Non-determinism with manifold and minkowski #4617
Labels
Comments
|
ping @ochafik |
|
If you enable manifold deterministic option: diff --git a/src/openscad.cc b/src/openscad.cc
index f4646bbf3..8ed411d8a 100644
--- a/src/openscad.cc
+++ b/src/openscad.cc
@@ -51,6 +51,10 @@
#include <vector>
#include <fstream>
+#ifdef ENABLE_MANIFOLD
+#include "manifold.h"
+#endif
+
#ifdef ENABLE_CGAL
#include "CGAL_Nef_polyhedron.h"
@@ -918,6 +922,10 @@ int main(int argc, char **argv)
init_mimalloc();
#endif
+#if ENABLE_MANIFOLD
+ manifold::ManifoldParams().deterministic = true;
+#endif
+
int rc = 0;
StackCheck::inst();
and disable parallel minkowski by setting |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Describe the bug
I was testing a scad file, which makes heavy use of minkowski, on the new manifold feature in nightlies.
Every render is giving different results.
To Reproduce
Steps to reproduce the behavior:
openscad-nightly --enable manifold ./test.scad -o out.stlLog from terminal:
Expected behavior
determinism, if possible?
Code reproducing the issue (click to expand)
Environment and Version info (please complete the following information):
Library Info (click to expand)
Additional context
The script is meant to experiment with morphology operations and composing them into a generic "roundover" operation which can fillet both convex and concave edges, without affecting the bounding box of the object.
In 2D, we can use offset and these are trivial to implement. But in 3D the closest thing we have to offset is minkowski with a sphere.
Since
minkowskiaddition can't directly simulate a negative offset, a number of additional operations are done to act as geometric complement, operating on that with positive offsets, etc.I was originally testing with an octahedral geodesic sphere from a library of mine, but replaced with standard sphere so there would be no external dependencies. Near the top of the script the
usedirectives can be un-commented and the commented line insidesphere_for_offsetcan be swapped if interested in trying. The same types of issues result either way.The text was updated successfully, but these errors were encountered: