Nodes Pyro

Flexible, production-quality fire and smoke shader.

Overview

The Pyro material’s interface may appear intimidating at first. However, it’s actually quite straightforward once you grasp how it’s organized. There are three main optical components to the material: smoke, fire, and scattering, corresponding to the first three tabs of the interface.

Wherever the material allows you to specify a color, it is split into two complementary parameters: the chromaticity of the color (its hue and saturation), and its brightness (also known as value, luminance, or intensity). More often than not, you will want to manipulate a color’s luminance separately from its hue or saturation. Especially in the Pyro material, you will often be dealing with very bright or very dark values, where the standard RGB swatch stops giving meaningful feedback. However, if you prefer, you can still ignore the separate brightness control (by leaving it at 1) and use the RGB swatch to express both a color (hue and saturation) and its intensity.

In addition, in places where you can specify colors, the interface includes “mod” parameters (the menus throughout the interface that default to “No mods”). You can completely ignore these controls until you get a feel for the main parameters, as the Pyro material does not depend on them. See modifiers for more information.

Choosing fields

The pop-up menus allowing you to choose fields contain several different types of fields:

Standard fields: A fixed list of fields that typically exist on the shaded object: density, temperature, heat, burn, and age.
Shader fields: fields generated internally by the shader. The speed field contains the magnitude of the vector-valued velocity field. The time field simply contains the current time in seconds.
Mod fields: these correspond to the Field1, Field2, Noise1, Noise2, etc. sub-tabs on the Modifiers tab when they are activated. See modifiers for more information. Use the Label parameter on the modifier to control how it appears in the menu.
User fields: these let you use other fields besides the standard names. First, in the geometry network, use a Name surface node to rename the field to one of user1, user2, etc. Then in the material, enable the field in the Settings / User fields tab. Then you can choose the field in the pop-up menu.

For example, you could assign the temperature field to be the smoke field by setting Density field to “temperature”.

Using scattering

The Pyro material can simulate scattering of light from the fire through smoke (to a rough, single scattering event approximation), that is, you can have the fire light up the smoke surrounding it. The material does this using point clouds.

Before rendering the final image, you need to have the Pyro material scatter points inside the smoke and fire volumes. The resulting point clouds are stored in two separate files for each frame, one for fire and one for smoke. Then, during shading, the Pryo material uses the point cloud files to simulate the emission of light from the fire field and the scattering of the light through the smoke field.

The Pyro material uses the fire point cloud to fill the fire volume with point light sources, with the intensity and color of each light derived from the parameters on the Fire tab. When calculating illumination of the smoke from fire emission, the material will loop over some subset of these fire point lights and accumulate their contribution just as it does for external lights. The amount of illumination contributed by each one of the fire points is subject to the same attenuation factors as the external sources: distance, the smoke’s own density, as well as any occluding objects embedded in the smoke. Each point in the smoke point cloud stores the resulting fire illumination. Then, while marching through the volume to shade it, the material will use a neighborhood of these precomputed fire illumination results, average them together, and use them in the final shading calculation.

See the scattering parameter help for more information.

How to generate scattering point clouds

The simplest way to generate point clouds is to render them from the Pyro material node’s parameter interface. This lets you generate all point clouds for a frame or animation at once, and re-generate as needed.

Advanced users may want to set up individual render output nodes to generate point clouds. This lets you generate point clouds for each frame before the final rendered image, and make Pyro shader point cloud generation part of a render dependency graph.

In general, the Volume Step Size used for generating the Pyro material’s point clouds can be a lot coarser that the step size for the final image. As a rough starting guideline, start with a step size around ten times coarser than the final render’s volume step size. For example, if the final render requires a Volume Step Size of 0.01, then you can set the step size for the scattering point clouds at 0.1 as a starting point, and refine based on look.

Depending on the relative smoke and fire content of the effect, you may want to generate the smoke point cloud with a different density than the fire point cloud (usually for efficiency, since a sparse cloud is faster to compute than a dense cloud). The instructions below include how to do this for the two methods of generating point clouds.

Tip

The scattered illumination (and attenuation) is based on the smoke density and fire intensity when the point clouds were saved. If you change any parameters that affect the smoke density or fire color or intensity, you must regenerate the point clouds. Changes that affect smoke color, shadow, or phase do not require regenerating the point clouds.

Tip

Point cloud files (extension .pc) are really geometry files (like .bgeo or .geo files). So, if you and want to visualize the point clouds, you can load the files into a Houdini using the File surface node.

