Regression: Broken vertex weight with mirror modifier #95541

New Issue

lygamac · 2022-02-05T22:56:31+01:00

lygamac commented

2022-02-05 22:56:31 +01:00

System Information
Operating system: Arch Linux 5.15.18-1-lts
Graphics card: AMD ATI Radeon RX 570

Blender Version
Broken: 3.0.1
Worked: 2.9.*

Behavior changed in 1a7757b0bc ...
... but that commit got reverted in e6cd4761e7
The change in behavior is still present in master though (as is the code -- got reintroduced with a0acb9bd0c)

Short description of error

When mirror modifier is on, the vertex at the mirror plane is only affected by the side that is not mirrored. The mirrored side does not affect vertexes at the mirror plane.

Only happens with new created armatures and new weights, files in older blender projects performs as expected (I'm not sure if it's because I missed some features update, but happens with default configuration).

More details can be found here.

Exact steps for others to reproduce the error

A minimal blend file is attached, which followed:

Create a new file
Enter the edit mode and make a loop cut to the default cube, and delete one half of if
Exit edit mode, create a mirror modifier
Create an armature, and apply automatic weights (this does not matter, it happens even with manually painted weights and rigify)
Try moving the two bones at the side

untitled.blend

**System Information** Operating system: Arch Linux 5.15.18-1-lts Graphics card: AMD ATI Radeon RX 570 **Blender Version** Broken: 3.0.1 Worked: 2.9.* Behavior changed in 1a7757b0bc ... ... but that commit got reverted in e6cd4761e7 The change in behavior is still present in master though (as is the code -- got reintroduced with a0acb9bd0c) **Short description of error** When mirror modifier is on, the vertex at the mirror plane is only affected by the side that is not mirrored. The mirrored side does not affect vertexes at the mirror plane. Only happens with new created armatures and new weights, files in older blender projects performs as expected (I'm not sure if it's because I missed some features update, but happens with default configuration). More details can be found [here](https://blender.stackexchange.com/posts/252025/edit). **Exact steps for others to reproduce the error** A minimal blend file is attached, which followed: 1. Create a new file 1. Enter the edit mode and make a loop cut to the default cube, and delete one half of if 1. Exit edit mode, create a mirror modifier 1. Create an armature, and apply automatic weights (this does not matter, it happens even with manually painted weights and rigify) 1. Try moving the two bones at the side [untitled.blend](https://archive.blender.org/developer/F12849207/untitled.blend)

lygamac commented

2022-02-05 22:56:31 +01:00

Added subscriber: @lygamac

lygamac commented

2022-02-05 23:07:52 +01:00

Seems broken in 3.0.0 too, I cannot check where 2.9.* works exactly due to Arch Linux dependencies system (got error with libtbb.so.2)

Seems broken in 3.0.0 too, I cannot check where 2.9.* works exactly due to Arch Linux dependencies system (got error with `libtbb.so.2`)

Omar Emara commented

2022-02-07 10:38:48 +01:00

Added subscriber: @OmarEmaraDev

Omar Emara commented

2022-02-07 10:38:48 +01:00

Changed status from 'Needs Triage' to: 'Archived'

Omar Emara closed this issue

2022-02-07 10:38:48 +01:00

Omar Emara commented

2022-02-07 10:38:48 +01:00

Thanks for the report, but the issue reported here is a request for modified/improved behavior and not a bug in current behavior. Closing as this bug tracker is only for bugs and errors.

For user requests and feedback, please use other channels: https://wiki.blender.org/wiki/Communication/Contact#User_Feedback_and_Requests

For more information on why this isn't considered a bug, visit: https://wiki.blender.org/wiki/Reference/Not_a_bug

This is caused by merging the vertices at the center. Blender averages the weights of vertices that got merged, so their weights will be different. I guess this behavior can be changed, but this is not a bug.

Thanks for the report, but the issue reported here is a request for modified/improved behavior and not a bug in current behavior. Closing as this bug tracker is only for bugs and errors. For user requests and feedback, please use other channels: https://wiki.blender.org/wiki/Communication/Contact#User_Feedback_and_Requests For more information on why this isn't considered a bug, visit: https://wiki.blender.org/wiki/Reference/Not_a_bug This is caused by merging the vertices at the center. Blender averages the weights of vertices that got merged, so their weights will be different. I guess this behavior can be changed, but this is not a bug.

Omar Emara commented

2022-02-07 11:43:25 +01:00

It seems the difference is not due to the modifier, rather, it seems the behavior of the automatic weight operator changed between versions. To get the same behavior in 3.0+, you need to enable the X Mirror option when computing the automatic weights. Not sure if this should be considered a bug.

Omar Emara commented

2022-02-07 11:44:38 +01:00

Changed status from 'Archived' to: 'Needs Triage'

Omar Emara reopened this issue

2022-02-07 11:44:38 +01:00

Philipp Oeser commented

2022-02-07 13:57:45 +01:00

Added subscriber: @lichtwerk

Philipp Oeser commented

2022-02-07 13:57:45 +01:00

In #95541#1303016, @OmarEmaraDev wrote:
It seems the difference is not due to the modifier, rather, it seems the behavior of the automatic weight operator changed between versions. To get the same behavior in 3.0+, you need to enable the X Mirror option when computing the automatic weights. Not sure if this should be considered a bug.

