Ctrl+c/v hotkeys have been enabled for the graph mode of the channel editor. They do the same as the C/P buttons in the upper-right corner. ie. copy the values at the current time and paste the previously copied values at the current time.

Ctrl+c/v hotkeys have been enabled for the graph mode of the channel editor. They do the same as the C/P buttons in the upper-right corner. ie. copy the values at the current time and paste the previously copied values at the current time.

Ctrl+c/v hotkeys have been enabled for the graph mode of the channel editor. They do the same as the C/P buttons in the upper-right corner. ie. copy the values at the current time and paste the previously copied values at the current time.

Ctrl+c/v hotkeys have been enabled for the graph mode of the channel editor. They do the same as the C/P buttons in the upper-right corner. ie. copy the values at the current time and paste the previously copied values at the current time.

Ctrl+c/v hotkeys have been enabled for the graph mode of the channel editor. They do the same as the C/P buttons in the upper-right corner. ie. copy the values at the current time and paste the previously copied values at the current time.

The Shader SOP now can have an arbitrary number of shaders applied in the single op. It also uses the new dynamic parm method of reference rather than the old shop path, allowing it to use relative references to SHOPs. Rather than default the group to an expression which reads the surface group, a toggle is now provided to override the group field with the surface group's field.

The Shader SOP now can have an arbitrary number of shaders applied in the single op. It also uses the new dynamic parm method of reference rather than the old shop path, allowing it to use relative references to SHOPs. Rather than default the group to an expression which reads the surface group, a toggle is now provided to override the group field with the surface group's field.

The Shader SOP now can have an arbitrary number of shaders applied in the single op. It also uses the new dynamic parm method of reference rather than the old shop path, allowing it to use relative references to SHOPs. Rather than default the group to an expression which reads the surface group, a toggle is now provided to override the group field with the surface group's field.

The Shader SOP now can have an arbitrary number of shaders applied in the single op. It also uses the new dynamic parm method of reference rather than the old shop path, allowing it to use relative references to SHOPs. Rather than default the group to an expression which reads the surface group, a toggle is now provided to override the group field with the surface group's field.

The Shader SOP now can have an arbitrary number of shaders applied in the single op. It also uses the new dynamic parm method of reference rather than the old shop path, allowing it to use relative references to SHOPs. Rather than default the group to an expression which reads the surface group, a toggle is now provided to override the group field with the surface group's field.

The File Still COP has been removed, as File COPs now interpret non-animated filename expressions as stills. Hip files with File Still COPs in them will be auto-converted to use File COPs on load.

The File Still COP has been removed, as File COPs now interpret non-animated filename expressions as stills. Hip files with File Still COPs in them will be auto-converted to use File COPs on load.

The File Still COP has been removed, as File COPs now interpret non-animated filename expressions as stills. Hip files with File Still COPs in them will be auto-converted to use File COPs on load.

The File Still COP has been removed, as File COPs now interpret non-animated filename expressions as stills. Hip files with File Still COPs in them will be auto-converted to use File COPs on load.

The File Still COP has been removed, as File COPs now interpret non-animated filename expressions as stills. Hip files with File Still COPs in them will be auto-converted to use File COPs on load.

These are just some notes on procedural geometry generators in mantra. Mantra can currently get geometry to render in two (or three) ways:

• The geometry can be read in directly (whether in-lined inthe IFD or from a disk file)

• The geometry can be read in from a disk fileconditionally. That is, if a ray intersects the boundingbox specified for the geometry, then the geometry isloaded on demand.

• Geometry can be generated on the fly using a proceduralobject.

These two mechanisms have different properties inside the rendering engine and these notes are intended to help identify the advantages and short-comings of each approach.

• When geometry is read in, mantra has to spend time parsing thegeometry file. The geometry uses memory up. Most of theseresources are wasted if the geometry isn't ever rendered (i.e.it's occluded or off screen).

However, a single piece of geometry can be instanced multiple times resulting in potentially huge memory savings. Savings are lessened if:

• The geometry has a lot of NURBs/Bezier surfaces or curves
• The geometry is displacement mapped

• The first method can be improved by referring to an externaldisk file and specifying a bounding box for the geometry. Inthis case, the cost of parsing of the geometry and memory useis only incurred if the geometry is actually rendered.

