Only search projects.blender.org
Log In
New Account
Home
My Page
Projects
Blender 2.x BF release
Summary
Activity
Tracker
SCM
Files
Blender 2.6 Bug Tracker: Browse
[#26035] Particles + raytrace BVH crash
Date:
2011-02-11 12:05
Priority:
3
State:
Closed
Submitted by:
Vijay Sridhara (
vsridhara
)
Assigned to:
Brecht Van Lommel (blendix)
Category:
Rendering
Status:
Fixed / Closed
Relates to:
Duplicates:
Patches:
Summary:
Particles + raytrace BVH crash
Detailed description
Hi,
I attached a coke can model, which upon rendering crashes Blender. I can repeat it every time.
Please check
Thanks
Vijay S
Followup
Message
Date
: 2011-02-11 14:32
Sender
:
Shinsuke Irie
This crash looks the same as #25871.
blender2.5: /tmp/buildd/blender2.5-2.56.1+svn34773/source/blender/render/intern/raytrace/reorganize.h:504: void VBVH_optimalPackSIMD<Node, TestCost>::calc_costs(Node*) [with Node = OVBVHNode, TestCost = PackCost]: Assertion `node->cut_cost[0] != (__builtin_inff())' failed.
Aborted
Date
: 2011-02-11 15:41
Sender
:
Campbell Barton
this is a duplicate or 25871, note that this only crashes with debug builds. you could do a release build instead as a workaround.
Date
: 2011-02-19 13:41
Sender
:
Ton Roosendaal
Not a duplicate, the 25871 one was because of displace texture, this file doesn't have displace.
Brecht: for me it doesn't crash, but keeps hanging for ever. When I reduce subdiv levels on can from 4 to 2 it goes fine.
Date
: 2011-02-23 10:55
Sender
:
Campbell Barton
looked into this further, particle systems seems at fault, attached a stripped down VSCoke12.blend
submitted a different report: 26190 for particle crash.
Some of the duplis are being instanced at nan,nan,nan location and it looks like this causes a raytracer assert?
we could just have the raytracer ignore objects which fail an 'isfinite' test in their transformation matrix but particles should be fixed too ofcourse.
Date
: 2011-02-23 13:01
Sender
:
Campbell Barton
jhk fixed the bug which causes the particles to have nan locations and crash however it would be good to fix this since its quite possible to make nan object transformations some other way.
for some reason the nan values only crash with duplis and not when expanded into normal objects.
attached a fix, but this is more an example then a propper fix since there are other places nan's may get into object matrices.
it does however fix the crash in the blendfile attached.
though I think a better fix would be to ignore any object or dupli with an invalid matrix.
Date
: 2011-03-12 00:14
Sender
:
Brecht Van Lommel
Fix in svn, added to bounding box checks where it already did similar things, so it's nicely centralized in raytree building code.
Attached Files:
Name
Date
Download
VSCoke.blend
2011-02-11 12:05
Download
VSCoke12.blend
2011-02-23 10:55
Download
26035_raytrace_fix.diff
2011-02-23 13:01
Download
Changes:
Field
Old Value
Date
By
status_id
Open
2011-03-12 00:14
blendix
close_date
None
2011-03-12 00:14
blendix
Status
Reopened
2011-03-12 00:14
blendix
File Added
15121: 26035_raytrace_fix.diff
2011-02-23 13:01
campbellbarton
File Added
15120: VSCoke12.blend
2011-02-23 10:55
campbellbarton
status_id
Closed
2011-02-19 13:41
ton
close_date
2011-02-11 15:41
2011-02-19 13:41
ton
assigned_to
none
2011-02-19 13:41
ton
summary
This blend file is crashing multiple times while rendering
2011-02-19 13:41
ton
Status
Rejected / Closed
2011-02-19 13:41
ton
status_id
Open
2011-02-11 15:41
campbellbarton
close_date
None
2011-02-11 15:41
campbellbarton
Status
New
2011-02-11 15:41
campbellbarton
File Added
14964: VSCoke.blend
2011-02-11 12:05
vsridhara
