Jawset Visual Computing

What is Jawset Visual Computing?

VOXEL-BASED FLUID GAS DYNAMICS

The Jawset Visual Computing simulation pipeline  implements a voxel-based solver based on Navier Stokes' incompressible equations. This means that it uses a voxel grid to describe the volumetric clouds of smoke and fire and solves the equations that describe the movement of the fluid in that grid. For each voxel, TurbulenceFD calculates the velocity of the fluid and various channels to describe properties such as temperature, smoke density, amount of fuel, etc. This simulation process produces a voxel grid for each frame, which is cached on disk for use by the Volumetric Renderer.

INTUITIVE WORKFLOW

To set up a fluid simulation, the artist uses any kind of geometric object or particle system to paint the sources of smoke, heat, fuel, etc. in space. The flow carries these emissions in a physically plausible way that creates the realistic appearance of fire, explosions, steam, clouds, dust, and more.

PULL ALL STOPS FROM YOUR CPU

The biggest technological challenge in fluid simulation is the handling of large amounts of data that a sequence of voxel grids requires. That's why Jawset Visual Computing's simulation pipeline  was designed from the ground up to optimize performance. This includes a careful selection of efficient numerical methods that provide high accuracy and stability throughout the simulation pipeline. And implementing this pipeline using the latest high-performance computing technology to exploit memory caches, multicore CPUs, and advanced vector instruction sets. For the artist, this means that more iterations can be performed in less time, making working with fluids more intuitive and productive.

Up to 12x GPU speed

Yes, twelve times! 10 minutes instead of 2 hours. And there's a simple reason for that: Today's high-end GPUs have 8 to 15 times the memory throughput of high-end CPUs. Jawset Visual Computing exploits this. It has a hybrid CPU/GPU simulation pipeline that achieves huge accelerations. Unlike some GPU-based tools, this isn't just a simplified version of CPU simulation. All features are supported with the same quality. When the GPU memory is exceeded, Jawset Visual Computing reverts to the on-the-fly CPU. This allows you to achieve near-real-time speeds for low resolutions and scale smoothly to high resolutions in the hundreds of millions of voxels. Instead of carefully changing the parameters, submitting the simulation work, and not seeing the results for hours, the fluid simulations can be adjusted in quick iterations with the artist observing the effect of the changes while the simulation is processing.

PHYSICALLY BASED FIRE PUMP

Getting the right colors is key when creating credible fire animations. You can project your color gradients manually for full artistic control. If you want realistic fire colors, the process of adjusting the colors directly can be time-consuming and tedious. Therefore, the fire shader simulates realistic colors of high dynamic range of fire, based on the Black Body Radiation model. This model is controlled by only two temperature values. It generates the colors that real fire would have at these temperatures. But Jawset Visual Computing doesn't stop you out there. You may want realistic colors, but you need more flexibility to adjust the huge dynamic range that fire has. Maybe give the reds a boost, compress the dynamic range a bit, or just use the generated colors as a starting point to edit the ones directly directly.

MULTIPLE DISPERSION

In a nutshell, multiple dispersion is the global illumination for smoke. It is a way to light the smoke in a more realistic and brighter way as it is illuminated from all directions. It also allows the fire to illuminate smoke from the inside, essential for realistic shading of explosions. Unlike many global lighting techniques, multiple scattering in Jawset Visual Computing does not add noise and therefore works well with animation. And the multiple dispersion rendering times in TurbulenceFD are really affordable. But if you're in a hurry, you can still make a compromise between the speed and detail of the lighting.

PARTICLE ADVECTION

The heart of fluid dynamics is the creation of a sequence of velocity fields that describe the characteristic and complex motion of the fluid. You can use Jawset Visual Computing's speed caches  to control the movement of particle systems. This allows you to supplement the voxel renderer with debris or sparks or just render the particles yourself.

ADAPTABLE CONTAINER

Jawset Visual Computing constantly tries to minimize the volume that needs to be processed to save memory and time. The velocity field is analyzed to ensure that only the parts of the volume are cut off that will not affect the flow in subsequent frames. If necessary, you can control the sensitivity of the cutout for each fluid channel.

