nuki

Doing some Bullet/Heightfield collision and it seems like bullet uses unusual basis vectors for the heightfield lookup? I have to transform the sim data (ry90, sx-1) coming out of the dopnet so it will line up with the heightfield again. Even though the terrain & packed objects show up in their correct orientation inside the dopnet. heightfield_collision.hiplc

To back up fencers suggestion, heres a part of the H18.5 keynotes where they do pyro trails: https://youtu.be/zMrvCWy85xM?t=6089

Yea, it seems offset by half voxelsize - wondering if its the sampling position of the noise. maybe the offset is needed so @P points def to voxel center. Resampling the source to the max bounds before applying the noise also works. I had no idea that I could change the Voxel Sampling Edge! Thanks a lot man!

I have a source vdb and a sparse pyro dopnet, the source being a thresholded noise with only 0.0 or 1.0 values. When enabling "solve on creation frame" I would expect the values of the source and the sim (with all advection disabled) to be the same. As you can see they're not, and I can only suspect the grids of the source and sim volumes are not lining up resulting in interpolated samples. left is a slice of the source, right is a slice of the sim For sake of avoiding interpolation of the values during sourcing, the source volume field should exactly match the sim field (voxel size and positions). So I attempted to resample my source volume with the max bounds, center and voxel size settings of the sim, but there are still lots of intermediate values generated during sourcing. Another approach was to use "gas match field" to match my flame field to my source vdb. That worked nicely but the field wont be resized anymore, which isnt good when advecting. So how could i align my volume sources voxels perfectly with my (sparse) pyro sim? pyro_match_source.hiplc

i suspect its just the mass of the bullet being too low. try increasing it in factors of 10 and see if it helps!

Like this - pyro_flare.hiplc You can improve that by making the volume source more interesting (noise) and playing with forces. For manipulating the vel field the "Volume Velocity" sop is a quick solution. If everything looks way too blocky, disable your scene view's display options/ textures/ limit 3d textures setting.

I've played around with the sparse pyro combustion model a bit and have a few question about the emission of smoke by flames. - I was under the impression that, when emitting smoke eg. in flame range 0-0.5 from a clear (full 1.0) burn field with a flame life of 2.0 there should be no smoke emitted at all in the first second? However, my flame field immediately drops down to values below 0.5 on the outer borders, even without any diffusion/dissipation going on. Hence theres smoke coating my flames from the start. How could my emitter (values 0 or range 1-2, sourced with max) produce values as low as 0.1 in the flame field on the second simulation frame? Maybe blurring by the advection? Or am I missing some field interaction here? - I also wanted to use a noise inside a sop solver to vary the flame density emission rates. While I could import the field with Dopimport Fields I could not find a way to apply that noise outside of the sop solver (bind it to the dopnet). Here's a small example scene to showcase the issue: sparsePyro-burn.hiplc And a quick render:

you could also use the RBD disconnected faces sop. It requires unpacking but works great oob

Nice man, great resources too!

Did you check the UV View (Space+5 in the Viewer)? Depending on your scale you might need to decrease the spacing on the UV Unwrap by a lot. Anyways, heres a file: frac_uv.hiplc

Found this in the hou docs: Filaments are closed polylines (polygonal curves) that represent vortex forces. The general filament setup works like this: Create a a geometry network to create the filament geometry. Import the filament geometry into the simulation network using the Source Filaments node. The Filament Solver node advects the points of the filaments to make them drift through space. While the Filament solver moves the filaments, the POP Advect by Filaments node applies vortex forces to surrounding particles, making them swirl around nearby filaments. source

looks like an alpha channel issue with the flames which get lost when blender premultilpies them. You could max() the result of a luma key with the alpha for a quick fix. Or fix the shader in houdini and re-render. Or not premult at all if possible.

length = sqrt(x²+y²+z²) the way i read it, length2 = x²+y²+z² in some cases (eg. when just calculating the ratio between two distances) you can simply ignore the square root calculation to increase your codes performance.

have you tried the sidefx labs quad remesher (requires license) or the instant meshes (open source) bridge?

for deforming geometry use the pointdeform sop. if you have multiple objects this is a great thread on setting it up and maximizing performance. transformpieces is for matching packed/rigid stuff (eg. the output of a bullet sim) . pointdeform.hiplc