To... Do this

To...	Do this
Set the filenames for point cloud files	In the Pyro material’s parameters, click the Scattering tab, then click then Point cloud files sub-tab. Set the Smoke PC file and Fire PC file parameters. Remember to include the frame number (e.g. `$F4`) in the filename if you are doing animation.
Generate point clouds from the Pyro material node	In the Pyro material’s parameters, click the Scattering tab, then click then Generate point clouds sub-tab. If you want the same step size for fire and smoke, set the Smoke step size and then set Generate mode to “Use smoke step for both”. If you want separate step sizes for smoke and fire, set Smoke step size and Fire step size, then set Generate mode to “Use individual steps”. See the parameter help for more information. If you are generating point clouds for animation, Turn on Frame range and set the start and end frame. Click Generate.
Generate point clouds as part of a render network (advanced)	The trick is to create separate render nodes for generating point clouds, where each render node uses a take in which the Pyro shader’s Shading mode is set to a different value. Create render nodes for writing out the point clouds. If you want the same step size for fire and smoke point clouds, create a single render node for writing out point clouds. If you want separate step sizes for smoke and fire, create two render nodes and set their Volume step size to different values for generating fire and smoke point clouds. Set the Properties / Render / Rendering engine parameter to either “Micropolygon Rendering” or “Micropolygon Physically Based Rendering”. Micropolygon renders will generate different point clouds than raytracing, and Pyro can compute scattering much more efficiently with the type of clouds generated by the micropolygon engines. Rename the render nodes to indicate the type(s) of point clouds they generate. Tip You can create three mantra nodes (for writing out the smoke cloud, fire cloud, or both) from the Pyro material’s parameter interface. Go to the Scattering / Generate point clouds sub-tab. Set Generate mode to “Create ROPs only” and click Generate. This will create three render output nodes in `/out`. This is useful when you want to control some aspect of the render nodes beyond what the Pyro interface exposes, but duplicate the exact settings the Pyro material uses otherwise. Warning Because the point clouds are rendered from the point of view of the camera selected in the render node, it is important that the entire volume be visible from that camera. If the frustum of the main render camera clips any part of the visible smoke/fire, then no points will be generated for that portion of the effect and you may get shading artifacts, such as dark shadows suddenly popping in and out of the scattered emission. You may need to create a separate camera for the point cloud render node(s). The point of view of the point cloud camera should be as close as possible to the final render camera’s, but with the entire effect (smoke and fire volumes) visible. The point cloud camera need not match the optical settings of the main render camera. Things like resolution, focal length, aperture, and so on, can all be different. Set up separate takes where the Pyro shader node has Shading mode on the Settings tab set to different values: Create a take for the final image in which Shading mode is set to “Normal shading”. If you want the same step size for fire and smoke point clouds, create another take in which Shading mode is set to “Write all point clouds”. If you want separate step sizes for smoke and fire, create two more takes: one in which the Shading mode is “Write smoke point cloud” and another in which Shading mode is “Write fire point cloud”. Rename the takes to indicate which Shading mode setting they contain. For each of the point render nodes you created, change its Render with take parameter to one of the takes containing different Shading mode settings. Once you have the render nodes set up, you can use the Render button on each node to re-generate point clouds, or wire them into a render dependency network to make generating point clouds part of a pipeline.

Set the filenames for point cloud files

In the Pyro material’s parameters, click the Scattering tab, then click then Point cloud files sub-tab.
Set the Smoke PC file and Fire PC file parameters. Remember to include the frame number (e.g. $F4) in the filename if you are doing animation.

Generate point clouds from the Pyro material node

In the Pyro material’s parameters, click the Scattering tab, then click then Generate point clouds sub-tab.
If you want the same step size for fire and smoke, set the Smoke step size and then set Generate mode to “Use smoke step for both”. If you want separate step sizes for smoke and fire, set Smoke step size and Fire step size, then set Generate mode to “Use individual steps”. See the parameter help for more information.
If you are generating point clouds for animation, Turn on Frame range and set the start and end frame.
Click Generate.

Generate point clouds as part of a render network (advanced)

The trick is to create separate render nodes for generating point clouds, where each render node uses a take in which the Pyro shader’s Shading mode is set to a different value.

Create render nodes for writing out the point clouds.
- If you want the same step size for fire and smoke point clouds, create a single render node for writing out point clouds.
- If you want separate step sizes for smoke and fire, create two render nodes and set their Volume step size to different values for generating fire and smoke point clouds.
- Set the Properties / Render / Rendering engine parameter to either “Micropolygon Rendering” or “Micropolygon Physically Based Rendering”. Micropolygon renders will generate different point clouds than raytracing, and Pyro can compute scattering much more efficiently with the type of clouds generated by the micropolygon engines.
- Rename the render nodes to indicate the type(s) of point clouds they generate.
Tip

You can create three mantra nodes (for writing out the smoke cloud, fire cloud, or both) from the Pyro material’s parameter interface. Go to the Scattering / Generate point clouds sub-tab. Set Generate mode to “Create ROPs only” and click Generate. This will create three render output nodes in /out. This is useful when you want to control some aspect of the render nodes beyond what the Pyro interface exposes, but duplicate the exact settings the Pyro material uses otherwise.

Warning

Because the point clouds are rendered from the point of view of the camera selected in the render node, it is important that the entire volume be visible from that camera. If the frustum of the main render camera clips any part of the visible smoke/fire, then no points will be generated for that portion of the effect and you may get shading artifacts, such as dark shadows suddenly popping in and out of the scattered emission.

You may need to create a separate camera for the point cloud render node(s). The point of view of the point cloud camera should be as close as possible to the final render camera’s, but with the entire effect (smoke and fire volumes) visible. The point cloud camera need not match the optical settings of the main render camera. Things like resolution, focal length, aperture, and so on, can all be different.
Set up separate takes where the Pyro shader node has Shading mode on the Settings tab set to different values:
- Create a take for the final image in which Shading mode is set to “Normal shading”.
- If you want the same step size for fire and smoke point clouds, create another take in which Shading mode is set to “Write all point clouds”.
- If you want separate step sizes for smoke and fire, create two more takes: one in which the Shading mode is “Write smoke point cloud” and another in which Shading mode is “Write fire point cloud”.
Rename the takes to indicate which Shading mode setting they contain.
For each of the point render nodes you created, change its Render with take parameter to one of the takes containing different Shading mode settings.

Once you have the render nodes set up, you can use the Render button on each node to re-generate point clouds, or wire them into a render dependency network to make generating point clouds part of a pipeline.

Using modifier fields

Tip

You can ignore the modifiers and the material will still work perfectly and be very useful. However, the modifiers are available when you need additional flexibility.