EMITTERS

Emitters are for fluid simulation what brushes are for painting. If an object is set on fire, it emits heat and a flame. Jawset Visual Computing allows you to use any geometric object or particle system to emit in fluid channels. This gives you maximum freedom of form and animation from your emitters. Working with emitters in TurbulenceFD is like animating the brush strokes that paint the sources of fire, smoke, etc. The fluid simulation takes your animated emission and creates a physically plausible flow from it.

COLLISION OBJECTS

Letting fluid flow interact with solid objects is useful in many scenes. From a solid, simple floor to vehicles moving through fire and smoke to animated characters catching fire. Not only is it necessary to integrate the simulation into an environment, but it creates a pleasant and natural turbulence in the sequence of an object. Collision objects can shake the fluid, wave to the side, or act as an obstacle. Jawset Visual Computing also supports collision objects with all kinds of complex animation, including MDD imports and objects controlled by rigid body dynamics.

SHADOW CURVE EDITOR

The heart of voxel shading are function curves (f-curves) that remap values such as temperature and density to intensity values used for opacity and color. Jawset Visual Computing has an f-curve editor that is specifically designed for voxel-based fluid shading. It allows for precise and intuitive control, making it very similar to the color correction workflow that many artists are already familiar with. And because f-curves need to be evaluated billions of times during rendering, a special type of spline curve was designed for TurbulenceFD that is particularly efficient at voxel rendering.

TURBULENCE MAPPING

Adding procedural noise to the fluid's velocity field is one way to get wavy flows that look more turbulent and more interesting. The controls work pretty much like a procedural noise shader, commonly found in texturing tools. However, adding the turbulence evenly throughout the entire volume will shake the core of an explosion as much as the farthest parts of the violent reaction. That doesn't make much sense. Thus, TurbulenceFD allows you to control where exactly you add curls to your flow using one of the fluid channels and a simple mapping curve. In this way, turbulence can be added only to certain regions, such as the core of an explosion or the hot part of a flame, for example.

VIEWPORT PREVIEW

The OpenGL-based visualization provides a detailed view of each of the fluid channels in real time. Several shading modes are supported in the preview. There is an analytic mode that provides a detailed view of the raw output of the simulation. And there are shaded modes that provide real-time feedback while enhancing the settings of each shader. In addition to the fully three-dimensional viewing modes, you can display a 2D slice of the voxel grid, oriented and positioned anywhere on the volume. This can be thought of as the magnifying glass of viewing modes and is comparable to a wire-framed view of geometric objects.

VOXEL NETWORK COMPRESSION

To help deal with big simulation data, TurbulenceFD features lossless data compression that is specifically designed for fluid data. It typically reduces voxel data to about 60% in practice.

IGNITION CONTROL

Controlling how fire is ignited and how quickly a flame spreads in TurbulenceFD is as easy and flexible as painting fluids with emitters. It can be based on any fluid channel, not just temperature. This avoids the equilibrium you must perform if the temperature also drives the buoyancy force that allows the hot gas to rise.

UP-RES'ING

Typically, you perform the simulation at a low or medium resolution that allows for fast iterations. Then you simply want to simulate again at a higher resolution to get the end result. But this can not only add high-resolution detail, but also slightly alter the motion on a large scale due to the numerical nature of the simulation. Up-Res'ing is a way to maintain the exact shape and movement of a low- or medium-resolution simulation and add only high-resolution detail to it. It's also faster than running a full simulation at the same high resolution.

SUB-GRID DETAIL

With the detail of the render-time sub-grid, TurbulenceFD further enhances the resolution boost approach. Instead of having to perform a second pass in the basic simulation, simply add the high-resolution details to the result at the time of rendering. For extreme configurations, this isn't as flexible as Up-Res'ing, but it also doesn't require additional simulation passing or additional cache memory.

System Requirements

Available for Windows and Mac