Volumetrics

This extension actually consists of 4 types of volumetrics:

  • "Constant density" - creates a cube shaped volumetric area in your scene which has a constant density. This is useful for very large atmospheric haze effects for example.
  • "Noise3D" - this also functions as the constant density type except it also adds an adjustable 3D noise to create density variations in the volume.
  • "Particle based" - this lets you use a particle simulation to define the volumetric density. You can either use particles directly in supported applications (such as Maya, 3ds Mac, Cinema4D, Softimage), or load an external particle file in several formats (.abc, .bin, .pxy, .prt, .rpc).
  • Volume file based - this works currently with the OpenVDB file format which is a format that describes a volume based on voxels and is not particle based. Usually these types of files can be created with RealFlow, Houdini, Maya. The advantage of using this format is that you can store very detailed and large smoke/explosion simulations without the need to store billions of individual particles.

 

Icon

Any Maxwell material (except emitters) can be applied to the volumetric object (or particles), but generally a normal lambert material works just fine.


How to use the Maxwell Volumetrics extension

The details will depend on what plugin you are using, or whether you are using Maxwell Studio, but in general, you add this extension as an object to your scene and it will be represented by a cube (in the case of Constant density and Noise3D), points that show the particle cloud in the case of Particle based, and a cube bounding box representation in the case of the Volume based type. Please see your specific plugin documentation for details on how to add an volumetric object to your scene. You then apply any Maxwell material to this object.

Constant density

This will actually be a separate cube shaped object in your scene, which can be manipulated like any other object (translated, rotated, scaled). The volumetrics will be visible only inside this cube shape. Please see your plugin documentation for specifics on how to add a Constant Density object to your scene. In Maxwell Studio, you can add it by right-clicking in the Object panel and choosing Create Extension Object>MaxwellVolumetric.

An example mimicking heavy atmospheric haze using the Constant Density cube encompassing the scene, and the resulting renders with/without the volumetrics. Note that a very low density of 0.003 was used since the Constant Density area was so large in this case (1km). The Maxwellsea extension was used to create the terrain.

Noise3D

As mentioned above, this type also uses a cubic shape to define the volume, but adds a 3D procedural noise to add some variation to the density. In the following examples, the density has been set very high to make it easier to see the changes to the noise parameters.

Seed

Define a starting seed for the randomization of the noise pattern

Detail

Create larger or smaller sized noise detail. In other applications this parameter is sometimes referred to as "wavelength".

Detail parameter from left to right: 2, 4, 8

Octaves

The number of times that the initial noise pattern should be repeated. With each repetition (octave) the noise will be half the size of the previous octave. The effect will look similar to having an initially smaller Detail, but the difference is that the larger initial Detail will still be visible, and the Octave parameter will add smaller noise to the initial larger noise pattern.

Octaves from left to right: 1,2,3. Detail level was kept at 3 and Persistance was kept at 0.6 for all renders.

 

To better visualize how the Octave parameter affects the noise pattern, lets look at it through a 2D cross section:

 

5 octaves of noise shown individually, and then added together to form the final noise. Notice how the final noise retains some of its original shape from the first octave (in this case a medium Persistance was used, see below)

 

Persistance

This parameter controls how "persistant" the noise is through each octave. With a high setting, the initial noise pattern quickly disappears, while the frequency of the noise increases:

Example of using 5 octaves, and a high Persistance such as 0.9 -notice in the result how the initial noise pattern is not really present anymore because of the high frequency/small wavelength noise

 

On the other hand, if the Persistance is very low, the noise pattern will dissipate very quickly with each octave:

Still using 5 octaves, but a small Persistance such as 0.1 - notice how the initial and final noise patterns are largely the same, no matter how many octaves you specify.

 

Icon

Relationship between Octaves and Persistance

From the diagrams above it follows that:

  • If you set the number of Octaves to 1, the Persistance parameter does not matter anymore.
  • If you set the Persistance value to 0 (or extremely low), the Octaves parameter does not matter anymore.

In both of these cases, it is only the Detail parameter that will influence the look of the noise.

Particle based

This type of volumetric will create density based on particles, either loading a particle file or by adding the Maxwell volumetric modifier to a particle simulation in your main application (please consult the documentation of your plugin on how to do this).

Icon

Particle file formats currently supported

  • .bin (RealFlow format), .abc (Alembic format), .pxy (RealFlow format), .prt (Krakatoa format), .rpc (RealFlow format)

For more information about these formats, please see this page in the RealFlow documentation.

 

Radius multiplier

For any density to be visible in the render, the particles must first of all have a certain initial radius. This is determined by you when you create the particle simulation, depending on which scale you work in, and at what scale you have exported your scene to Maxwell. This parameter lets you multiply that radius - in case the particles have a too small radius, or you want to change the look of the volumetric. Increasing this parameter has the effect of making a more "puffy" less defined look.

If nothing appears in the render, this is the first parameter you should change. Generally, the idea is to avoid seeing individual particles and instead have a continuous looking smoke.

