Blender Git Commits

Blender Git "cloth-improvements" branch commits.

Page: 2 / 6

March 17, 2017, 01:42 (GMT)
Fix basemesh target object poll function
March 16, 2017, 20:56 (GMT)
Merge branch 'master' into cloth-develop
March 15, 2017, 08:36 (GMT)
Implement rest planarity factor scaling
March 14, 2017, 05:50 (GMT)
Implement combined vertex weight by group influeces (Weight Keys)
March 14, 2017, 05:50 (GMT)
Implement weight keys for pinning
March 14, 2017, 05:50 (GMT)
Refactor: remove unnecessary re-initialization

Cloth was being re-initialized whenever pinning settings were changed.
This has been necessary in the past, as pins were only computed during
initialization, but ever since animated vertex weight support was added,
by having pins recalculate on every step, this re-initialization has
become unnecessary.
March 14, 2017, 04:00 (GMT)
Refactor: remove goal springs in favor of per vert computation
March 13, 2017, 04:31 (GMT)
Implement animated cloth velocity initialization

This allows the initial cloth velocity to be set by the vertex
velocities in the underlying animation.
March 12, 2017, 19:26 (GMT)
Add support for external basemesh object

This allows the usage of a different object as the rest shape, which
enables more flexibility regarding animated rest shapes.
March 4, 2017, 04:38 (GMT)
Remove a few warnings and some cleanup
March 4, 2017, 04:12 (GMT)
Implement normal override for the new collision function
March 4, 2017, 03:24 (GMT)
Replace plNearestPoints with a deticated solution

Implement a dedicated collision detection function (was previously
relying on Bullet's generic `plNearestPoints`).
This function computes all the collision data to be used for response:
coordinates, distance, direction vector.

This new function has three advantages:

* Remove a dependency from cloth simulation (Bullet).
* Give more pleasing collision results (this function is tailored
specifically for our collision response method).
* Much faster computation (not benchmarked extensively, but observed
overal simulation time was cut roughly in half with "collision-heavy"
simulations).
March 3, 2017, 23:41 (GMT)
Simplify impulse clustering pointer stuff
March 3, 2017, 23:36 (GMT)
Expose slice intersection function (math_geom)
February 12, 2017, 03:48 (GMT)
Fix stack overflow in spring angle calculation
February 4, 2017, 05:57 (GMT)
Fix regression on plasticity caused by sewing springs

I must have been sleeping when I committed the fix for the sewing
springs regression, causing another silly regression on plasticity...
February 3, 2017, 21:59 (GMT)
Add collision response quality control
February 1, 2017, 16:00 (GMT)
Implement separate normal override and single sided collisions

This adds single sided collisions without necessarily using normal
override, and adds a separate normal override option. (code can do with
some cosmetic cleanup)

Also removed a printf I had forgotten.
February 1, 2017, 06:09 (GMT)
Implement collision impulse based adaptive step size calculation
February 1, 2017, 03:17 (GMT)
Apply all object collisions simultaneously

This makes all multi-object collisions be solved simultaneously, as not
to give any of them priority over the others (as in the sequential
collision solve that was implemented)

Note that self collisions have not been integrated, and are instead
solved independently after the object collisions. This is so that they
don't have to fight against the object collisions, as those are
generally more robust, and might cause issues with the self collisions.
By: Miika HämäläinenLast update: Nov-07-2014 14:18MiikaHweb | 2003-2021