bevel error #45919

Closed
opened 2015-08-27 04:46:47 +02:00 by Dave Holloway · 15 comments

bevel error.jpg

win 8.1
gtx 750ti

In recent build bot builds I am getting bad result with the bevel tool around edge loops, not matching at the end.
More verts seems to increase the chance of it happening.

The c150645 build seemed ok.

![bevel error.jpg](https://archive.blender.org/developer/F228362/bevel_error.jpg) win 8.1 gtx 750ti In recent build bot builds I am getting bad result with the bevel tool around edge loops, not matching at the end. More verts seems to increase the chance of it happening. The c150645 build seemed ok.
Author

Changed status to: 'Open'

Changed status to: 'Open'
Author

Added subscriber: @daveh

Added subscriber: @daveh

Added subscriber: @Sergey

Added subscriber: @Sergey

Please always provide .blend and exact steps needed to reproduce the issue. So far i can not confirm the issue with own tests.

Please always provide .blend and exact steps needed to reproduce the issue. So far i can not confirm the issue with own tests.
Howard Trickey was assigned by Aaron Carlisle 2015-08-27 15:55:19 +02:00
Member

Added subscriber: @Blendify

Added subscriber: @Blendify
Author

Sorry, I should have given a more thorough explanation.

I simply added a circle.
Increased the verts to 64.
Extruded up, and filled with ngon.
Then bevel.

It seems to be forming a spiral, rather than a chamfer.

Sorry, I should have given a more thorough explanation. I simply added a circle. Increased the verts to 64. Extruded up, and filled with ngon. Then bevel. It seems to be forming a spiral, rather than a chamfer.
Member

I can confirm. For reference, here is a model built according to Dave's instructions:
bevspiral.blend

I did recently put some limits on a 'width adjustment' pass, which tries to deal with conflicting bevel requirements from adjacent vertices. This was to fix a problem where the bevel width grew hugely when there was a very long chain of edges. I would have thought no adjustment would be needed in this case, but clearly I am wrong. I will investigate. Thanks for the report.

I can confirm. For reference, here is a model built according to Dave's instructions: [bevspiral.blend](https://archive.blender.org/developer/F228511/bevspiral.blend) I did recently put some limits on a 'width adjustment' pass, which tries to deal with conflicting bevel requirements from adjacent vertices. This was to fix a problem where the bevel width grew hugely when there was a very long chain of edges. I would have thought no adjustment would be needed in this case, but clearly I am wrong. I will investigate. Thanks for the report.

Added subscriber: @tompov

Added subscriber: @tompov

I can confirm this.
It appeared between 24e56d4 and 58c9a0b
Perhaps with c1c07b6 ?
Tom

I can confirm this. It appeared between 24e56d4 and 58c9a0b Perhaps with c1c07b6 ? Tom
Member

P.S., What I had intended was that one could get the old unrestricted width adjustment path by turning on the 'clamp overlap' option, and indeed that does make for an even bevel here. The problem is that the clamp stops the beveling at a fairly small value and I don't understand why (well, I have a suspicion). Will have to also investigate that.

P.S., What I had intended was that one could get the old unrestricted width adjustment path by turning on the 'clamp overlap' option, and indeed that does make for an even bevel here. The problem is that the clamp stops the beveling at a fairly small value and I don't understand why (well, I have a suspicion). Will have to also investigate that.
Howard Trickey removed their assignment 2015-08-28 14:01:16 +02:00
Campbell Barton was assigned by Howard Trickey 2015-08-28 14:01:16 +02:00
Member

Added subscriber: @howardt

Added subscriber: @howardt
Member

Thanks Tom for suggesting the possible cause of the problem. That is indeed the change that is causing the problem.

Campbell, your change to make epsilon tests in isect_line_line_v3 is testing against too large an epsilon, I think.

In the case tickled here, the V1, V2, V2, V4 args are:
(-0.881921,0.471397,2.432115) (-0.932373,0.387223,2.432115) (-0.791522,0.428641,2.432115) (-0.833481,0.339928,2.432115)

which leads to these values for a, b, ab, and div in the isect_line_line_v3 code:
a=(-0.050452,-0.084174,0.000000) b=(-0.041958,-0.088713,0.000000) ab=(0.000000,0.000000,0.000944) div=0.000001

These are not collinear lines -- they have a 5.625 degree angle between them. Yet the test of div against the default epsilon in the lib calls them collinear. This is all kind of tricky, because we shouldn't be testing against absolute epsilons, since that makes modeling at a small scale hard. One suggested change is to use epsilon * epsilon as the thing to test against instead of epsilon, since the thing it is testing is the square of a length. Making that change fixes this bug, but thought I'd let you decide what to do here.