Modifier overview

The material’s main parameters let you shading existing volume fields, but often those fields are not shaped exactly right to shade them as you want. You don’t want to have to recalculate an entire simulation just to adjust some field slightly, or generate a new field which is just a simple transformation of an existing field. Modifiers let you do useful field manipulations at shading time:

Field Shaping	The falloff at the edge of a field (its contour) my be too soft and requires sharpening, or it’s too sharp and needs to be softened. This kind of “softening” can also be the result of natural diffusion within the simulation and the correction may need to be applied gradually over time. The field’s range may drift too much throughout the simulation, or it may simply be too large or too small to work with comfortably in the context of shading. For example, a field which we want to use as a ramp lookup may have a sim-generated range which is much smaller (or much larger) than what you can comfortably work with using the ramp interface. Such fields need to be remapped to a different range. A field (`temperature` for example) may be unbounded and need to be clamped before you can work with it.
High Frequency Detailing	Increasing the volume resolution of a simulation is not always practical because of slower processing speed. In the context of smoke and fire, small-scale detailing usually means the addition of layers of noise. The material can add high resolution noise at shading time for a fraction of the cost of actually simulating it. The catch is you don’t know beforehand where this noise-based detailing may be needed. As a result, the shader-based noise needs to be assignable to not only the incoming fields themselves (to assist with shaping), but also to the main parameters
Field-Driven Parameterization	For example, you want the smoke to look darker when it’s close to the flames, but lighter as it “cools off”, so you need to drive the smoke’s Brightness parameter based on some transformation of the `temperature` field. It’s now a magical smoke which needs to be smooth when moving fast, but have lots of noise detail when drifting normally or moving very little. So now you need to drive one (or possibly several) parameters of the noise based on several similar (but not identical) variations on the smoke’s speed. Wanting the incoming fields (or variations of them) and/or the shader-generated noises to drive the material’s main parameters is very common in production.

To support this wide variety of field modifications and noise generation, Pyro provides a bank of “Modifiers”. There are six field modifiers which can be used to shape or remap incoming fields, and four noise generators which can also optionally be driven by either the incoming fields or one of the six field modifiers. Any one of these modifiers, once activated, will be available to drive any one (or several) of the principal parameters which supports modification (any parameter which is accompanied by a “Mod” pop-up menu).

Modifier organization

The Pyro material only works with scalar (single floating point number) fields.
Field modifiers operate on simulation fields. Their output is yet another scalar field which represents a modified version of the base field.
Noise generators do not necessarily depend on any field, though they can optionally be constrained to exist within a field (either a simulation field or the output of one of the field modifiers). You can assign their output to any mod-able parameter, except the (very few) parameters which strictly expect a field (such as the Density Field parameter on the Smoke tab).
Field modifiers and noise generators normally produce scalar output. However, you can use the scalar output as to look up a value in a color/vector ramp to produce vector/color output from a modifier. Currently there are only a few mod-able vector parameters, all colors (such as the Smoke Color parameter).

How to set up modifiers

The Pyro material has two kinds of modifiers: field modifiers and noise modifiers. Field modifiers let you modify the values of a field from the sim before you use them in the shader. Noise modifiers let you add noise to a field or all fields before you use it/them in the shader.

To... Do this

To...	Do this
Modify the smoke and fire fields	By default, the source fields on the Smoke and Fire tabs are set to use fields from the simulation named `density` and `temperature` directly. Click the Modifiers tab, then click one of the Field sub-tabs. Turn on Active field modifier and give the modifier a descriptive name in the Label text box. Set the Base field to the field you want to modify, such as `density` for smoke. Use the controls on the Field sub-tab to modify the incoming field. For example, you can use Range mapping and/or the Interpolation ramp to remap the source smoke density. Go back to the Smoke or Fire tab and change the source field to use the modifier field instead of the actual source field. Tip Instead of using actual fields (such as `density` and `temperature`) for your smoke and fire fields, assign those fields to modifier fields and use the modifier fields instead. Usually when your sim values come into the material, the raw flame is a bit high, not high enough, or doesn’t have the right weight. By remapping the incoming range you can specify at what density the flame starts appearing and how it is distributed into the 0-1 range. Using the ramp you can then redistribute the intensity. This is very useful for both fire and smoke.
Modify a value using a field	Most parameters in the material have a pop-up menu to let you choose a modifier field as a multiplier for the parameter’s value. For example, you can use the Density amplitude parameter on the Smoke field with a modifier to remap smoke density. Set it to a high number (e.g. `10`), then use the modifier field’s ramp to control the amplitude.
Have fire light up the surrounding smoke	See the Scattering parameters.

Modify the smoke and fire fields

By default, the source fields on the Smoke and Fire tabs are set to use fields from the simulation named density and temperature directly.
Click the Modifiers tab, then click one of the Field sub-tabs. Turn on Active field modifier and give the modifier a descriptive name in the Label text box.
Set the Base field to the field you want to modify, such as density for smoke.
Use the controls on the Field sub-tab to modify the incoming field. For example, you can use Range mapping and/or the Interpolation ramp to remap the source smoke density.
Go back to the Smoke or Fire tab and change the source field to use the modifier field instead of the actual source field.

Tip

Instead of using actual fields (such as density and temperature) for your smoke and fire fields, assign those fields to modifier fields and use the modifier fields instead.

Usually when your sim values come into the material, the raw flame is a bit high, not high enough, or doesn’t have the right weight. By remapping the incoming range you can specify at what density the flame starts appearing and how it is distributed into the 0-1 range. Using the ramp you can then redistribute the intensity. This is very useful for both fire and smoke.

Modify a value using a field

Most parameters in the material have a pop-up menu to let you choose a modifier field as a multiplier for the parameter’s value.

For example, you can use the Density amplitude parameter on the Smoke field with a modifier to remap smoke density. Set it to a high number (e.g. 10), then use the modifier field’s ramp to control the amplitude.

Have fire light up the surrounding smoke

See the Scattering parameters.

Important parameters

Important smoke parameters:

Density amplitude	Very important to achieve both wispy or billowy smoke.
Smoke brightness	Drastically changes brightness/exposure.
Shadow density	Very important to get a lighter smoke without blowing out the brightness. Less contrasty smoke.
Scattering Phase	Controls how deep light penetrates into the smoke. Small changes to this make a big difference.

Important fire parameters:

Fire Intensity	Especially important when scattering.
Noise Modifiers	Try adding a noise modifier with the default settings: tons of noise.
Temperature	Increasing this will create a hotter, whiter flame.
Red Boost	Increasing this will create a hotter whiter flame, without affecting the temperature values.

Important scattering parameters:

Fire Intensity	A high fire intensity will really increase the amount of light emitted.
Distance Attenuation	How far the light penetrates the point cloud.
Density Attenuation	How much the light penetrates the point cloud through dense areas.

