3D print Toolbox Intersections gives error #37583

New Issue

Mikhail Rachinskiy · 2013-11-22T17:35:21+01:00

2013-11-22 17:35:21 +01:00

Blender Version
Broken: Blender 2.69 ac021a4 Build Bot Official

Short description of error
Intersections check breaks on Ngons

Also I found that Intersections falsely triggered on particular models (blend file attached)

Exact steps for others to reproduce the error
Enable the Mesh: 3D Print Toolbox addon.
Open attached blend file, and press intersections button (falsely triggered)
intersect.blend

Create a Cylinder primitive and press the Intersections button. (gives eror)

Traceback (most recent call last):
  File "C:\Users\Mikhail\Downloads\26\2.69\scripts\addons\object_print3d_utils\o
perators.py", line 163, in execute
    return execute_check(self, context)
  File "C:\Users\Mikhail\Downloads\26\2.69\scripts\addons\object_print3d_utils\o
perators.py", line 117, in execute_check
    self.main_check(obj, info)
  File "C:\Users\Mikhail\Downloads\26\2.69\scripts\addons\object_print3d_utils\o
perators.py", line 158, in main_check
    faces_intersect = mesh_helpers.bmesh_check_self_intersect_object(obj)
  File "C:\Users\Mikhail\Downloads\26\2.69\scripts\addons\object_print3d_utils\m
esh_helpers.py", line 124, in bmesh_check_self_intersect_object
    face_map_index = {i: face_map_index_org[face_map.get(f, f)] for i, f in enum
erate(bm.faces)}
  File "C:\Users\Mikhail\Downloads\26\2.69\scripts\addons\object_print3d_utils\m
esh_helpers.py", line 124, in <dictcomp>
    face_map_index = {i: face_map_index_org[face_map.get(f, f)] for i, f in enum
erate(bm.faces)}
KeyError: <BMFace(0x0000000009F2F4C8), index=20, totverts=3>

Updated steps for others to reproduce the error

import bpy, bmesh
bpy.ops.mesh.primitive_cylinder_add(vertices=5)
bm = bmesh.new()
bm.from_mesh(bpy.context.object.data)
face_map_index_org = {f: i for i, f in enumerate(bm.faces)}
ret = bmesh.ops.triangulate(bm, faces=bm.faces)
face_map = ret["face_map"]
face_map_index = {i: face_map_index_org[face_map.get(f, f)] for i, f in enumerate(bm.faces)}

**Blender Version** Broken: Blender 2.69 ac021a4 Build Bot Official **Short description of error** Intersections check breaks on Ngons Also I found that Intersections falsely triggered on particular models (blend file attached) **Exact steps for others to reproduce the error** Enable the *Mesh: 3D Print Toolbox* addon. Open attached blend file, and press intersections button (falsely triggered) [intersect.blend](https://archive.blender.org/developer/F29986/intersect.blend) Create a Cylinder primitive and press the Intersections button. (gives eror) ``` Traceback (most recent call last): File "C:\Users\Mikhail\Downloads\26\2.69\scripts\addons\object_print3d_utils\o perators.py", line 163, in execute return execute_check(self, context) File "C:\Users\Mikhail\Downloads\26\2.69\scripts\addons\object_print3d_utils\o perators.py", line 117, in execute_check self.main_check(obj, info) File "C:\Users\Mikhail\Downloads\26\2.69\scripts\addons\object_print3d_utils\o perators.py", line 158, in main_check faces_intersect = mesh_helpers.bmesh_check_self_intersect_object(obj) File "C:\Users\Mikhail\Downloads\26\2.69\scripts\addons\object_print3d_utils\m esh_helpers.py", line 124, in bmesh_check_self_intersect_object face_map_index = {i: face_map_index_org[face_map.get(f, f)] for i, f in enum erate(bm.faces)} File "C:\Users\Mikhail\Downloads\26\2.69\scripts\addons\object_print3d_utils\m esh_helpers.py", line 124, in <dictcomp> face_map_index = {i: face_map_index_org[face_map.get(f, f)] for i, f in enum erate(bm.faces)} KeyError: <BMFace(0x0000000009F2F4C8), index=20, totverts=3> ``` **Updated steps for others to reproduce the error** ``` import bpy, bmesh bpy.ops.mesh.primitive_cylinder_add(vertices=5) bm = bmesh.new() bm.from_mesh(bpy.context.object.data) face_map_index_org = {f: i for i, f in enumerate(bm.faces)} ret = bmesh.ops.triangulate(bm, faces=bm.faces) face_map = ret["face_map"] face_map_index = {i: face_map_index_org[face_map.get(f, f)] for i, f in enumerate(bm.faces)}

