Regression: Script using bmesh.ops.wireframe and bmesh.ops.bevel does not work the same in 3.1 and in 3.01. #96308

New Issue

Christopher S Nichols · 2022-03-10T19:29:33+01:00

2022-03-10 19:29:33 +01:00

System Information
Operating system: platform: 'Windows-10-10.0.19042-SP0'
Graphics card: Radeon RX 570 Series

Blender Version
Broken: version: 3.1.0, branch: master, commit date: 2022-03-08 18:16, hash: c77597cd0e, type: release
Worked: version: 3.0.1, branch: master, commit date: 2022-01-25 17:19, hash: dc2d180181, type: release

Short description of error
In 3.1 it acts like a simple bevel not properly adding the offset of the wireframe added to the mesh. In 3.01 I get a defined edge that is offset and beveled. In both, I keep the original geometry.

Exact steps for others to reproduce the error

Paste the provided script in Text editor
Execute

import bpy
import bmesh
from bpy import context as C

obj = C.active_object
me = obj.data
bm = bmesh.new()
bm.from_mesh(me)

bmesh.ops.wireframe(bm, faces=bm.faces, thickness=0.02, offset=.1, 
                    use_replace=False, use_boundary=True, use_even_offset=True, use_relative_offset=True, material_offset=-1)

bmesh.ops.bevel(bm, geom=bm.edges, offset=.05, segments=6, profile=.5, affect='EDGES', clamp_overlap=True, material=-1)

bm.to_mesh(me)
me.update()
bm.free()

**System Information** Operating system: platform: 'Windows-10-10.0.19042-SP0' Graphics card: Radeon RX 570 Series **Blender Version** Broken: version: 3.1.0, branch: master, commit date: 2022-03-08 18:16, hash: c77597cd0e15, type: release Worked: version: 3.0.1, branch: master, commit date: 2022-01-25 17:19, hash: dc2d18018171, type: release **Short description of error** In 3.1 it acts like a simple bevel not properly adding the offset of the wireframe added to the mesh. In 3.01 I get a defined edge that is offset and beveled. In both, I keep the original geometry. **Exact steps for others to reproduce the error** - Paste the provided script in Text editor - Execute ``` import bpy import bmesh from bpy import context as C obj = C.active_object me = obj.data bm = bmesh.new() bm.from_mesh(me) bmesh.ops.wireframe(bm, faces=bm.faces, thickness=0.02, offset=.1, use_replace=False, use_boundary=True, use_even_offset=True, use_relative_offset=True, material_offset=-1) bmesh.ops.bevel(bm, geom=bm.edges, offset=.05, segments=6, profile=.5, affect='EDGES', clamp_overlap=True, material=-1) bm.to_mesh(me) me.update() bm.free() ```

Christopher S Nichols commented

2022-03-10 19:29:33 +01:00

Added subscriber: @Armageddon29

blender-admin commented

2022-03-10 19:29:33 +01:00

#96693 was marked as duplicate of this issue

blender-admin commented

2022-03-10 19:29:33 +01:00

#96522 was marked as duplicate of this issue

blender-admin commented

2022-03-10 19:29:33 +01:00

#96305 was marked as duplicate of this issue

Pratik Borhade commented

2022-03-11 06:02:07 +01:00

Added subscriber: @PratikPB2123

Pratik Borhade commented

2022-03-11 06:02:07 +01:00

Changed status from 'Needs Triage' to: 'Needs User Info'

Pratik Borhade commented

2022-03-11 06:02:07 +01:00

Hi, thanks for the report. Issue appears to be related to #96305 (Applying Wireframe modifier does not apply correctly)
Can you confirm?

Hi, thanks for the report. Issue appears to be related to #96305 (Applying Wireframe modifier does not apply correctly) Can you confirm?

Christopher S Nichols commented

2022-03-11 06:22:31 +01:00

Actually, I checked adding the wireframe and the bevel modifiers and they work just fine in 3.1. It appears to mess up only in the bmesh script and only in 3.1. I also noticed that if I save the file in 3.1 and then open it up in 3.01 the script doesn't work then either. If I open a new file in 3.01 and copy the script it will work. I know you updated python, but I don't see how that could affect it. I am puzzled because it doesn't make sense to me why a file saved in 3.1 and opened in 3.01 would treat the code differently than when it was written the same in 3.01. I will leave it to someone more capable than I to ferret out the cause. If you need any more info, let me know.

Sincerely

Chris Nichols

Actually, I checked adding the wireframe and the bevel modifiers and they work just fine in 3.1. It appears to mess up only in the bmesh script and only in 3.1. I also noticed that if I save the file in 3.1 and then open it up in 3.01 the script doesn't work then either. If I open a new file in 3.01 and copy the script it will work. I know you updated python, but I don't see how that could affect it. I am puzzled because it doesn't make sense to me why a file saved in 3.1 and opened in 3.01 would treat the code differently than when it was written the same in 3.01. I will leave it to someone more capable than I to ferret out the cause. If you need any more info, let me know. Sincerely Chris Nichols

Pratik Borhade commented

2022-03-11 07:21:45 +01:00

Changed status from 'Needs User Info' to: 'Needs Developer To Reproduce'

Pratik Borhade commented

2022-03-11 07:21:45 +01:00

Added subscriber: @HooglyBoogly

Pratik Borhade commented

2022-03-11 07:21:45 +01:00

Actually, I checked adding the wireframe and the bevel modifiers and they work just fine in 3.1.

Hi, this did not work for me when I apply both modifiers (works until they are in modifiers stack)

I also noticed that if I save the file in 3.1 and then open it up in 3.01 the script doesn't work then either.