Density amplitude and volume step size

Fields generated by a simulation rarely have values as high as 1 inside a voxel. The Pyro shader’s default density scaling of 500 was chosen to push the typically low-valued simulation densities toward a very dense, sooty smoke by default.

This default is too high for if you begin with a very dense field. The result can be a very noisy image. The problem is the field is so dense that it’s becoming solid, and the boundary between the “air” that surrounds a solid and its surface is infinitesimally small: there’s nothing for a long time and then, suddenly, there is a surface. For the renderer to catch this sudden transition from “nothing” to “something”, it needs to take very small steps, otherwise it might overshoot the surface boundary and miss it by a small amount.

As mantra takes relatively large steps through space scanning for “density”, it overshoots the boundary by differing amounts each time and determines that the boundary happened slightly earlier or later in each case. Therefore, some of these approximated boundary positions end up receiving shadows and some don’t, resulting in an noisy image.

You can try several solutions to this problem, each with a different impact on render times and visual quality:

Remove shadows from the light. This gets rid of the problem, but is not very useful since you usually want shadows in your smoke.
Lower the Pyro material’s Density amplitude until it becomes “soft” enough that the current Volume Step Size setting in the mantra node has no problem catching it.
Decrease the Volume Step Size setting in the mantra node until it is fine enough to catch our (unusually solid) boundary.

The most common solution is a combination of choices #2 and #3: lower the density as you shorten the step until you find a place that is acceptable in terms of render time and image quality.

Parameters

Smoke

See how to use modifiers for information about the “mod” parameters. See color correction for information about the color correction parameters.

Enable smoke visibility

Turns the visibility of smoke on or off. Note that this does not control whether the smoke calculations occur, just whether to display their output. This is useful for hiding smoke you can see other things, such as fire, more clearly.

Density field

The name of the field that represents the smoke density. All volume/fluid simulations will generate at lease one field named density, and for pyrotechnic effects, you can usually use this to represent smoke or some byproduct of combustion.

Density amplitude

A scaling factor for the density values in the field. Smoke fields generated by a simulation rarely have values as high as 1 inside a voxel. The Pyro material’s default density scale of 500 is designed to push typically low simulation densities toward a very dense, sooty smoke.

The default value may be too high if you already begin with a dense field. The result will sometimes be a very noisy image.

(image)

See density amplitude and volume step size above for more information.

Use point color (tint with Cd)

When this option is on, the material will use the Cd (diffuse color) attribute on the volume to tint the color you set in the Smoke color parameter. When this option is off, the material ignores the volume’s color and only uses the Smoke color parameter.

(Technically, it’s a “primitive” color, not a point color, since volumes are primitives and don’t have points, but surface shaders use “point color”, so the Pyro material uses it for the sake of familiarity.)

Smoke color

Modifies the reflectance color of the smoke. Note that if Use point color (tint with Cd) is on, this color will be tinted by the color of the volume geometry.

You can use the separate Smoke brightness parameter to control the brightness. This is often more useful, since you will be dealing with very bright or very dark values, where the standard RGB swatch stops giving meaningful feedback. However, if you prefer, you can still ignore the separate brightness control (by leaving it at 1) and use the RGB swatch to express both a color (hue and saturation) and its intensity.

Smoke brightness

Modifies the brightness of the smoke.

Shadow color

Modifies the color of shadows in the smoke.

You can use the separate Shadow density parameter to control the shadow intensity. This is often more useful, since you will be dealing with very bright or very dark values, where the standard RGB swatch stops giving meaningful feedback. However, if you prefer, you can still ignore the separate shadow density control (by leaving it at 1) and use the RGB swatch to express both a color (hue and saturation) and its intensity.

Shadow density

Modifies the intensity of smoke shadows. This is the complementary concept to “brightness” or “albedo” in the Smoke brightness parameter above.

Scattering phase

Note

Scattering Phase is currently not supported for PBR rendering. It is only available for raytracing and micropolygon rendering.

By default, the Pyro material assumes that light scatters evenly in all directions (isotropic scattering). In reality, depending on the type of particles suspended in the volume, the direction of light scattering may show a bias. It might, for example, scatter more often in the same direction as the incoming light (“forward scattering”) or in the opposite direction of the incoming light (“backward scattering” or “retro-reflection”).

The Pyro material simulates this behavior (to a rough, single scattering event approximation). When Scatter phase is 0, it creates isotropic scattering. Positive values of Scatter phase (from just above 0, to 1) create increasing degrees of forward scattering. Negative values (from just below 0, to -1) create increasing degrees of backward scattering.

This parameter has three components to give you separate control over scattering of the Red, Green, and Blue channels. You could, for example, let the blue channel scatter slightly more forward than the red or green channels. If you choose to modify them separately, you should be aware that you're introducing direction-based coloring to the volume, meaning that the color of the smoke depends on both the viewing direction and the direction of the incoming illumination.

	Backlit side of a complex smoke cloud, isotropic scattering.
	With 50% forward scattering, the light travels further into the cloud. This would be appropriate if, for example, the “smoke” were composed of water vapor instead of “sooty” carbon particles.
	Here the blue channel scatters slightly more forward than the other two (RGB = 0.49, 0.5, 0.52), resulting in a subtly blue-tinted cloud (which on the lit side would appear slightly red/yellow).

Fire

See how to use modifiers for information about the “mod” parameters. See color correction for information about the color correction parameters.

Fire is shaded as a constant color as it is considered an emitter (or an illumination source). It is disabled by default because there’s no established standard for the naming of fields which represent fire: It could be “heat” or “temperature” or “burn” or some combination of all of these, or indeed, be a “user” field. The default field is “Bound: temperature”.

Because it is an emitter, the majority of the fire tab’s interface deals with how to map a field’s value to a color and an intensity. A secondary aspect of shading fire has to do with perceived opacity.