With the X Mirror option enabled, I get exactly the same results as in 2.93 (where the option also existed) and that is : no influence on the mirror plane
Will check which commit changed behavior here.

> In #95541#1303016, @OmarEmaraDev wrote: > It seems the difference is not due to the modifier, rather, it seems the behavior of the automatic weight operator changed between versions. To get the same behavior in 3.0+, you need to enable the X Mirror option when computing the automatic weights. Not sure if this should be considered a bug. With the `X Mirror` option enabled, I get exactly the same results as in 2.93 (where the option also existed) and that is : no influence on the mirror plane Will check which commit changed behavior here.

lygamac commented

2022-02-07 15:35:21 +01:00

Thanks for the responce, but for

you need to enable the X Mirror option when computing the automatic weights

I'm afraid the problem is not with automatic weights, as I tried to paint the weights myself, and the result was the same: No influence on the mirror plane when mirrored

Thanks for the responce, but for > you need to enable the X Mirror option when computing the automatic weights I'm afraid the problem is not with automatic weights, as I tried to paint the weights myself, and the result was the same: No influence on the mirror plane when mirrored

lygamac commented

2022-02-07 15:47:56 +01:00

For the same example file, but manually painted weights on the original side:

For the same example file, but manually painted weights on the original side: ![image.png](https://archive.blender.org/developer/F12852719/image.png)

Philipp Oeser commented

2022-02-07 15:55:07 +01:00

Changed status from 'Needs Triage' to: 'Confirmed'

Philipp Oeser commented

2022-02-07 15:55:07 +01:00

Added subscribers: @mont29, @ideasman42

Philipp Oeser commented

2022-02-07 15:55:07 +01:00

This changed in 1a7757b0bc ...
... but that commit got reverted in e6cd4761e7

The change in behavior is still present though (as is the code -- bisecting between spits out aa1c44a113?), not sure what is going on with git.

CC @mont29
CC @ideasman42
could you check?

This changed in 1a7757b0bc ... ... but that commit got reverted in e6cd4761e7 The change in behavior is still present though (as is the code -- bisecting between spits out aa1c44a113?), not sure what is going on with git. CC @mont29 CC @ideasman42 could you check?

Omar Emara commented

2022-02-07 15:57:04 +01:00

We are now mostly considering why there is a difference between older versions and newer versions. In this example, does it work differently on older versions?

Philipp Oeser commented

2022-02-07 15:58:28 +01:00

In #95541#1303245, @OmarEmaraDev wrote:
We are now mostly considering why there is a difference between older versions and newer versions. In this example, does it work differently on older versions?

Yes, the mirror modifer behaves differently

> In #95541#1303245, @OmarEmaraDev wrote: > We are now mostly considering why there is a difference between older versions and newer versions. In this example, does it work differently on older versions? Yes, the mirror modifer behaves differently

Philipp Oeser commented

2022-04-28 10:20:51 +02:00

Raising priority since it is a recent regression, feel free to lower again if this feel unappropriate.

Philipp Oeser changed title from ~~Broken vertex weight with mirror modifier~~ to Regression: Broken vertex weight with mirror modifier

2022-04-28 10:21:03 +02:00

Henrik D. commented

2022-05-02 16:33:55 +02:00

Added subscriber: @HDMaster84

Henrik D. commented

2022-05-02 16:33:55 +02:00

In #95541#1303232, @lichtwerk wrote:
This changed in 1a7757b0bc ...
... but that commit got reverted in e6cd4761e7

The change in behavior is still present though (as is the code -- bisecting between spits out aa1c44a113?), not sure what is going on with git.

CC @mont29
CC @ideasman42
could you check?

Since the revert #91444 is now an open bug again.

> In #95541#1303232, @lichtwerk wrote: > This changed in 1a7757b0bc ... > ... but that commit got reverted in e6cd4761e7 > > The change in behavior is still present though (as is the code -- bisecting between spits out aa1c44a113?), not sure what is going on with git. > > CC @mont29 > CC @ideasman42 > could you check? Since the revert #91444 is now an open bug again.

Bastien Montagne commented

2022-05-02 16:51:44 +02:00

1a7757b0bc was reverted by e6cd4761e7, and then re-introduced by a0acb9bd0c but for new modifiers only. This is why older existing files behave as expected, while newly created ones have the issue.

@HDMaster84 so no, #91444 is not open bug again, it's just a bug that cannot be fixed automatically for existing files, users need to manually do it if they want to

1a7757b0bc was reverted by e6cd4761e7, and then re-introduced by a0acb9bd0c but for new modifiers only. This is why older existing files behave as expected, while newly created ones have the issue. @HDMaster84 so no, #91444 is not open bug again, it's just a bug that cannot be fixed automatically for existing files, users need to manually do it if they want to

Bastien Montagne self-assigned this 2022-05-02 17:33:56 +02:00

Bastien Montagne commented

2022-05-02 17:33:56 +02:00

Quick follow up, am on it, way the vgroups are merged at the center needs to be changed in new case....

blender-admin commented

2022-05-03 10:30:50 +02:00

This issue was referenced by 20f819d708

This issue was referenced by 20f819d7082f0f7582a7a9e7f7be66e2b59aae63

Bastien Montagne commented

2022-05-03 10:34:15 +02:00

Changed status from 'Confirmed' to: 'Resolved'

Bastien Montagne closed this issue

2022-05-03 10:34:15 +02:00

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: Broken vertex weight with mirror modifier #95541