Mikhail Rachinskiy commented

2013-11-22 17:35:21 +01:00

Changed status to: 'Open'

Campbell Barton was assigned by Mikhail Rachinskiy

2013-11-22 17:35:21 +01:00

Mikhail Rachinskiy commented

2013-11-22 17:35:21 +01:00

Added subscriber: @MikhailRachinskiy

blender-admin commented

2013-11-22 17:35:21 +01:00

#38241 was marked as duplicate of this issue

blender-admin commented

2013-11-22 17:35:21 +01:00

#39037 was marked as duplicate of this issue

Campbell Barton removed their assignment 2013-11-25 07:23:01 +01:00

Campbell Barton commented

2013-11-25 07:24:25 +01:00

Added subscribers: @(Deleted), @ideasman42

Campbell Barton commented

2013-11-25 07:24:25 +01:00

@dalai_testing, The Python exception looks like its caused by recent changes to triangulate (since 2.69 release), looks like mapping isn't stored correctly anymore.

I'll check on the incorrect intersections in intersect.blend

@dalai_testing, The Python exception looks like its caused by recent changes to triangulate (since 2.69 release), looks like mapping isn't stored correctly anymore. I'll check on the incorrect intersections in `intersect.blend`

Dalai Felinto self-assigned this 2013-11-25 17:25:41 +01:00

Dalai Felinto commented

2013-11-25 17:25:41 +01:00

Removed subscriber: @(Deleted)

Dalai Felinto commented

2013-11-25 19:16:29 +01:00

Simpler way to reproduce the problem:

import bpy, bmesh
bpy.ops.mesh.primitive_cylinder_add()
bm = bmesh_new()
bm.from_mesh(bpy.context.object.data)
face_map_index_org = {f: i for i, f in enumerate(bm.faces)}
ret = bmesh.ops.triangulate(bm, faces=bm.faces)
face_map = ret["face_map"]
face_map_index = {i: face_map_index_org[face_map.get(f, f)] for i, f in enumerate(bm.faces)}

Simpler way to reproduce the problem: ``` import bpy, bmesh bpy.ops.mesh.primitive_cylinder_add() bm = bmesh_new() bm.from_mesh(bpy.context.object.data) face_map_index_org = {f: i for i, f in enumerate(bm.faces)} ret = bmesh.ops.triangulate(bm, faces=bm.faces) face_map = ret["face_map"] face_map_index = {i: face_map_index_org[face_map.get(f, f)] for i, f in enumerate(bm.faces)} ```

Dalai Felinto commented

2013-11-25 21:23:15 +01:00

@ideasman42 can you land me a hand here?

Check the file: tri-print.blend (open it, run the script in the Text Editor and check the console).

The code basically returns:

original faces:

0 <BMFace(0x10716b218), index=0, totverts=5>


new faces:

0 <BMFace(0x10716b218), index=0, totverts=3>
1 <BMFace(0x10716b250), index=1, totverts=3>
2 <BMFace(0x10716b288), index=2, totverts=3>


dead faces:

{<BMFace dead at 0x10fa138a0>: 0}

Which means that the original stored face pointer is still valid, but the value stored in face_map_index_org changes somehow. That's what is producing the crash. But I'm a bit clueless on what can be causing that.

@ideasman42 can you land me a hand here? Check the file: [tri-print.blend](https://archive.blender.org/developer/F32966/tri-print.blend) (open it, run the script in the Text Editor and check the console). The code basically returns: ``` original faces: ``` -------------- 0 <BMFace(0x10716b218), index=0, totverts=5> ``` new faces: ``` --------- 0 <BMFace(0x10716b218), index=0, totverts=3> 1 <BMFace(0x10716b250), index=1, totverts=3> 2 <BMFace(0x10716b288), index=2, totverts=3> ``` dead faces: ``` ---------- ``` {<BMFace dead at 0x10fa138a0>: 0} ``` Which means that the original stored face pointer is still valid, but the value stored in **face_map_index_org** changes somehow. That's what is producing the crash. But I'm a bit clueless on what can be causing that.