In the real world, fire is not opaque, and so its opacity should be set to zero. This is Pyro’s default behaviour: transparent fire. However, real fire may sometimes appear opaque due to its intense luminance (or, less often, because of matter suspended in the fire which has not yet combusted). If you choose to control its apparent opacity solely by its intensity, then you have to be prepared to deal with output images which contain a very large dynamic range. In addition, if you leave fire completely transparent and your fire volume is very deep (in relative units), you may find that the intensity levels accumulating over the depth of the volume are hard to control practically. These reasons combine to sometimes create a need for “artistic” control over the perceived opacity of the fire. To this end, the Pyro material exposes some opacity controls you can use to assign an artistic (physically incorrect) opacity to the fire field.

Enable fire visibility	Turns the visibility of fire on or off. Note that this does not control whether the fire calculations occur, just whether to display their output. This is useful for hiding fire you can see other things, such as smoke, more clearly.
Temperature field	The name of the field to use for the fire’s temperature.
Fire intensity	Increases or decreases the brightness of the fire.
Opacity field	The name of a field to use for the fire’s opacity. The default “No mods” gives a physically correct fire with no opacity.
Opacity amplitude	A scaling factor for the value of the opacity field. The default `0` gives a physically correct fire with no opacity.
Opacity blend rate	The Pyro material keeps an internal mask for the extent of the flames (the volume within which they are incandescent). This control allows you to “feather” the fire opacity at the envelope boundaries of the flames. The blend rate has a nominal range of 0-1 and is a rate, meaning that the higher the number, the faster the opacity will take effect at the border of the flames, and the lower the number, the more “feathered” or gradual the opacity blend will be from the outside edge of the flame toward its interior.
Color model	There are two approaches mapping the value of the temperature field to an output color: A physically-based approach which assigns color (chromaticity) and intensity based on interpreting the field as a temperature distribution, and an artistic approach which simply maps color and intensity directly from the underlying field values. The default is the physical model.
Use point color (tint with Cd)	When this option is on, the material will use the `Cd` (diffuse color) attribute on the volume to tint the fire color. When this option is off, the material ignores the volume’s color and only uses the calculated fire color. (Technically, it’s a “primitive” color, not a point color, since volumes are primitives and don’t have points, but surface shaders use “point color”, so the Pyro material uses it for the sake of familiarity.)

Physical model

The physical model derives a color and intensity based on the current temperature (as given by the value of the temperature field, scaled by the Temperature parameter below). This model uses Planck’s formulas for both the color and intensity of a perfect emitter (also known as a “black body”) at a certain temperature, given in degrees Kelvin.

The black body model is generally a good starting point for a color spread for a flame. The drawback is it’s based on measurements of near-perfect emitters/absorbers, which doesn’t necessarily apply to flames (though it’s not far off). In the case of flames, the visible color is more likely the result of chemical reactions than black body emission. For example, the bottom of a candle flame appears blue because it’s more oxygenated than the rest of the flame (and burns more efficiently). Therefore, the Pyro material gives you controls to adjust the red/blue content of physically modeled flames.

	Fire field at 3000 degrees K, Red boost = 0, Blue boost = 0
	Red boost = 4, blue boost = 0
	Red boost = 0, blue boost = 4

Temperature	The temperature (in degrees Kelvin) corresponding to a value of `1` in the field. As a rough reference for what these temperatures mean, thermal emission becomes visible at around 500°K (e.g. the temperature at which an iron rod will start looking “red hot”). Temperatures for typical flames can range from around 2000°K (candle flame) to 3000°K (blowtorch, limelight) and higher. The Pyro material uses a fairly “hot” temperature as a default. This is only a guess based on observation of a few temperature fields generated by simulations, which don’t always reach 1. It is entirely possible that the incoming fields may go well above one, in which case you should reduce the default.
Red boost	Artificially boosts the red (colder) end of the flame’s spectrum. This parameter has a lower bound of 0, but is unbounded on the positive side. However, practical values should rarely go above 4-8.
Blue boost	Artificially boosts the blue (hotter) end of the flame’s spectrum, essentially sliding the blackbody temperature up into the extreme heat end of the spectrum. Although this operates only on the perceived chromaticity (hue + saturation), it is likely that the overall intensity of the flame will also increase because as the color goes to blue it passes through white, which is much more luminous than fully saturated red. So, you will need to adjust the fire intensity when using either of these “boost” parameters. It’s possible that if you boost the blues, the core of the flame may be too bright and be clamped too much. You can use the Soft Clip parameter to address this.
Soft clip	Also known as a “knee” operator (although this uses is a cheaper algorithm because it has to run for every shaded point). If the hot portion of the flame starts becoming so bright that it ends up clamping too much, you can use soft-clip it to some ceiling. This is used to compress high intensities that can’t be displayed by a some device (like a TV or monitor). This material already applies a tone mapping operator to the black body intensities (which have an insane dynamic range). Nevertheless, there are cases where the accumulated brightness of the fire will “blow up” beyond any displayable value. This typically happens when the fire occupies a very large volume and has zero opacity, with the result that every step along the march through the volume will simply add itself to all the previous values.
Start	The intensity value at which compression will start. Keep in mind that these are values within each step of the volume march (that is, per voxel), not the absolute accumulated value at the end of the march, so it will typically be a small value around 0.2 to 0.5.
Suppression	The amount by which to suppress the output. This is an unbounded positive. It may need to be set quite high depending on the situation, and can be thought of as an exponent. The higher it goes, the closer the result will approach a standard “clamp” at the value given in the Start parameter. It is, actually, a “soft” clamp. The Start value indicates where the clipping or clamping will start, and the Suppression parameter determines how strongly values above Start will be compressed. In practice, it is better to see how intense the flame is in context (that is, with the smoke enabled) before making any decisions on whether to clamp its intensity.

Artistic model

The horizontal axis of the ramps corresponds to field values from 0 to 1. Any values in the field lower than 0 or higher than 1 will be clamped, and so use the values at the left or right end of the ramp.

