EWA sampling is designed for downsampling images, i.e. scaling down the size of
input image pixels, which happens regularly in compositing. While the standard
sampling methods (linear, cubic) work reasonably well for linear
transformations, they don't yield good results in non-linear cases like
perspective projection or arbitrary displacement. EWA sampling is comparable to
mipmapping, but avoids problems with discontinuities.
To work correctly the EWA algorithm needs partial derivatives of the mapping
functions which convert output pixel coordinates back into the input image
space (2x2 Jacobian matrix). With these derivatives the EWA algorithm
projects ellipses into the input space and accumulates colors over their
area. This calculation was not done correctly in the compositor, only the
derivatives du/dx and dv/dy were calculation, basically this means it only
worked for non-rotated input images.
The patch introduces full derivative calculations du/dx, du/dy, dv/dx, dv/dy for
the 3 nodes which use EWA sampling currently: PlaneTrackWarp, MapUV and
Displace. In addition the calculation of ellipsis area and axis-aligned
bounding boxes has been fixed.
For the MapUV and Displace nodes the derivatives have to be estimated by
evaluating the UV/displacement inputs with 1-pixel offsets, which can still have
problems on discontinuities and sub-pixel variations. These potential problems
can only be alleviated by more radical design changes in the compositor
functions, which are out of scope for now. Basically the values passed to the
UV/Displacement inputs would need to be associated with their 1st order
derivatives, which requires a general approach to derivatives in all nodes.
Now cut lines detect vertices that they pass (almost) exactly over
and snap to them, to avoid making verts vert close to other ones.
Added radius arg to BKE_bmbvh_ray_cast so that can detect an obscuring
face when the ray might otherwise go exactly between two triangles.
Needed an isect_line_tri_epsilon function for similar reason.
Fixes last part of bug #35002. Other knife bugs still present but
getting this commit in now before continuing bug fixing.
- plane_point_side_v3(), a bit like line_point_side_v2()
- isect_point_planes_v3(), moved from paint_hide.c
functions to convert between point/normal pairs.
- plane_from_point_normal_v3()
- plane_to_point_normal_v3()
- IsectLLPt2Df
- isect_point_quad_uv_v2
- isect_point_face_uv_v2
These are obsoleted by resolve_tri_uv, resolve_quad_uv
also add attributes for unused function results for some math functions.
Now we do simple triangulation and calculate signed area of triangles to
account for concave polygons.
This only works correct for planar polygons but gives better results
overall.
projecting it. The original paper suggests to simply interpolate between
the two points of an edge if the distance of the point to that edge is
smaller than a threshold.
* Fixed both 3D and 2D code to utilize this. Possibly other places in
blender where this scaling is done will have to be adjusted.
* Changed vertex interpolation to use 2D interpolation, since it already
did projection on plane and 2d calculations are faster.
* Also added notifier on hard recalc when uvcalc_transfor_correction is
used. Results in instant feedback on UV editor when edge sliding.
