Regression: Strange crease when using proportional editing with the "connected only" option turned on (edge selected) #89968

New Issue

Zach Hixson · 2021-07-19T19:30:21+02:00

Zach Hixson commented

2021-07-19 19:30:21 +02:00

System Information
Operating system: Windows-10-10.0.19041-SP0 64 Bits
Graphics card: GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.72

Blender Version
Broken: version: 2.93.0, branch: master, commit date: 2021-06-02 11:21, hash: 84da05a8b8

Pretty sure this is caused by 21b9231d7f

Short description of error
When using proportional editing, there is a strange pinching/creasing that is going on where it should be smooth. It seems to only happen when "Connected Only," is turned on

Exact steps for others to reproduce the error

Open new file
Add plane
Add a whole bunch of edge loops
Enable proportional editing + "Connected Only," option
Grab one edge and move it up || down

Test File:
#89968.blend

**System Information** Operating system: Windows-10-10.0.19041-SP0 64 Bits Graphics card: GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 461.72 **Blender Version** Broken: version: 2.93.0, branch: master, commit date: 2021-06-02 11:21, hash: `84da05a8b8` Pretty sure this is caused by 21b9231d7f **Short description of error** When using proportional editing, there is a strange pinching/creasing that is going on where it should be smooth. It seems to only happen when "Connected Only," is turned on **Exact steps for others to reproduce the error** 1. Open new file 2. Add plane 3. Add a whole bunch of edge loops 4. Enable proportional editing + "Connected Only," option 5. Grab one edge and move it up || down ![ex.png](https://archive.blender.org/developer/F10228314/ex.png) Test File: [#89968.blend](https://archive.blender.org/developer/F10228725/T89968.blend)

Zach Hixson commented

2021-07-19 19:30:21 +02:00

Added subscriber: @ZachHixson

blender-admin commented

2021-07-19 19:30:21 +02:00

#102031 was marked as duplicate of this issue

Pratik Borhade commented

2021-07-20 06:11:07 +02:00

Added subscriber: @PratikPB2123

Pratik Borhade commented

2021-07-20 06:11:07 +02:00

Can confirm. Not sure if this is known issue or reported before. Will check before confirming.

Richard Antalik commented

2021-07-20 06:12:44 +02:00

Added subscriber: @iss

Richard Antalik commented

2021-07-20 06:12:44 +02:00

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

Richard Antalik commented

2021-07-20 06:12:44 +02:00

Can you provide sample file that demonstrates this issue well?

Pratik Borhade commented

2021-07-20 06:25:34 +02:00

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

Pratik Borhade commented

2021-07-20 06:25:34 +02:00

Hi @iss , added the file. Don't see the similar issue reported before.

@iss , can you also reproduce?

Hi @iss , added the file. Don't see the similar issue reported before. @iss , can you also reproduce?

Richard Antalik commented

2021-07-20 06:40:09 +02:00

In #89968#1193474, @PratikPB2123 wrote:
@iss , can you also reproduce?

I can reproduce, but only in 2.93 not in 3.0 so this seems to be fixed already, can you check that?

> In #89968#1193474, @PratikPB2123 wrote: > @iss , can you also reproduce? I can reproduce, but only in 2.93 not in 3.0 so this seems to be fixed already, can you check that?

Pratik Borhade commented

2021-07-20 07:15:42 +02:00

can you check that?

Created the file with 3.0.0 itself so yes I can reproduce (118803893e)

Also checked with debug build (ff01070b5c)

>can you check that? Created the file with 3.0.0 itself so yes I can reproduce (`118803893e`) Also checked with debug build (`ff01070b5c`)

Philipp Oeser commented

2021-07-20 14:56:21 +02:00

Added subscribers: @brecht, @lichtwerk

Philipp Oeser commented

2021-07-20 14:56:21 +02:00

Pretty sure this is caused by 21b9231d7f.

