When creating tile data include only triangles which have got intersection
with tile's rectangle only. This saves quite a lot of per-pixel iterations
through triangles which simply can not affect on current tile.
In fact, it's AABB check is used here. It could be improved further, but
it'll slowdown tile data generation with questionable speedup.
Another major slowdown is in fact caused by voronoi triangulation code.
Currently it's used naive algorithm which is O(N^2) where N is number
of edges. Added few euristics there and removed unused part of code, which
gave quite noticeable speedup already.
This could be improved further, but this node is not ment to be used for
lots of markers. It's also generates wrong triangulation when there're
many sites used. Need to be investigated further.
switch channel
This was caused by a typo, which would end up toggling the expanded status of
the Grease Pencil datablock (gpd->flag & 4) instead of the layer active status
(gpl->flag & 4).
Revert part of r48105 (calling mouse_mesh() in mouse_mesh_loop() is not a good idea, as it might select another element ouside the selected loop, and is anyway overkill!), added lighter code that simply checks for the nearest vertex/face of current edge.
Separate X and Y passes of blurring like it's done for flat
gaussian blur. This reduces computing difficulty from size^2
to 2*size without any visual changes in matte.
When removing a skin or multires modifier, it skips deletion of the
associated CustomData layer if the object has any other modifiers of
that type. This check has been extended to all objects that use the
object's data.
Similarly, deleting higher multires levels and multires subdivision
will not update the maximum level of any other multires modifiers on
objects that link to the same mesh.
Note that modifier_apply_obdata() doesn't need any changes as it
does not allow applying to multi-user data.
Object joining has also been modified to synchronize multires levels
objects that share a mesh. This is needed because joining can
subdivide or delete levels in order to match the maximum level of the
join-from object to the join-to object.
Fixes bug [#31880] instance multiresolution modifier error.
http://projects.blender.org/tracker/index.php?func=detail&aid=31880&group_id=9&atid=498
Reviewed by Sergey:
http://codereview.appspot.com/6332047/
Notes:
*This implements a quite simple algorithm, which simply checks angles (actually, absolute cosines) of created tri and remaining face (which may be a tri, quad, or more NGon), so that both are "best" (ie avoid as much as possible too much narrow/wide corners), and also checks the new edge is OK (i.e. does not goes "out" of original face).
*Incidently, it fixes a typo in that bm_face_goodline() func!
*It's quite performant (a bit quicker than previous code, as far as I have tested it) and prevent creation of completely flat triangles as much as possible, but it's far from being a "best" solution (as it is still a "progressive" one)!
*It also introduces a new math func (in BLI_math_vector.h), cos_v3v3v3, which computes cosine (ie dot product of normalized vectors) and is roughly a quicker replacement for angle_v3v3v3, when real angles are not needed.
In fact fixed in easiest way -- always re-calculate knots array
on topology changes.
After some discussion with Ton we agreed on that having manually
editable knots is not intuitive and user should only define
cyclic/endpoints flags and knots would be re-calculated based
on this flags.
This means that it's unnecessary to have special logic for knots
manipulating in some topology changing tools and they could just
re-calculate knots for the whole nurb, without worrying that knots
could have been manually edited.
Now these links are drawn in the same way as all others. Also they now display highlighting correctly when inserting a node into a link, making it easier to avoid unwanted insertions.
input image to be calculated in some cases, use only actual area which is
needed to calculate current tile.
Seems to be giving some % of speedup. Verified result of keying before
this patch and after this patch and they were identical, so hopefully
now area of interest is indeed correct.
Screens are usually doesn't have overexposured pixels and all
saturation / gradient math was written assuming that all channels
are withing 0 .. 1 range and in cases when some channel exceeds
this range matte could be completely wrong.
Added special check for overesposure and assume such pixels as
definitely foreground.
Also fixed minimal value for edge kernel size.
Right now this is being fixed by not allowing the exact same action (action, start/end frames, speed, etc) to be played if it's already playing. Hopefully this will not cause more issues than it solves.
a translate node were connected with the same complex node (like lens
distortion).
Added a check to see if the list of buffers are available to resolve
this issue.