As with the first case, mantra is able to re-use the geometry for multiple instances resulting in much better memory performance.

• Procedural geometry is handled slightly differently in mantrawith regards to instancing. Procedurals cannot currently beshared between instances. Thus, procedurals can consume a lotmore memory than referencing disk files.

However, procedurals are able to generate geometry on the fly. This means that they can generate geometry suitable for the frame being rendered. For example, a procedural tree might generate open polygon curves if the tree is very small in screen space, but generate highly detailed tubes if the tree is in very close to the camera. The procedural also has the ability to change shaders based on the level of detail required for rendering.

So in short, there are advantages to using procedurals, but at the cost of a potentially larger memory footprint for renderings. Especially where multiple instances are being rendered.

These are just some notes on procedural geometry generators in mantra. Mantra can currently get geometry to render in two (or three) ways:

• The geometry can be read in directly (whether in-lined inthe IFD or from a disk file)

• The geometry can be read in from a disk fileconditionally. That is, if a ray intersects the boundingbox specified for the geometry, then the geometry isloaded on demand.

• Geometry can be generated on the fly using a proceduralobject.

These two mechanisms have different properties inside the rendering engine and these notes are intended to help identify the advantages and short-comings of each approach.

• When geometry is read in, mantra has to spend time parsing thegeometry file. The geometry uses memory up. Most of theseresources are wasted if the geometry isn't ever rendered (i.e.it's occluded or off screen).

However, a single piece of geometry can be instanced multiple times resulting in potentially huge memory savings. Savings are lessened if:

• The geometry has a lot of NURBs/Bezier surfaces or curves
• The geometry is displacement mapped

• The first method can be improved by referring to an externaldisk file and specifying a bounding box for the geometry. Inthis case, the cost of parsing of the geometry and memory useis only incurred if the geometry is actually rendered.

As with the first case, mantra is able to re-use the geometry for multiple instances resulting in much better memory performance.

• Procedural geometry is handled slightly differently in mantrawith regards to instancing. Procedurals cannot currently beshared between instances. Thus, procedurals can consume a lotmore memory than referencing disk files.

However, procedurals are able to generate geometry on the fly. This means that they can generate geometry suitable for the frame being rendered. For example, a procedural tree might generate open polygon curves if the tree is very small in screen space, but generate highly detailed tubes if the tree is in very close to the camera. The procedural also has the ability to change shaders based on the level of detail required for rendering.

So in short, there are advantages to using procedurals, but at the cost of a potentially larger memory footprint for renderings. Especially where multiple instances are being rendered.

These are just some notes on procedural geometry generators in mantra. Mantra can currently get geometry to render in two (or three) ways:

• The geometry can be read in directly (whether in-lined inthe IFD or from a disk file)

• The geometry can be read in from a disk fileconditionally. That is, if a ray intersects the boundingbox specified for the geometry, then the geometry isloaded on demand.

• Geometry can be generated on the fly using a proceduralobject.

These two mechanisms have different properties inside the rendering engine and these notes are intended to help identify the advantages and short-comings of each approach.

• When geometry is read in, mantra has to spend time parsing thegeometry file. The geometry uses memory up. Most of theseresources are wasted if the geometry isn't ever rendered (i.e.it's occluded or off screen).

However, a single piece of geometry can be instanced multiple times resulting in potentially huge memory savings. Savings are lessened if:

• The geometry has a lot of NURBs/Bezier surfaces or curves
• The geometry is displacement mapped

• The first method can be improved by referring to an externaldisk file and specifying a bounding box for the geometry. Inthis case, the cost of parsing of the geometry and memory useis only incurred if the geometry is actually rendered.

As with the first case, mantra is able to re-use the geometry for multiple instances resulting in much better memory performance.

• Procedural geometry is handled slightly differently in mantrawith regards to instancing. Procedurals cannot currently beshared between instances. Thus, procedurals can consume a lotmore memory than referencing disk files.