The default settings are a crude approximation of the physical model, simply as a plausible starting point for an artists to start modifying.

As with other color controls in the material, color and intensity have separate controls for practical reasons, but if you want you can embed intensity in the “hue” ramp.

Temperature to hue mapping	The color of the flame for field values from `0` to `1`.
Temperature to intensity mapping	The intensity of the flame color for field values from `0` to `1`. Note that the Fire Intensity parameter (near the top of the Fire tab) is the final, global intensity adjustment, so this ramp can be within a comfortable 0-1 range, as a “relative intensities” ramp.

Scattering

See using scattering above.

See also the Scatter phase parameter on the Smoke tab. See color correction for information about the color correction parameters.

Tip

Scattering

Whether to simulate and display scattered light emission from fire through the smoke.

Point cloud files

Smoke PC file	The cache file for the smoke scattering point cloud. Remember to include the frame number (e.g. `$F4`) in the filename if you are doing animation.
Fire PC file	The cache file for the fire scattering point cloud. Remember to include the frame number (e.g. `$F4`) in the filename if you are doing animation.

Generate point clouds

See how to generate scattering point clouds for more information.

Smoke step size

Volume step size to use when generating the smoke point cloud. The smaller this number, the more dense the point cloud. A good rule of thumb is to start this value 10 times larger than the volume step size you are using for rendering the final image, then adjust based on the look.

Fire step size

You only need to set this parameter if you want a different step size for the smoke and fire point clouds (set Generate mode to “Use individual steps”). Otherwise, set Generate mode to “Use smoke step for both” and just set the Smoke step size parameter.

Camera

The camera to use when “rendering” the point clouds. This can be the same camera you use to render the final image, but it should have the entire effect in view, since the point cloud will only be built in areas visible to the camera. You may need to create a new camera for building point clouds that has the same point of view as the main camera but pulled back to show the entire effect volume. See how to generate scattering point clouds for more information.

Frame range

Turn this on and set the frame range to generate point clouds for multiple frames of animation. If this option is off, clicking Generate will only generate point clouds for the current frame.

Generate mode

Use smoke step for both	Use the value of the Smoke step size parameter as the volume step size and generate the smoke and fire point clouds at the same time.
Use individual steps	Use the values of the Smoke step size and Fire step size parameters to generate smoke and fire point clouds with different step sizes.
Create ROPs only	Advanced: This setting won’t actually generate any point cloud files. Instead, it will create render nodes in `/out` appropriate for generating point clouds, using the current values of the Pyro material. You will still need to customize the render nodes as described under how to generate scattering point clouds.

Generate button

Click to generate the point clouds necessary for simulating scattering. (If Generate mode is “Create ROPs only”, this will create render nodes in /out, not generate point clouds).

Amplitude

A global “dimmer” dial: it controls the overall intensity of scattered light.

	Amplitude = 0.5
	Amplitude = 1.0
	Amplitude = 2.0

Tip

In the real world, density attenuation and distance attenuation are linked. In the Pyro material, they're decoupled to allow for more artistic control.

Density atten

As light travels inside a dense medium it progressively loses intensity as bits of it are absorbed or scattered away. This parameter scales this “attenuation due to density”. Values greater than 1 mean density attenuation happens faster than it would in the real world given the density, and values less than 1 mean less attenuation than in the real world.

	Density atten = 0.5
	Density atten = 1.0
	Density atten = 1.5

Distance atten

As in the real world, the light emitted by the point lights in the fire cloud attenuates with distance according to the inverse square law: the rate of attenuation is inversely proportional to the square of the distance. This parameter scales that attenuation. Values greater than 1 mean distance attenuation happens faster than it would in the real world given the distance (perceived as shortening the scattering distance), and values less than 1 mean less attenuation than in the real world (perceived as increasing the scattering distance).

Scaling distance attenuation retains the overall shape/distribution of the scattering, but scales the perceived length/size up or down (whereas density attenuation behaves more like an intensity dimmer).

	Distance atten = 0.5
	Distance atten = 1.0
	Distance atten = 2.0

Scattering phase

Controls the directionality of scattered fire light. 0 means isotropic scattering, positive numbers mean forward scattering, negative numbers mean backward scattering.

Max search radius

How far the scattering will spread, in world space units. Depending on the radius, this distance may encompass thousands or hundreds of thousands of points. Use the Max contributors parameter to limit the number of points considered for efficiency.

Max contributors

Of all points in the Max search radius, the material will consider this many of the closest points.

There is a point where additional light contributions stop having a significant effect on the overall result. However, this number is hard to determine without testing (it depends on several factors), so you should try to not limit the number of contributors at the beginning. Therefore, you should try starting with this parameter at a large number, like 10000.

Filter points

As the Pyro material marches through the volume, it computes the scattering contribution by averaging (or blurring) a neighborhood of precomputed point cloud points. This parameter is the number of cloud points in that “blur filter”. Higher values give blurrier results. The default (8 points) should be adequate in most cases for point clouds generated with a micropolygon renderer. For efficient shading, you should use the smallest number which still produces a smooth result. If the value is too small, you’ll seeing grid-like artifacts in the scattered light.

The following image shows filter points settings of (left to right, top to bottom) 1, 2, 4, 8, 16, and 32 in a constant cloud.

Note

If you are using your own render nodes to generate the point clouds using raytracing instead of micropolygon rendering for some reason, you will need to use a filter size of between 30 and 80 points due to the randomness of point distribution in raytraced point clouds.

Atten method

How the material calculates the attenuation and occlusion of light emitted from the fire field. The three choices represent trade-offs between speed and accuracy.

Approximated, No shadows

The default, cheapest approximation. This method assumes a constant density for each cloud point and will not calculate occlusion (shadows) from objects embedded in the smoke. It is the fastest to compute, and in cases when the smoke’s density doesn’t vary too much, it results in a fairly close approximation to real attenuation.

Approximated, Hard shadows only

This is the same approximation as “Approximated, No shadows”, but it also computes shadows from solid objects in the smoke (no transparencies).

