[Geometry] Bugfixes and add unit tests#5987
Open
fredroy wants to merge 9 commits intosofa-framework:masterfrom
Open
[Geometry] Bugfixes and add unit tests#5987fredroy wants to merge 9 commits intosofa-framework:masterfrom
fredroy wants to merge 9 commits intosofa-framework:masterfrom
Conversation
fredroy
added
pr: fix
Contributor
Author
|
[ci-build][with-all-tests] |
fredroy
force-pushed
the
add_tests_geometry
branch
from
ae273ee to
69cba4f
Compare
bakpaul
reviewed
Mar 26, 2026
| return false; | ||
|
|
||
| const auto Ln0p0 = sofa::type::dot(AB, AC); | ||
| if (Ln0p0 < 0 || Ln0p0 > Ln0n1) // out of bounds [n0; n1] |
Contributor
There was a problem hiding this comment.
Funny that the AI didn't catch this, but here because AB is not normalized the check is wrong.
Suggested change
| if (Ln0p0 < 0 || Ln0p0 > std::pow(Ln0n1,2) ) // out of bounds [n0; n1] |
Contributor
Author
There was a problem hiding this comment.
but Ln0n1 = sofa::type::dot(AB, AB); so I dont see why you do std::pow(Ln0n1,2)....
Comment on lines
+94
to
+113
| // In 3D, use signed areas via dot product with triangle normal | ||
| // to get correct sign for outside-triangle points | ||
| const auto N = sofa::type::cross(n1 - n0, n2 - n0); | ||
| const auto NdotN = sofa::type::dot(N, N); | ||
|
|
||
| if (NdotN < std::numeric_limits<T>::epsilon() * std::numeric_limits<T>::epsilon()) // triangle is flat | ||
| { | ||
| return sofa::type::Vec<3, T>(-1, -1, -1); | ||
| } | ||
|
|
||
| sofa::type::Vec<3, T> baryCoefs(type::NOINIT); | ||
| baryCoefs[0] = sofa::type::dot(N, sofa::type::cross(n2 - n1, p0 - n1)) / NdotN; | ||
| baryCoefs[1] = sofa::type::dot(N, sofa::type::cross(n0 - n2, p0 - n2)) / NdotN; | ||
| baryCoefs[2] = 1 - baryCoefs[0] - baryCoefs[1]; | ||
|
|
||
| if (fabs(baryCoefs[2]) <= std::numeric_limits<T>::epsilon()){ | ||
| baryCoefs[2] = 0; | ||
| } | ||
|
|
||
| return baryCoefs; |
Contributor
There was a problem hiding this comment.
Ok, before this the barycentric mapping was a bit wrong when the point wasn't on the triangle. Because of 104, the value of the third barycentric coordinate was always underestimated because the other ones where overestimated.
Contributor
There was a problem hiding this comment.
@epernod this fix the cases when the point is not on the same plane as the triangle. But does it ever happen ? Anyway I think that the api should not expect anything and check because otherwise we would rely on the user to make the check before.
fredroy
force-pushed
the
add_tests_geometry
branch
from
fcc869f to
9cd1ebb
Compare
if (denominator < EQUALITY_THRESHOLD) the denominator is a signed dot product. When negative (edge crosses plane from the "back"), this is always true, so the function rejects ~half of all valid intersections.
1-The 2D path computes alpha (parameter on edge AB) and checks 0 <= alpha <= 1, but never computes or checks beta (parameter on edge CD). Reports false intersections when the infinite line CD crosses segment AB but outside segment CD. 2-if (alphaDenom < std::numeric_limits<T>::epsilon()) same pattern as Bug 1. The cross product alphaDenom is signed. When negative (non-collinear edges in one orientation), incorrectly reports them as collinear.
solveLCP returns true when solver fails to converge (max iterations reached)
While isPointInTriangle still works (it only checks sign pattern), any code that uses the returned coordinates for interpolation, projection to the nearest point on the triangle, or distance computation gets wrong values. This is a public utility function likely consumed by many parts of the codebase.
…o bypass the plane check (as before)
…or for beta) Co-authored-by: Paul Baksic <30337881+bakpaul@users.noreply.github.com>
fredroy
force-pushed
the
add_tests_geometry
branch
from
9cd1ebb to
5a274e4
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fix various Edge and triangles functions
and more unit tests for Sofa.Geometry module
[with-all-tests]
By submitting this pull request, I acknowledge that
I have read, understand, and agree SOFA Developer Certificate of Origin (DCO).
Reviewers will merge this pull-request only if