randomly boolean behaviour #30565

New Issue

Tim L · 2012-03-16T17:13:11+01:00

Tim L commented

2012-03-16 17:13:11 +01:00

%%%1) am an end-user issues with Blender binaries

it is a real bug
it is intended to work by development
it is repeatable consistently
Checked on another computer (by rking on irc://freenode/blender).
previous release dose not have this problem.
Report
OS version=OSX 10.7.3
3D card=AMD Radeon HD 6770M, 512MB
RAM=16GB
CPU=i5 4 Core 2.7 GHz
Blender version(s) you tested=

osx-2.61=parts of the boolean don't work randomly
windows-SVN=fail to open file,
windows-2.62=fail to open file,
windows-2.61=works

minimum instructions:

new file
make 2 more cubes and 2 spheres
move all objects to 0,0,0
make the cubes .5x.5x1 .5x1x.5 1x.5x.5
make the second sphere .88x.88x.88
apply a boolean difference on the first sphere to all other objects

must change the rotation of cube 3 to 0.0001 for it to work
must change size of second sphere to .8x.8x.8 for it to work

%%%

%%%1) am an end-user issues with Blender binaries 2) it is a real bug it is intended to work by development it is repeatable consistently Checked on another computer (by rking on irc://freenode/blender). previous release dose not have this problem. 3) Report OS version=OSX 10.7.3 3D card=AMD Radeon HD 6770M, 512MB RAM=16GB CPU=i5 4 Core 2.7 GHz Blender version(s) you tested= ``` osx-2.61=parts of the boolean don't work randomly windows-SVN=fail to open file, windows-2.62=fail to open file, windows-2.61=works ``` minimum instructions: ``` new file make 2 more cubes and 2 spheres move all objects to 0,0,0 make the cubes .5x.5x1 .5x1x.5 1x.5x.5 make the second sphere .88x.88x.88 apply a boolean difference on the first sphere to all other objects ``` must change the rotation of cube 3 to 0.0001 for it to work must change size of second sphere to .8x.8x.8 for it to work %%%

Tim L commented

2012-03-16 17:13:11 +01:00

Changed status to: 'Open'

Brecht Van Lommel commented

2012-03-19 14:50:07 +01:00

%%%We've got a new boolean implementation, which may succeed/fail in different situations than the previous one. I get this message in the console:

WARNING! <-0.50000010838224251408235,-0.449834598124567386356887,-0.560210702961086992601736>==<-0.50000010838224251408235,-0.449834598124567386356887,-0.560210702961086992601736>
WARNING! <-0.50000010838224251408235,-0.449834598124567386356887,-0.560210702961086992601736>==<-0.50000010838224251408235,-0.449834598124567386356887,-0.560210702961086992601736>
uncontained hole loop does not share vertices with any face loop!
CSG failed, exception /Users/brecht/dev/git-blender/blender/extern/carve/lib/triangulator.cpp:899 "didn't manage to link up hole!"
Unknown internal error in boolean

After doing Remove Doubles it works better, the corners on the inside of this mesh seems to have quite a few duplicate vertices which may be confusing the boolean code. Assigned to Sergey, not sure if this is considered a bug.%%%

%%%We've got a new boolean implementation, which may succeed/fail in different situations than the previous one. I get this message in the console: WARNING! <-0.50000010838224251408235,-0.449834598124567386356887,-0.560210702961086992601736>==<-0.50000010838224251408235,-0.449834598124567386356887,-0.560210702961086992601736> WARNING! <-0.50000010838224251408235,-0.449834598124567386356887,-0.560210702961086992601736>==<-0.50000010838224251408235,-0.449834598124567386356887,-0.560210702961086992601736> uncontained hole loop does not share vertices with any face loop! CSG failed, exception /Users/brecht/dev/git-blender/blender/extern/carve/lib/triangulator.cpp:899 "didn't manage to link up hole!" Unknown internal error in boolean After doing Remove Doubles it works better, the corners on the inside of this mesh seems to have quite a few duplicate vertices which may be confusing the boolean code. Assigned to Sergey, not sure if this is considered a bug.%%%

Sergey Sharybin commented

2012-03-19 20:10:27 +01:00

%%%It's a failure of edge classifier. Me and Tobias are aware of this and issue and it was already reported here and assigned to me. Will ask Tobias if he've got some progress on resolving crash on windows (in fact it's assert abort in stl) and if not will add workaround which should make it behave in the same way as linux.
Noticed another issue which is related on shading. Its' a regression caused by rev44960. Already working on this.
Anyway, thanks for the report, closing.%%%

%%%It's a failure of edge classifier. Me and Tobias are aware of this and issue and it was already reported here and assigned to me. Will ask Tobias if he've got some progress on resolving crash on windows (in fact it's assert abort in stl) and if not will add workaround which should make it behave in the same way as linux. Noticed another issue which is related on shading. Its' a regression caused by rev44960. Already working on this. Anyway, thanks for the report, closing.%%%

Sergey Sharybin commented

2012-03-19 20:10:28 +01:00

Changed status from 'Open' to: 'Archived'

Sergey Sharybin closed this issue

2012-03-19 20:10:28 +01:00

Sean Roach commented

2012-03-20 00:21:33 +01:00

%%%I think I ran into the same problem from another angle, and felt it worth registering for.

I just tried to make a groin arch, and while boolean/difference modifier DID work, it produced a Very Messy set of vertices. I ended up with multiple vertices in spots.
Earlier, when I tried to create the same arch, but had my overlapping cylinder meshes the same size as the cube I was modifying, it crashed. Increasing the length it protruded from the cube on 5 sides, from exactly zero to a positive number fixed that.

I don't know if this will matter much longer, or if ngons will render it a moot point.

I have the .blend with the dirty vertices still.%%%

%%%I think I ran into the same problem from another angle, and felt it worth registering for. I just tried to make a groin arch, and while boolean/difference modifier DID work, it produced a Very Messy set of vertices. I ended up with multiple vertices in spots. Earlier, when I tried to create the same arch, but had my overlapping cylinder meshes the same size as the cube I was modifying, it crashed. Increasing the length it protruded from the cube on 5 sides, from exactly zero to a positive number fixed that. I don't know if this will matter much longer, or if ngons will render it a moot point. I have the .blend with the dirty vertices still.%%%

Sign in to join this conversation.

No Label

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

randomly boolean behaviour #30565