Of course I could explicitly pass in a much smaller epsilon from the bevel code, but this seems like a problem in the library that should be fixed there.

Thanks Tom for suggesting the possible cause of the problem. That is indeed the change that is causing the problem. Campbell, your change to make epsilon tests in isect_line_line_v3 is testing against too large an epsilon, I think. In the case tickled here, the V1, V2, V2, V4 args are: (-0.881921,0.471397,2.432115) (-0.932373,0.387223,2.432115) (-0.791522,0.428641,2.432115) (-0.833481,0.339928,2.432115) which leads to these values for a, b, ab, and div in the isect_line_line_v3 code: a=(-0.050452,-0.084174,0.000000) b=(-0.041958,-0.088713,0.000000) ab=(0.000000,0.000000,0.000944) div=0.000001 These are not collinear lines -- they have a 5.625 degree angle between them. Yet the test of div against the default epsilon in the lib calls them collinear. This is all kind of tricky, because we shouldn't be testing against absolute epsilons, since that makes modeling at a small scale hard. One suggested change is to use epsilon * epsilon as the thing to test against instead of epsilon, since the thing it is testing is the square of a length. Making that change fixes this bug, but thought I'd let you decide what to do here. Of course I could explicitly pass in a much smaller epsilon from the bevel code, but this seems like a problem in the library that should be fixed there.

Changed status from 'Open' to: 'Resolved'

Changed status from 'Open' to: 'Resolved'

Reverted the part of c1c07b68b8 which caused the problem

Applied an alternative fix to isect_tri_tri_epsilon_v3,
since using line-intersections wasn't very stable.
see 6a53e658d3)

Reverted the part of c1c07b68b8 which caused the problem Applied an alternative fix to `isect_tri_tri_epsilon_v3`, since using line-intersections wasn't very stable. see 6a53e658d3)
Member

Thanks, Campbell.
I confirm that this fixes the bevel problem that was the initial subject of this bug report, so agree that the issue is Resolved.

Thanks, Campbell. I confirm that this fixes the bevel problem that was the initial subject of this bug report, so agree that the issue is Resolved.
Sign in to join this conversation.
No Label
Interest
Alembic
Interest
Animation & Rigging
Interest
Asset Browser
Interest
Asset Browser Project Overview
Interest
Audio
Interest
Automated Testing
Interest
Blender Asset Bundle
Interest
BlendFile
Interest
Collada
Interest
Compatibility
Interest
Compositing
Interest
Core
Interest
Cycles
Interest
Dependency Graph
Interest
Development Management
Interest
EEVEE
Interest
EEVEE & Viewport
Interest
Freestyle
Interest
Geometry Nodes
Interest
Grease Pencil
Interest
ID Management
Interest
Images & Movies
Interest
Import Export
Interest
Line Art
Interest
Masking
Interest
Metal
Interest
Modeling
Interest
Modifiers
Interest
Motion Tracking
Interest
Nodes & Physics
Interest
OpenGL
Interest
Overlay
Interest
Overrides
Interest
Performance
Interest
Physics
Interest
Pipeline, Assets & IO
Interest
Platforms, Builds & Tests
Interest
Python API
Interest
Render & Cycles
Interest
Render Pipeline
Interest
Sculpt, Paint & Texture
Interest
Text Editor
Interest
Translations
Interest
Triaging
Interest
Undo
Interest
USD
Interest
User Interface
Interest
UV Editing
Interest
VFX & Video
Interest
Video Sequencer
Interest
Virtual Reality
Interest
Vulkan
Interest
Wayland
Interest
Workbench
Interest: X11
Legacy
Blender 2.8 Project
Legacy
Milestone 1: Basic, Local Asset Browser
Legacy
OpenGL Error
Meta
Good First Issue
Meta
Papercut
Meta
Retrospective
Meta
Security
Module
Animation & Rigging
Module
Core
Module
Development Management
Module
EEVEE & Viewport
Module
Grease Pencil
Module
Modeling
Module
Nodes & Physics
Module
Pipeline, Assets & IO
Module
Platforms, Builds & Tests
Module
Python API
Module
Render & Cycles
Module
Sculpt, Paint & Texture
Module
Triaging
Module
User Interface
Module
VFX & Video
Platform
FreeBSD
Platform
Linux
Platform
macOS
Platform
Windows
Priority
High
Priority
Low
Priority
Normal
Priority
Unbreak Now!
Status
Archived
Status
Confirmed
Status
Duplicate
Status
Needs Info from Developers
Status
Needs Information from User
Status
Needs Triage
Status
Resolved
Type
Bug
Type
Design
Type
Known Issue
Type
Patch
Type
Report
Type
To Do
No Milestone
No project
No Assignees
6 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: blender/blender#45919
No description provided.