Blender Git Loki

Fix T46437: Make progress report py helper resitent to 'zero steps' passed value...

To be backported, should we need an 'a' release.

Fix T46441: texture paint soften brush at 0 strength works at full strength.

small fix for potentially unintended calculation

changed calculation of stability factor for compounds and adapted UI a bit (for compounds)

disabled damage propagation (and settings) in code and ui, was slow and not really necessary, also fixed a crash which happened when freed bullet manifolds were used again.

tiny fixes for autohide and attempt to get rid of propagateDamage crashes (again)

attempted fixes for : autohide, damage propagation in compounds (was crashing after some time) and compound parameter changes do reset the cache now, too

calculating minimum impulse per object now, also made solver iteration settings not visible for compounds any more (they relate only to constraints)

attempt to fix some (stack overflow related ?) crashes

tiny fix, forgot to assign ids to meshislands in splitshards to island mode

small fix for ui description of mass_threshold_factor

new parameter "stability factor" (for constraints, compounds) and resetting all shards by default now on refracture, except in mousebased fracture (faster to keep unchanged shards there)

some enhancement attempts for compounding, supposed to be better controllable in some cases, but still odd behavior with multiple compound objects

fix for triggers and multiple compound objects (but compound building is slower again, due to necessary broadphase handles for triggers)

exposed a couple of compound damage propagation parameters, need to test still to find good values

windows compile fix

crash fix in propagateDamage, checking for index being smaller than connection size now

speed enhancement for building compounds (omitting broadphase handles) and successful test with clusters, values need a bit tweaking but basic principle works

added very basic damage propagation scheme to compoundbased fracture (direct damage is too small, can be increased this way, but you mostly get single parts falling off only)

some fixes for compound method, but still not very well controllable