This is not clear to me either. Maybe python 3.10 upgrade is responsible? (just a guess)

Speaking about the original report, I'm pretty sure its a duplicate of #96305 (Applying Wireframe modifier does not apply correctly)
But I'd like to confirm this from @HooglyBoogly

> Actually, I checked adding the wireframe and the bevel modifiers and they work just fine in 3.1. Hi, this did not work for me when I apply both modifiers (works until they are in modifiers stack) > I also noticed that if I save the file in 3.1 and then open it up in 3.01 the script doesn't work then either. This is not clear to me either. Maybe python 3.10 upgrade is responsible? (just a guess) Speaking about the original report, I'm pretty sure its a duplicate of #96305 (Applying Wireframe modifier does not apply correctly) But I'd like to confirm this from @HooglyBoogly

Christopher S Nichols commented

2022-03-11 17:48:48 +01:00

To whom it may concern, I did the modifiers (wireframe, bevel), and once applied (which I did not do before sorry) it didn't work. Since I try to work non-destructively, it did not occur to me to apply them.

Hans Goudey commented

2022-03-11 18:55:10 +01:00

Added subscriber: @ideasman42

Hans Goudey commented

2022-03-11 18:55:10 +01:00

This is related to #96305, but since this involves the Python API and the other doesn't, I think we should keep both open.

The issue is that the BMesh created by BM_mesh_bm_from_me doesn't always have valid normals. I think that's a good thing, because when creating a BMesh, the normals might not be necessary.
It often breaks though, because BMesh has no way to track whether the normals are dirty or not, and existing code creating a BMesh often assumed that normals were in a valid state.

I can think of a few solutions:

Always calculate normals when creating a BMesh.

I don't think this is a good option.

Change BMeshFromMeshParams.calc_face_normal to ``BMeshFromMeshParams.calc_normals` and also calculate vertex normals when this is true.

A good option I think.

Add a calc_vertex_normal field to BMeshFromMeshParams and related API functions. Set this to true wherever calc_face_normal is already true.

This is slightly more complicated than option 2, but has the benefit that we can only calculate normals for one layer if necessary (which becomes more important when face corner normals are handled in a similar way in the future).

Add dirty normal tracking to BMesh and do all the normal calculation automatically.

Maybe a "better" solution to have the design more similar to Mesh, but this isn't something I'm that interested in spending time on now.

@ideasman42 I think I'd go with option 3. With your go-ahead, I'd create a patch for this.

This is related to #96305, but since this involves the Python API and the other doesn't, I think we should keep both open. The issue is that the BMesh created by `BM_mesh_bm_from_me` doesn't always have valid normals. I think that's a good thing, because when creating a BMesh, the normals might not be necessary. It often breaks though, because BMesh has no way to track whether the normals are dirty or not, and existing code creating a BMesh often assumed that normals were in a valid state. I can think of a few solutions: 1. Always calculate normals when creating a BMesh. - I don't think this is a good option. 2. Change `BMeshFromMeshParams.calc_face_normal` to ``BMeshFromMeshParams.calc_normals` and also calculate vertex normals when this is true. - A good option I think. 3. Add a `calc_vertex_normal` field to `BMeshFromMeshParams` and related API functions. Set this to true wherever `calc_face_normal` is already true. - This is slightly more complicated than option 2, but has the benefit that we can only calculate normals for one layer if necessary (which becomes more important when face corner normals are handled in a similar way in the future). 4. Add dirty normal tracking to BMesh and do all the normal calculation automatically. - Maybe a "better" solution to have the design more similar to Mesh, but this isn't something I'm that interested in spending time on now. @ideasman42 I think I'd go with option 3. With your go-ahead, I'd create a patch for this.

Hans Goudey commented

2022-03-11 18:57:10 +01:00

Changed status from 'Needs Developer To Reproduce' to: 'Confirmed'

Hans Goudey commented

2022-03-11 18:57:10 +01:00

Actually, I think I'll merge the other report here, because these reproduction steps are simpler and get at the problem a bit better. Fixing the issue would fix both issues anyway.

Hans Goudey commented

2022-03-11 18:57:24 +01:00

Added subscriber: @Burhursta

Hans Goudey self-assigned this 2022-03-11 18:57:31 +01:00

Pratik Borhade commented

2022-03-16 14:34:31 +01:00

Added subscriber: @Rafamizi

Hans Goudey commented

2022-03-22 00:06:17 +01:00

Added subscriber: @xidingart

blender-admin commented

2022-03-22 15:33:50 +01:00

This issue was referenced by fb2cb0324a

This issue was referenced by fb2cb0324a391041db3fdc6c66ce992adf9598aa

blender-admin commented

2022-03-22 15:33:50 +01:00

This issue was referenced by 64cd927519

This issue was referenced by 64cd927519748bbd50b856c25f7e4eac17d8c780

Hans Goudey commented

2022-03-22 15:34:41 +01:00

Changed status from 'Confirmed' to: 'Resolved'

Hans Goudey closed this issue

2022-03-22 15:34:41 +01:00

Philipp Oeser changed title from ~~Script using bmesh.ops.wireframe and bmesh.ops.bevel does not work the same in 3.1 and in 3.01.~~ to Regression: Script using bmesh.ops.wireframe and bmesh.ops.bevel does not work the same in 3.1 and in 3.01.

2022-03-22 16:07:23 +01:00

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

Regression: Script using bmesh.ops.wireframe and bmesh.ops.bevel does not work the same in 3.1 and in 3.01. #96308