Blender Git Commits

interpolating phi later in step

interpolation needs to come later to avoid weird mesh resizing in first frames (occurred between frame 1 and 2)

use larger surface distance value so that obvels cover larger area. manta can then better extrpolate those obvels

small fix for frame number: manta starts on frame 0, not on 1

save and load obstacle velocities (for standalone mode)

added cleanup routine in destructor for multigrid

updated manta pp files

removed dummy fractions grid from smoke script

updated manta pp files and adapted obstacle velocity setter in smoke/liquid script

important: boundaryWidth for liquids is 2 instead of default 1

added adaptive time stepping option to ui

added support for liquid moving obstacles

using same obvel setter as in smoke script. fractions left out for now

removed high res clean up functions for liquid

in theory with correct obvels no need to clear phi inside obs. if still needed, then maybe do this later in blender c code

added guard for smoke/liquid sampling/distance functions

in smoke domains no need to get mesh distances, in liquid domains no need to calculate smoke influence factor

removed flow type guard and instead made FLUID destructor less restrictive

required if multiple domains in viewport. domains do not know what flow objects they have inside

fix for wrong framenr argument in step function

improved raycasting for mesh distances again

also includes name and formatting cleanup

improved levelset calculation function

now using better estimates for min distance and also 14 instead of just 6 raydirs for raycasts (added 8 quadrant diagonals)

removed unused phi reset from obstacle update function

added ids to all grids, vars, funcs, etc in manta scripts

having unique python variables allows having multiple domains in the viewport at once

split up fluid solver and helper variable setup

fixed and cleaned up inflow grid

inflow grid is necessary when running manta script in mantaflow standalone mode