Real

Actually traces occlusion to each fire light source instead of approximating, which gives the highest quality results but can be very expensive. Use this method if there is a great deal of variation in the smoke density, and the approximation starts looking unconvincing, or there are transparent objects in the smoke.

In practice, due to the order in which the sources are visited, it isn’t necessary to trace for every single source, so the material has a Quality parameter which allows you to control the frequency of these trace calls.

Approximated, No Shadows. Cost: 1.00

Approximated, Hard Shadows Only. Cost: 2.06

Real, quality set to 10. Cost: 26.67

Quality

Available when Atten method is “Real”. A value from 0 to 100. Traces occlusion once every N sources, where N is 100/Quality. The default 0.8 is useful in cases that don’t require precise shadows (just density attenuation). If you have detailed embedded objects with crisp shadows, increase quality to around 10.

Modifiers

See how to use modifier fields above.

Field modifiers

Activate field modifier	Make this field modifier “active”, so it shows up in the pop-up menus of modifiers next to mod-able parameters.
Label	A short, descriptive label for the field. This is how the field will appear in the pop-up menus of modifiers next to mod-able parameters. This string can include spaces and other non-alphanumeric characters.
Comment	An optional longer description of the modifier and its purpose. This value is not used by the material, it’s simply a convenient place to record information.
Base field	The field on the object to modify, for example `temperature`.
Contour	The Contour modifier operates on low field values and is useful to “sharpen” the contours of a smoke cloud. You can think of it as a “soft floor” operator. Values below the Width will be compressed according to Sharpness, where the higher the sharpness, the more abrupt the transition between 0 and the width will be. The following images show a density field (with Density Amplitude set to 500) where the field values were rendered directly on the left, and then modified with contour sharpening (Width = 0.02, Sharpness = 20) on the right. (If you use sharp transitions like this, you’ll probably need to lower the Volume Step Size on the Mantra render node or you may see shadowing artifacts). Turn on the checkbox to enable this modification.
Soft clip	The soft clip modifier operates on high field values and is useful to compress very high values. You can think of it as a “soft ceiling”. Values above Start will be progressively compressed according to Suppression such that, the higher the suppression, the more it will resemble a clamp operation. The following images show a temperature field which was rendered directly on the left, and then modified with soft clipping (Start = 0.5, Suppression = 20) on the right. Turn on the checkbox to enable this modification.
Range mapping	Range mapping simply fits one range to another, identically to the “fit” functions, VOPs, and operators available throughout Houdini. It maps the Field Range’s Min value to the Target Range’s Min value, and the Field Range’s Max value to the Target Range’s Max value. Values between these extremes will, by default, be interpolated linearly. However, you can use the Interpolation ramp to set a custom interpolation curve. Tip Depending on the Min/Max settings of the field and target ranges, it may be tricky to predict the values produced by the interpolation ramp. Just remember: wherever the ramp has a value of zero, you will get the target range’s Min value, and wherever the ramp value is one, you will get the target Range’s Max value. The range mapping is only active within the specified Field Range. Field values outside this range will clamp to the Min/Max values of the Target Range. Therefore, it is usually best to set as large a Field Range as possible and then shape the output using the ramp, using the target range as more of a scaling factor. (images) Turn on the checkbox to enable this modification.
Interpolation	The interpolation curve to use when Range mapping is on.

Noise modifiers

The Pyro material divides the visual characteristics of the noise into two categories: The basis (the fundamental type of noise), and the fractal construction (the method use to construct fractals using the basis) of the noise.

