On a 128x128x128 grid. Using incompressible, inviscid Navier-Stokes equations with Semi-Lagrangian advection to simulate the field and using half-angle slicing technique (+self-shadows) for volume rendering.
@youchaney I did not try this on a really high-res grid (This example wan on a 128x128x128 grid), so detail gets better on high-res grids, and with probably more slices per volume (so they are closely spaced between each other). The other important factor is the resolution of the light's shadow accumulator buffer.
I believe this technique was mostly created for real-time volume rendering, so obviously it is not physically perfect to catch all details.
@youchaney I did not try this on a really high-res grid (This example wan on a 128x128x128 grid), so detail gets better on high-res grids, and with probably more slices per volume (so they are closely spaced between each other). The other important factor is the resolution of the light's shadow accumulator buffer.
I believe this technique was mostly created for real-time volume rendering, so obviously it is not physically perfect to catch all details.
Nice . Does the self shadowing technique allow for sharp detail ? We will take our base volume and use a 3d fractal multiplier to put extra detail into the volume . Some optimize shadow routines miss this detail and a full on ray march is sloooow . Does the Half-angle technique hold up the sharp detail ?
@OBOGAN I used a technique called Half-angle slicing originally created by Kniss et al in 2003: . "A Model for Volume Lighting and Modeling."
But Simon Green from nVidia wrote a technical paper on Volumetric Particle Shadows where he used this technique to create shadows. It is not really nVidia or hardware specific.
@youchaney I did not try this on a really high-res grid (This example wan on a 128x128x128 grid), so detail gets better on high-res grids, and with probably more slices per volume (so they are closely spaced between each other). The other important factor is the resolution of the light's shadow accumulator buffer.
I believe this technique was mostly created for real-time volume rendering, so obviously it is not physically perfect to catch all details.
skremon 1 year ago
@youchaney I did not try this on a really high-res grid (This example wan on a 128x128x128 grid), so detail gets better on high-res grids, and with probably more slices per volume (so they are closely spaced between each other). The other important factor is the resolution of the light's shadow accumulator buffer.
I believe this technique was mostly created for real-time volume rendering, so obviously it is not physically perfect to catch all details.
skremon 1 year ago
Nice . Does the self shadowing technique allow for sharp detail ? We will take our base volume and use a 3d fractal multiplier to put extra detail into the volume . Some optimize shadow routines miss this detail and a full on ray march is sloooow . Does the Half-angle technique hold up the sharp detail ?
youchaney 1 year ago
@OBOGAN I used a technique called Half-angle slicing originally created by Kniss et al in 2003: . "A Model for Volume Lighting and Modeling."
But Simon Green from nVidia wrote a technical paper on Volumetric Particle Shadows where he used this technique to create shadows. It is not really nVidia or hardware specific.
skremon 1 year ago
You speck about Volumetric Particle Shadows exposed by nvidia ?
OBOGAN 1 year ago