1
0
mirror of https://git.dev.opencascade.org/repos/occt.git synced 2025-04-09 18:50:54 +03:00

3 Commits

Author SHA1 Message Date
msv
eca0539260 0028081: Fix regression in HLR introduced by the fix for 28053
The whole piece of the walking line between two vertices is no more checked for having length greater than Precision::Confusion(). Instead, only full duplicate points (comparing with gp::Resolution()) are removed, allowing the resulting line have any non-null length.

The test bugs modalg_6 bug27720_5 has been reverted to its original state.
2016-11-17 14:48:57 +03:00
msv
4af9e8a8b7 0028053: Regressions in HLR when FPE signals are enabled
Protect Contap_Contour from creation of null-length curves.
2016-11-08 19:29:56 +03:00
jgv
8d795b5130 0027720: HLRBrep_Algo BSpline missing edges
The algorithm that builds outlines ("silhouettes") makes an outline in 2d parametric space of the surface starting from some previously detected point where normal is orthogonal to direction of view. So, the surface is previously discretized into (m*n) sample points and some of them become starting points for future outlines.

    If the surface has non-uniform parametrization and/or some local extremums of curvature, the outlines can not be built without breaks, so there are several groups of consequent outlines in this case. Unfortunately, it leads to the situation when current number of sample points becomes insufficient to build all the parts of outlines.

    The idea is to detect the "holes" between already constructed parts of outlines and then complete the construction.

New auxiliary draw command for testing of HLR.

Correction according to the remarks.

Update of test case according to the developer's directive
2016-09-23 09:57:50 +03:00