Dalai Felinto commented

2013-11-25 23:39:49 +01:00

@ideasman42 I added a workaround here: http://developer.blender.org/D47

Dalai Felinto commented

2013-11-25 23:40:32 +01:00

I mean, D47 is a workaround (I shouldn't have pasted the entire link, only the number is fine)

I mean, [D47](https://archive.blender.org/developer/D47) is a workaround (I shouldn't have pasted the entire link, only the number is fine)

Dalai Felinto commented

2013-11-26 04:03:24 +01:00

Apparently the problem is that bmesh_face_swap_data and BM_face_kill are mangling the variables of f and f_new in bmesh_polygon.c.

Basically bpy_bm_generic_invalidate is being called for f_new, but &BPy_BMGeneric is the same for both f and f_new, so f ends up with an invalid BPy_BMGeneric->bm (even though BPy_BMGeneric->f is valid).

See the backtrace I get up to bpy_bm_generic_invalidate is called:
http://www.pasteall.org/47617

I tried a soft kill face solution - D50 just for testing, and although the original face no longer shows as 'dead', we still get a Key not found error. We could solve that in the addon if we stored str(f) instead of f as key, but if we do that we may as well do as in D47 and spare Blender core from being patched.

Apparently the problem is that **bmesh_face_swap_data** and **BM_face_kill** are mangling the variables of *f* and *f_new* in bmesh_polygon.c. Basically **bpy_bm_generic_invalidate** is being called for f_new, but &*BPy_BMGeneric* is the same for both *f* and *f_new*, so f ends up with an invalid BPy_BMGeneric->bm (even though BPy_BMGeneric->f is valid). See the backtrace I get up to **bpy_bm_generic_invalidate** is called: http://www.pasteall.org/47617 I tried a soft kill face solution - [D50](https://archive.blender.org/developer/D50) just for testing, and although the original face no longer shows as 'dead', we still get a Key not found error. We could solve that in the addon if we stored **str(f)** instead of **f** as key, but if we do that we may as well do as in [D47](https://archive.blender.org/developer/D47) and spare Blender core from being patched.

Campbell Barton commented

2013-11-26 21:48:14 +01:00

Added subscriber: @dfelinto

Campbell Barton commented

2013-11-26 21:48:14 +01:00

@dfelinto,
Simply not-swapping customdata seems to work.

adding this line at the end of bmesh_face_swap_data:
SWAP(void *, f_a->head.data, f_b->head.data);

@dfelinto, Simply not-swapping customdata seems to work. adding this line at the end of `bmesh_face_swap_data`: `SWAP(void *, f_a->head.data, f_b->head.data);`

blender-admin commented

2013-11-26 22:11:34 +01:00

This issue was referenced by blender/blender-addons-contrib@3b24eaad9a

This issue was referenced by blender/blender-addons-contrib@3b24eaad9af8025cb38879cbc345f7debceba5a9

blender-admin commented

2013-11-26 22:11:34 +01:00

This issue was referenced by blender/blender@3b24eaad9a

This issue was referenced by blender/blender@3b24eaad9af8025cb38879cbc345f7debceba5a9

Campbell Barton commented

2013-11-26 22:12:34 +01:00

Changed status from 'Open' to: 'Resolved'

Campbell Barton closed this issue

2013-11-26 22:12:34 +01:00

Campbell Barton commented

2013-11-26 22:12:34 +01:00

Closed by commit blender/blender@3b24eaad9a.

Mikhail Rachinskiy commented

2013-12-03 03:54:30 +01:00

Hey Campbell,

intersections are good with Ngons now, but they are still falsely triggered with attached model.
Could you please look in to it?

Hey Campbell, intersections are good with Ngons now, but they are still falsely triggered with attached model. Could you please look in to it?

