpuBufferObject or SoGpuBufferObject. (SoGLBufferObject with target = ARRAY_BUFFER is also allowed but not recommended.)
      • colorStride

        public final SoSFInt32 colorStride
        Stride in bytes between the first component of two consecutive colors.
        Default is 0. e.g: If the colors are composed of 3 float components the stride should be 3 * sizeof(float). If vertices are packed in the same buffer the stride should be 3 * sizeof(float) + 3 * sizeof(float), the second part stands for the extra data padding.

        Note: When the values are packed (only color values in the buffer) the value 0 can be used and OpenGL will compute the stride value.

      • colorOffset

        public final SoSFInt32 colorOffset
        Offset in bytes to the first color value in the buffer. Default is 0.
      • colorComponentsCount

        public final SoSFInt32 colorComponentsCount
        Number of components in each color value. Default is 3 (i.e. red, green and blue)
      • texCoordsStride

        public final SoMFInt32 texCoordsStride
        Stride in bytes between the first component of two consecutive texture coordinates.
        Default is 0. e.g: If each element is composed of 2 float components the stride should be 2 * sizeof(float). If vertices are packed in the same buffer the stride should be 2 * sizeof(float) + 3 * sizeof(float), the second part stands for the extra data padding.

        Note: When the values are packed (only texture coordinates in the buffer) the value 0 can be used and OpenGL will compute the stride value.

      • texCoordsOffset

        public final SoMFInt32 texCoordsOffset
        Offset in bytes to the first texture coordinate in the buffer. Default is 0.
      • texCoordsComponentsCount

        public final SoMFInt32 texCoordsComponentsCount
        Number of components in each texture coordinate. Default is 2 (i.e. S and T)
    • Constructor Detail

      • SoBufferedShape

        public SoBufferedShape()
        Default constructor.