diff options
| author | Alexander Gavrilov <angavrilov@gmail.com> | 2018-07-07 23:21:20 +0300 |
|---|---|---|
| committer | Alexander Gavrilov <angavrilov@gmail.com> | 2018-07-19 17:58:27 +0300 |
| commit | 0bf8096501a8e7883f4061ba3e425966ba7517cd (patch) | |
| tree | 0535d2a4df15dc45c02b114bdd34f615f3a0d61a /source/blender/blenkernel | |
| parent | 51f14cfa43e5fe4bdea6fbd5c668a44919657f99 (diff) | |
Resolve the opposite vector ambiguity in Damped Track constraint.
Damped Track by specification attempts to arrive at the desired
direction via the shortest rotation. However with opposite vectors
there are infinitely many valid 180 degree rotations. Currently
it gives up and does nothing.
I think that it would be more reasonable to resolve the ambiguity
arbitrarily, so that Damped Track won't have a weird dead zone.
To make it more predictable I use a local axis.
In addition, the singularity area vicinity has some floating
point precision problems that result in significant jitter.
This applies workarounds for two causes of instability.
Differential Revision: https://developer.blender.org/D3530
Diffstat (limited to 'source/blender/blenkernel')
| -rw-r--r-- | source/blender/blenkernel/intern/constraint.c | 32 |
1 files changed, 30 insertions, 2 deletions
diff --git a/source/blender/blenkernel/intern/constraint.c b/source/blender/blenkernel/intern/constraint.c index 2a5a0cf9ae7..7d861658993 100644 --- a/source/blender/blenkernel/intern/constraint.c +++ b/source/blender/blenkernel/intern/constraint.c @@ -3709,7 +3709,7 @@ static void damptrack_evaluate(bConstraint *con, bConstraintOb *cob, ListBase *t * - the min/max wrappers around (obvec . tarvec) result (stored temporarily in rangle) * are used to ensure that the smallest angle is chosen */ - cross_v3_v3v3(raxis, obvec, tarvec); + cross_v3_v3v3_hi_prec(raxis, obvec, tarvec); rangle = dot_v3v3(obvec, tarvec); rangle = acosf(max_ff(-1.0f, min_ff(1.0f, rangle))); @@ -3717,7 +3717,35 @@ static void damptrack_evaluate(bConstraint *con, bConstraintOb *cob, ListBase *t /* construct rotation matrix from the axis-angle rotation found above * - this call takes care to make sure that the axis provided is a unit vector first */ - axis_angle_to_mat3(rmat, raxis, rangle); + float norm = normalize_v3(raxis); + + if (norm < FLT_EPSILON) { + /* if dot product is nonzero, while cross is zero, we have two opposite vectors! + * - this is an ambiguity in the math that needs to be resolved arbitrarily, + * or there will be a case where damped track strangely does nothing + * - to do that, rotate around a different local axis + */ + float tmpvec[3]; + + if (fabsf(rangle) < M_PI - 0.01f) { + return; + } + + rangle = M_PI; + copy_v3_v3(tmpvec, track_dir_vecs[(data->trackflag + 1) % 6]); + mul_mat3_m4_v3(cob->matrix, tmpvec); + cross_v3_v3v3(raxis, obvec, tmpvec); + + if (normalize_v3(raxis) == 0.0f) { + return; + } + } + else if (norm < 0.1f) { + /* near 0 and Pi arcsin has way better precision than arccos */ + rangle = (rangle > M_PI_2) ? M_PI - asinf(norm) : asinf(norm); + } + + axis_angle_normalized_to_mat3(rmat, raxis, rangle); /* rotate the owner in the way defined by this rotation matrix, then reapply the location since * we may have destroyed that in the process of multiplying the matrix |