Activate field modifier	Make this field modifier “active”, so it shows up in the pop-up menus of modifiers next to mod-able parameters.
Label	A short, descriptive label for the noise. This is how the field will appear in the pop-up menus of noises next to mod-able parameters. This string can include spaces and other non-alphanumeric characters.
Comment	An optional longer description of the noise and its purpose. This value is not used by the material, it’s simply a convenient place to record information.
Exists in field	Set to “None” to have the noise exist everwhere in space, or choose a field to have the noise exist within some sub-range of the field, where the field’s values act as an envelope for the noise’s amplitude.
Active field range	The range of values from the input field (set in the Exists in field parameter) to map into the noise function.
Amplitude distribution within active field range	This ramp lets you scale the noise amplitude based on the values from the input field (set in the Exists in field parameter). For example, if the input field has a lot of high values, you can scale down the high end to make the output noise more uniform.
Noise basis	The type of noise to use as the basis for generating the noise fractal. You can use the basis noise as-is by setting Fractal type to “None”.
Invert noise	Inverts the output values of the noise function, so high valued areas become low and low valued areas become high.
4D noise	For noise bases that support 4D noise (noise that evolves in time), turn this checkbox on to use the 4D version. The default is to use 3D noise because 4D noise is more expensive to compute.
Frequency	Increasing frequency decreases the size of noise features.
Offset	Translates the noise in space.
Filter width	Controls the scale of the noise’s anti aliasing filter. The default filter size is calculated from the other settings and you will normally not need to adjust this, but there may be situations where the built-in anti-aliasing mechanism may under- or over-estimate the filter size. In those cases, you can use this parameter to scale the calculated filter size.
Exponent	Raises the value of the noise basis to the given exponent. You can use this to increase or decrease contrast without clipping. Values above 1 will increase contrast, values above 0 and less than 1 will decrease contrast.
Flow	Available when Noise basis is “Flow”. This is like a separate dimension, similar to time for 4D noises, which when animated can make the noise appear to “flow”. This is why this parameter defaults to “current time” (`$T`). When this parameter is exactly 0, the Flow basis is identical to the Perlin basis.
Fractal type	The method for building a fractal from the noise basis. You can set this to “None” to see the noise basis on its own. Note All fractals are constructed by adding a signal with a scaled amplitude and frequency back onto itself an infinite number of times. This makes them both self similar and infinitely detailed, at least in theory. In practice, the number of loops is constrained by the amount of detail that can be displayed – there’s a limit to how much high frequency detail can be expressed by a single pixel. In the Pyro material, the “signal” that keeps feeding back onto itself is the noise basis.
Max octaves	Each feedback iteration of the fractal is called an octave (because the frequency is typically doubled at each step, as in a musical octave). Since it’s a counter, the number of octaves would normally be an integer, but the Pyro material allows you to use floating point values to support driving the parameter with a modifier. For example, 3.5 octaves means 3 full iterations plus the fourth iteration at half its weight (a linear interpolation between octaves 3 and 4). All noise parameters in Pyro are continuous. Note that it’s “Max Octaves” not “Octaves”. This is because the actual number of octaves will depend on what can be practically displayed given the current camera settings, resolution, distance from volume, and so on without showing aliasing artifacts. The Max Octaves parameter says to not go beyond this number of octaves.
Lacunarity	Each octave (iteration of the fractal feedback loop) typically doubles the frequency of the signal. However, this doesn’t necessarily have to be the case: The multiplier could conceivably be something other than 2. This scaling factor is called “lacunarity” because the visual effect is that as this factor grows, the number of “gaps” between features appear to grow (lacunae is a fancy word for gaps). Lacunarity is a measure of how a fractal fills space. In the Pyro material, it is the frequency scaling factor per octave. (This parameter’s default is `2.01234` instead of `2.0` because, depending on the implementation, you will sometimes see the grid pattern of the basis lattice when the frequency factor is set to exactly `2.0`.)
Roughness	Sometimes called “gain”, this is the amplitude scaling factor per octave. This is typically set to 0.5 because in the standard fractal construction, as we double the frequency, we halve the amplitude. However, you can But again, as with “lacunarity”, there’s no rule that says this must be so. Therefore, it is exposed as a parameter.
Displacement	Available when Fractal type is “Multiplicative”, “Displaced”, or “Gradient Advection”. The parameter has a slightly different meanings for each. The Multiplicative fractal concentrates detail in places where the basis has high values and dampens detail where the basis has low values. This is why it is normally used for terrain generation, as it adds detail to the top of mountains but keeps the valleys smooth. The Displacement parameter controls the overall scale of that detail displacement. High values (above 1) will augment it – that is: The peaks will become noisier faster and the valleys will be dampened more, though there will be fewer of them. For the Displaced fractal, the Displacement parameter literally controls the amount that the spatial sample point will be displaced from its natural location. Visually, this means that the higher the number, the more “veiny” the appearance. For the Gradient Advection fractal the Displacement parameter is identical to that of the Displacement fractal. The only difference is the direction in which the sample point is displaced: the gradient of the basis instead of an arbitrary direction.
Lookup ramp for color/vector output	The output of the noise generator is always scalar (single floating point values). For the few places in the material that need vector values instead of floating point (i.e. colors), this ramp lets you map output values to colors.
Output range	Minimum and maximum output values of the noise function. For example, to arrive at something closer to what the Anti Aliased Noise VOP puts out, we can change the Output Range to . By default all noises are normalized to a nominal output range of 0-1. Changing the output range to include negative values is perfectly safe and should not produce illegal output. The Pyro material will not allow fields which shouldn’t be negative (such as density) to drop below zero.
Final amplitude	Scales the output of the noise function.

Settings

Shading mode	Switches between normal shading and the generation of point clouds. See how to generate scattering point clouds.
Current time/Current frame	Used by the material for 4D noise and solving dual rest position fields. You should leave these at their defaults (`$T` and `$FF`) unless you know you need to change them for some reason.
Dual rest	Rest fields will become less defined as they diffuse, creating an undesirable stretching of any shader-based noise that is dependent on them. This option counteracts this by maintaining two rest fields which reset their values at different times. The material interpolates between the two fields to give the impression of a stable rest position.
Start frame	You do not normally need to change this parameter. However, if you change the Start frame parameter on the Rest tab of the Pyro solver node in the DOP network for the effect this material is applied to, you will need to change this parameter to match.
Reset rate	The number of frames between when the material resets a rest fields. The material staggers the reset points of the two fields so the reset of an individual rest field is not visible.

User fields

If you need to use fields from the simulation besides the common ones already listed in the field pop-up menus (e.g. density, temperature), you can use a Name surface node to rename the field to one of user1, user2, etc. Then in the material, use the controls on this sub-tab to add the field(s) into the material’s pop-up menus.

Include in menus	Turn this on to add a given `usern` field to the pop-up menus of available fields in the material’s interface.
Label	A short, descriptive label for the field. This is how the field will appear in the pop-up field menus. This string can include spaces and other non-alphanumeric characters.
Comment	An optional longer description of the field and its purpose. This value is not used by the material, it’s simply a convenient place to record information.

Output variables

A list of variables exported by this material.

Color correction controls

Each of the principal tabs, Smoke, Fire, and Scattering, has a block of color correction parameters. These are conveniences to allow in-shader tweaking of the final composition of each component. It is not always possible to delegate all correction to the compositing stage when working with volumes because the components influence each other during evaluation, so it is sometimes useful to have some basic correction control at the shading stage.

These corrections are applied in the order they appear in the interface. That is, HSV, then Contrast, then Gamma, and finally Tint.

Beyond that, the meaning of these corrections should hopefully be self explanatory as they are no different than the similarly named corrections found in the Halo (or, indeed, any other) compositing environment. Perhaps the only thing worth mentioning is the HSV correction as it is compressed into a single vector parameter.

Enable color correction	When this option is on, the material uses the parameters below to modify the final composition of the Smoke, Fire, or Scattering.
HSV	Rotates the hue. This value is normalized 0-1, so a value of 1 rotates the hue by 360° (a full rotation back to the starting hue). This value is not bounded and can be either positive or negative. For example, a value of -0.5 will rotate the hue by -180° and be equivalent to -1.5, -2.5, -3.5, etc.
Contrast	Increases or decreases the contrast.
Gamma	Modifies the gamma.
Tint	Tints the output color with a second color.

Usages in other examples

Example name	Example for