Campbell Barton commented

2013-12-03 04:02:29 +01:00

@MikhailRachinskiy, sure but in future prefer to open a new report (since this is a different issue).

Campbell Barton commented

2013-12-03 04:02:45 +01:00

Changed status from 'Resolved' to: 'Open'

Campbell Barton reopened this issue

2013-12-03 04:02:45 +01:00

Dalai Felinto removed their assignment 2013-12-03 05:05:32 +01:00

Campbell Barton was assigned by Sergey Sharybin

2013-12-04 15:49:50 +01:00

Sergey Sharybin commented

2013-12-04 15:49:50 +01:00

Added subscriber: @Sergey

Sergey Sharybin commented

2013-12-04 15:49:50 +01:00

@ideasman42, will you look into this?

Mikhail Rachinskiy commented

2013-12-11 06:46:28 +01:00

Hey guys!

In the meantime I have found what triggers false intersections, it's geometry distance from the origin, yep.
As soon as you move geometry closer to the origin, intersections wont be triggered anymore.

To reproduce it, you need to:

create a sphere
enter edit mode
move sphere geometry 20 units away
make intersections check (resulting in 94 intersections)

Hey guys! In the meantime I have found what triggers false intersections, it's geometry distance from the origin, yep. As soon as you move geometry closer to the origin, intersections wont be triggered anymore. To reproduce it, you need to: 1. create a sphere 2. enter edit mode 3. move sphere geometry 20 units away 4. make intersections check (resulting in 94 intersections)

Mikhail Rachinskiy commented

2013-12-13 09:12:04 +01:00

Guys?

Bastien Montagne commented

2014-03-08 15:20:06 +01:00

◀ Merged tasks: #38241.

Bastien Montagne commented

2014-03-08 15:20:33 +01:00

Added subscriber: @mont29

Bastien Montagne commented

2014-03-08 15:20:33 +01:00

◀ Merged tasks: #39037.

Campbell Barton commented

2014-05-12 16:31:12 +02:00

Changed status from 'Open' to: 'Resolved'

Campbell Barton closed this issue

2014-05-12 16:31:12 +02:00

Campbell Barton commented

2014-05-12 16:31:12 +02:00

Closed by commit blender/blender-addons-contrib@3b24eaad9a.

Mikhail Rachinskiy commented

2014-05-13 08:10:40 +02:00

Why this task marked as resolved while issue is still appears?

Sergey Sharybin commented

2014-05-13 09:54:38 +02:00

Changed status from 'Resolved' to: 'Open'

Sergey Sharybin reopened this issue

2014-05-13 09:54:38 +02:00

Sergey Sharybin commented

2014-05-13 09:54:38 +02:00

We had a bit of an accident mentioned there: http://lists.blender.org/pipermail/bf-committers/2014-May/043538.html Some of the re-opened tasks might be closed now unfortunately.

Re-opening it again. Sorry for the confusion.

We had a bit of an accident mentioned there: http://lists.blender.org/pipermail/bf-committers/2014-May/043538.html Some of the re-opened tasks might be closed now unfortunately. Re-opening it again. Sorry for the confusion.

Mikhail Rachinskiy commented

2014-05-19 08:59:29 +02:00

Please fix this bug before 2.71 release, it's crucial for 3D printing.

Dalai Felinto commented

2014-05-22 15:31:29 +02:00

This seems to be fixed here. Which version did you use to check it? Get a fresh copy of Blender from builder.blender.org

Mikhail Rachinskiy commented

2014-05-22 20:42:55 +02:00

Buildbot win64 90db85a, May 22 2014 - bug is still appears.

#38241 gives a prettty good description and blend file to test this out, but it has been marked as a duplicate to this task.

Buildbot win64 90db85a, May 22 2014 - bug is still appears. #38241 gives a prettty good description and blend file to test this out, but it has been marked as a duplicate to this task.

Campbell Barton commented

2014-08-11 03:19:33 +02:00

Changed status from 'Open' to: 'Resolved'

Campbell Barton closed this issue

2014-08-11 03:19:33 +02:00

Campbell Barton commented

2014-08-11 03:19:33 +02:00

Closing this task then, leave #38241.

Sign in to join this conversation.