It works well (even in this example) if a single vertex is selected.
Can confirm it gets bit wonky if a region is selected though.

From the D10068: Transform: approximate geodesic distances for proportional edit connected mode:

The algorithm can be further improved for the case where a region is selected
rather than a single vertex. The boundary of the region is not smooth since it
zigzags along the topology, and that leads to non-smooth distances as well.
Less so for geodesic distances but still visible. Some type of blurring of the
boundary before propagation would help, but I consider that outside the scope
of this patch.

@brecht: Known Issue?

Pretty sure this is caused by 21b9231d7f. It works well (even in this example) if a single vertex is selected. Can confirm it gets bit wonky if a region is selected though. From the [D10068: Transform: approximate geodesic distances for proportional edit connected mode](https://archive.blender.org/developer/D10068): ``` The algorithm can be further improved for the case where a region is selected rather than a single vertex. The boundary of the region is not smooth since it zigzags along the topology, and that leads to non-smooth distances as well. Less so for geodesic distances but still visible. Some type of blurring of the boundary before propagation would help, but I consider that outside the scope of this patch. ``` @brecht: `Known Issue`?

Philipp Oeser changed title from ~~Strange crease when using proportional editing with the "connected only," option turned on~~ to Strange crease when using proportional editing with the "connected only," option turned on (region selected)

2021-07-20 14:57:52 +02:00

Germano Cavalcante commented

2021-07-20 16:17:08 +02:00

Added subscriber: @mano-wii

Germano Cavalcante commented

2021-07-20 16:17:08 +02:00

This indeed seems to be a limitation in the algorithm that estimates the geodesic distance of the vertices. (seen in geodesic_distance_propagate_across_triangle).
If I understand the code correctly, when the sum of the two geodesic distances of the triangle's reference vertices is less than the distance between these vertices, a fallback distance is returned, (which in this case coincidentally is the real distance).
But when the sum of the distances is greater than the distance between the vertices, the computed distance is equivalent to the one shown in the image below, which is smaller than the fallback, and so the propagation is done differently.

Perhaps it would be possible to improve this by indicating a direction vector for each vertices.
Or compute a different value when the geodetic distance is smaller than the one between the vertices?
Anyway, it doesn't seem like a trivial change.

This indeed seems to be a limitation in the algorithm that estimates the geodesic distance of the vertices. (seen in `geodesic_distance_propagate_across_triangle`). If I understand the code correctly, when the sum of the two geodesic distances of the triangle's reference vertices is less than the distance between these vertices, a fallback distance is returned, (which in this case coincidentally is the real distance). But when the sum of the distances is greater than the distance between the vertices, the computed distance is equivalent to the one shown in the image below, which is smaller than the fallback, and so the propagation is done differently. ![image.png](https://archive.blender.org/developer/F10229973/image.png) Perhaps it would be possible to improve this by indicating a direction vector for each vertices. Or compute a different value when the geodetic distance is smaller than the one between the vertices? Anyway, it doesn't seem like a trivial change.

Brecht Van Lommel commented

2021-07-20 18:25:47 +02:00

The code includes propagation across edges also, not just across triangles.

There may be some more complex cases where it's harder, but at least in this simple case it should be able to find accurate distances that way.

It's also a regression compared to earlier Blender versions, which was not known.

The code includes propagation across edges also, not just across triangles. There may be some more complex cases where it's harder, but at least in this simple case it should be able to find accurate distances that way. It's also a regression compared to earlier Blender versions, which was not known.

Philipp Oeser changed title from ~~Strange crease when using proportional editing with the "connected only," option turned on (region selected)~~ to Regression: Strange crease when using proportional editing with the "connected only" option turned on (edge selected)

2022-10-24 10:03:21 +02:00

Philipp Oeser commented

2022-10-24 10:03:52 +02:00

Added subscriber: @Twixuss

Philipp Oeser removed the

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Regression: Strange crease when using proportional editing with the "connected only" option turned on (edge selected) #89968