However, procedurals are able to generate geometry on the fly. This means that they can generate geometry suitable for the frame being rendered. For example, a procedural tree might generate open polygon curves if the tree is very small in screen space, but generate highly detailed tubes if the tree is in very close to the camera. The procedural also has the ability to change shaders based on the level of detail required for rendering.

So in short, there are advantages to using procedurals, but at the cost of a potentially larger memory footprint for renderings. Especially where multiple instances are being rendered.

These are just some notes on procedural geometry generators in mantra. Mantra can currently get geometry to render in two (or three) ways:

• The geometry can be read in directly (whether in-lined inthe IFD or from a disk file)

• The geometry can be read in from a disk fileconditionally. That is, if a ray intersects the boundingbox specified for the geometry, then the geometry isloaded on demand.

• Geometry can be generated on the fly using a proceduralobject.

These two mechanisms have different properties inside the rendering engine and these notes are intended to help identify the advantages and short-comings of each approach.

• When geometry is read in, mantra has to spend time parsing thegeometry file. The geometry uses memory up. Most of theseresources are wasted if the geometry isn't ever rendered (i.e.it's occluded or off screen).

However, a single piece of geometry can be instanced multiple times resulting in potentially huge memory savings. Savings are lessened if:

• The geometry has a lot of NURBs/Bezier surfaces or curves
• The geometry is displacement mapped

• The first method can be improved by referring to an externaldisk file and specifying a bounding box for the geometry. Inthis case, the cost of parsing of the geometry and memory useis only incurred if the geometry is actually rendered.

As with the first case, mantra is able to re-use the geometry for multiple instances resulting in much better memory performance.

• Procedural geometry is handled slightly differently in mantrawith regards to instancing. Procedurals cannot currently beshared between instances. Thus, procedurals can consume a lotmore memory than referencing disk files.

However, procedurals are able to generate geometry on the fly. This means that they can generate geometry suitable for the frame being rendered. For example, a procedural tree might generate open polygon curves if the tree is very small in screen space, but generate highly detailed tubes if the tree is in very close to the camera. The procedural also has the ability to change shaders based on the level of detail required for rendering.

So in short, there are advantages to using procedurals, but at the cost of a potentially larger memory footprint for renderings. Especially where multiple instances are being rendered.

These are just some notes on procedural geometry generators in mantra. Mantra can currently get geometry to render in two (or three) ways:

• The geometry can be read in directly (whether in-lined inthe IFD or from a disk file)

• The geometry can be read in from a disk fileconditionally. That is, if a ray intersects the boundingbox specified for the geometry, then the geometry isloaded on demand.

• Geometry can be generated on the fly using a proceduralobject.

These two mechanisms have different properties inside the rendering engine and these notes are intended to help identify the advantages and short-comings of each approach.

• When geometry is read in, mantra has to spend time parsing thegeometry file. The geometry uses memory up. Most of theseresources are wasted if the geometry isn't ever rendered (i.e.it's occluded or off screen).

However, a single piece of geometry can be instanced multiple times resulting in potentially huge memory savings. Savings are lessened if:

• The geometry has a lot of NURBs/Bezier surfaces or curves
• The geometry is displacement mapped

• The first method can be improved by referring to an externaldisk file and specifying a bounding box for the geometry. Inthis case, the cost of parsing of the geometry and memory useis only incurred if the geometry is actually rendered.

As with the first case, mantra is able to re-use the geometry for multiple instances resulting in much better memory performance.

• Procedural geometry is handled slightly differently in mantrawith regards to instancing. Procedurals cannot currently beshared between instances. Thus, procedurals can consume a lotmore memory than referencing disk files.

However, procedurals are able to generate geometry on the fly. This means that they can generate geometry suitable for the frame being rendered. For example, a procedural tree might generate open polygon curves if the tree is very small in screen space, but generate highly detailed tubes if the tree is in very close to the camera. The procedural also has the ability to change shaders based on the level of detail required for rendering.

So in short, there are advantages to using procedurals, but at the cost of a potentially larger memory footprint for renderings. Especially where multiple instances are being rendered.