Icon

Be careful not to raise this parameter too high. If you start noticing that the scene voxelisation time increases a lot, it usually means the radius multiplier is set too high

From left to right, changing the radius multiplier: Value of 2 (perhaps a bit too much), 1 (seems a good value in this case), 0.3 (too small)

Density multiplier

You can increase or decrease the amount of density that each particle will contribute with this parameter. Note that the final density will still be limited by the "Max final density" parameter.

Values of 2, 5, 100 ("Max final density" parameter was kept at 35 for all renders)

Min final density

The minium density in a particular cell of the volumetric object. In most cases you should leave this value at 0 as you don't want any visible density where there are no particles.

Max final density

The max allowable density in a particular cell of the volumetric object. Note that this parameter is also influenced by the Density Multiplier - a small density multiplier will not make the volumetric look more dense, even if the Max final density is very large. In fact, you should avoid setting this setting very high (for example more than a few hundred), especially if the density multiplier value is small. This will result in a similar look as if a smaller max final density value was used, with the downside that the render time will be substantially slower. As you can see in the example renders below, if the Density multiplier is already fairly large, then a Max final density value of around 100 will be enough even for the most "solid" volumetric simulations.

Values of 1, 10, 100 (Density multiplier was kept at 25 for all renders)

Cell size

The cell size should be considered the "resolution" of the volumetric object. The parameter uses a real world unit of 1m. The smaller the cell size, the finer the resolution. Just as a bitmap image can reveal more detail when zoomed in if it contains a lot of pixels. The cell size parameter is necessary in order to keep the RAM usage under control. So the final RAM consumption is not influenced by the number of particles in your simulation, but instead on the number of cells which are created at render time, which define the resolution of the volumetric object. 

 

Icon

Decrease this setting carefully as very small changes can cause your RAM consumption to increase a lot.

 

How cell density is determined

It is important to understand that what you finally see in the render is the contribution of the density of each cell, NOT particles directly turned into density. So how is the amount of density for each cell determined? Each cell looks from it's corners into the cell to see how far away a particle is in regards to each corner. Depending on the particle radius, density multiplier, and max final density - some density may be shown, or nothing at all.

The above illustration shows a particle that happens to have the same size as the cell. But even in this case, it isn't certain that this cell will show any density if the Density Multiplier and Final Density settings are too small, because the cell still has a certain distance to search before finding this particle. The longer this search distance is, the more faded the density becomes.

Let's see some more examples:

A 10cm large cube of particles, where each particle is 2cm in diameter, and each cell is 1cm in size (setting of 0.01). The cells are small, so a smaller search distance inside the cell is needed to find a particle, thus we see some density created in each cell in this case. Notice that we are not seeing the round shape of the particles, but rather the square shape of each of the cells. The cells (the "resolution") would need to be even smaller to start mimicking the round shape of the particles.

 

Here the cell size was increased to 2cm. Since the distance each cell has to search has increased, nothing appears in the render. We could raise the Radius Multiplier in this case to make the particles bigger, or lower the cell size again to make some density reappear. We could in fact also raise the cell size even more, and some density would appear again. See the next illustration for why this happens.

 

Now the cell size is 3cm and we start seeing some density again. This can seem very unintuitive at first, but looking at the grid of cells, it is clear that now some corners of some cells are actually closer to a particle compared to when the grid was set to 2cm. So we start seeing a very rough, square representation of these particles, because the cell size is so large.

 

Density multiplier as a "falloff"

All renders with a cell size of 2mm, Max Final Density at 500. Only the Density Multiplier changed: 10000, 3000, 1000.

This series of images is ment to show two things. First, that with a small enough cell size, each particle can be very well defined (in these renders the cell size was reduced to 2mm (a setting of 0.002). Second, how the Density Multiplier parameter can act as a sort of falloff between areas of density and no density. The Max Final Density was 500 for all renders. You can see that as the Density Multiplier decreases, there is a larger area of falloff between no density to the areas of maximum density. When the Density Multiplier is very high, there is a very sharp falloff, making the volumetric shape have much sharper contours.

 

Extra particles

This parameter adds extra particles around the simulated particles. This way, a much lighter simulation with a small amount of particles can produce a more dense cloud by multiplying them at render time. The new particles are affected by the speed of the original ones, so your particles must have some speed or this parameter won't have any influence. Note that more RAM will not be used if you create extra particles, this depends again on the final number of cells created by the cell size parameter. If you do use a high dispersion though, it is likely that more cells will be created since the total bounding box of the particle cloud will increase, thus increasing the amount of RAM needed for rendering.

Extra particles set to 0, 10, 200. Dispersion was set to 0.1 (10cm) in this case. RAM usage only very slightly increased even with 200 extra particles per particle. The right most image shows the particle radius increased to avoid seeing the individual particles.

 

Extra particles dispersion

The distance (in meters) that the extra particles should be allowed to be randomly distributed around a sphere based on the original position of their "parent" particle.

Extra particles deformation

The new particles can be generated randomly inside a sphere based on the position of the "parent" particle, or inside an ellipsoid with this value of eccentricity. The ellipsoid is aligned along the direction of the parent particles speed.

Motion blur factor

You can control the amount of motion blur visible on the particles, separately from the scenes main motion blur settings. If you want the particles to look blurrier as they move, increase this parameter.

 

 

Related content: