Bug: Endpoint curve node version compatibilty #107433

Closed
opened 2023-04-28 13:29:46 +02:00 by Arye Ramaty · 3 comments
Contributor

System Information
Operating system: Linux-5.19.0-40-generic-x86_64-with-glibc2.35 64 Bits
Graphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 510.108.03

Blender Version
Broken: version: 3.5.0 Alpha, branch: master, commit date: 2022-11-21 01:31, hash: rB7a6cdeb24279
Worked: version: 3.5.0 Alpha, branch: arcpatch-D16471, commit date: 2022-11-18 20:37, hash: rB48cfe11fa60d

Behavior changed in e83f46ea76.

Short description of error
The GN setup functions differently in versions 3.41 and 3.5. After gradually freezing the GN setup and adding it as mesh data, I discovered that one of these three nodes - Capture Attribute, Endpoint Selection, or Mesh to Curve - is the cause. The images attached illustrate that the same node setup results in a different number of vertices.

3.4 3.5
GN_Bug_3.41.png GN_Bug_3.5.png

Exact steps for others to reproduce the error
Download the attached blend file,
Open it in Blender 3.4 and 3.5.
See the differences in behavior.
[Based on the default startup or an attached .blend file (as simple as possible)]
GN bug report.blend

**System Information** Operating system: Linux-5.19.0-40-generic-x86_64-with-glibc2.35 64 Bits Graphics card: NVIDIA GeForce RTX 2060/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 510.108.03 **Blender Version** Broken: version: 3.5.0 Alpha, branch: master, commit date: 2022-11-21 01:31, hash: `rB7a6cdeb24279` Worked: version: 3.5.0 Alpha, branch: arcpatch-D16471, commit date: 2022-11-18 20:37, hash: `rB48cfe11fa60d` Behavior changed in e83f46ea7630163ae836f04cff867eee99032efa. **Short description of error** The GN setup functions differently in versions 3.41 and 3.5. After gradually freezing the GN setup and adding it as mesh data, I discovered that one of these three nodes - Capture Attribute, Endpoint Selection, or Mesh to Curve - is the cause. The images attached illustrate that the same node setup results in a different number of vertices. | 3.4 | 3.5 | | -- | -- | | ![GN_Bug_3.41.png](/attachments/e71fadba-d550-4e50-8a8b-254448d7e3a1) | ![GN_Bug_3.5.png](/attachments/f35f4012-bbce-4d8e-844d-5bbfa7de803f) | **Exact steps for others to reproduce the error** Download the attached blend file, Open it in Blender 3.4 and 3.5. See the differences in behavior. [Based on the default startup or an attached .blend file (as simple as possible)] [GN bug report.blend](/attachments/aecf7f94-f1ff-4e78-a42a-25165014297a)
Arye Ramaty added the
Severity
Normal
Type
Report
Status
Needs Triage
labels 2023-04-28 13:29:47 +02:00
3.4 3.6
image image
| 3.4 | 3.6 | | -- | -- | | ![image](/attachments/a5b0b512-75a6-4390-ba80-9c25dc5e07c1) | ![image](/attachments/5632b801-3a0d-4345-bc9f-64b7dd291539) |
Member

Behavior changed in e83f46ea76.

Since this was a "identical results compared to the BMesh version" commit, will raise prio to check results again.

CC @wannes.malfait
CC @HooglyBoogly

Behavior changed in e83f46ea7630163ae836f04cff867eee99032efa. Since this was a "identical results compared to the BMesh version" commit, will raise prio to check results again. CC @wannes.malfait CC @HooglyBoogly
Member

I don't think this is actually a "bug". Let me explain why:

I did some research into what exactly was causing the difference in output. Since the relevant commit was found to be a refactor of the edge split node, the problem should occur there. This is what I found:

  • The output of the "new" edge split node is correct: I manually checked that each edge that should get split gets split, and that the number of vertices and edges is correct. I then verified this with an older version of blender using the "old" edge split node. The number of vertices and edges is exactly the same, and the connectivity is also the same.
  • There is a difference in the output of the two versions of the node: the order of the edges is different! This is something that happens due to a slightly different algorithm being used to sort edges around a vertex.
  • Later on in the tree, some nodes produce different outputs because they implicitly rely on the ordering of the edges (like the mesh to curve node for the orientation of the curves). This causes the final output to be different between the different versions of blender.

In any case, there is no real way to "fix" this difference without reverting to the older version of the edge split node, which is not desirable. After all the newer version is much faster, and can take advantage of multiple cpu cores. Since geometry nodes tries to avoid the use of indices as much as possible anyway, I don't think that this report is a compelling enough reason to revert.

I don't think this is actually a "bug". Let me explain why: I did some research into what exactly was causing the difference in output. Since the relevant commit was found to be a refactor of the edge split node, the problem should occur there. This is what I found: - The output of the "new" edge split node is correct: I manually checked that each edge that should get split gets split, and that the number of vertices and edges is correct. I then verified this with an older version of blender using the "old" edge split node. The number of vertices and edges is exactly the same, and the connectivity is also the same. - There is a difference in the output of the two versions of the node: the order of the edges is different! This is something that happens due to a slightly different algorithm being used to sort edges around a vertex. - Later on in the tree, some nodes produce different outputs because they implicitly rely on the ordering of the edges (like the mesh to curve node for the orientation of the curves). This causes the final output to be different between the different versions of blender. In any case, there is no real way to "fix" this difference without reverting to the older version of the edge split node, which is not desirable. After all the newer version is much faster, and can take advantage of multiple cpu cores. Since geometry nodes tries to avoid the use of indices as much as possible anyway, I don't think that this report is a compelling enough reason to revert.
Blender Bot added
Status
Archived
and removed
Status
Confirmed
labels 2023-05-05 22:03:53 +02:00
Sign in to join this conversation.
No Label
Interest
Alembic
Interest
Animation & Rigging
Interest
Asset System
Interest
Audio
Interest
Automated Testing
Interest
Blender Asset Bundle
Interest
BlendFile
Interest
Code Documentation
Interest
Collada
Interest
Compatibility
Interest
Compositing
Interest
Core
Interest
Cycles
Interest
Dependency Graph
Interest
Development Management
Interest
EEVEE
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
Viewport & EEVEE
Interest
Virtual Reality
Interest
Vulkan
Interest
Wayland
Interest
Workbench
Interest: X11
Legacy
Asset Browser Project
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
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
Module
Viewport & EEVEE
Platform
FreeBSD
Platform
Linux
Platform
macOS
Platform
Windows
Severity
High
Severity
Low
Severity
Normal
Severity
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
4 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#107433
No description provided.