Blender 4.0 creates many invalid materials, corrupts file, Eevee crashes #114428

New Issue

Closed

opened 2023-11-02 20:46:17 +01:00 by Josef-Nilsen · 12 comments

Josef-Nilsen commented

2023-11-02 20:46:17 +01:00

System Information
Operating system: macOS 14
Graphics card: M1 (Metal, GPU-only)

Blender Version
Broken: f6c52849b5 (would be nice if this could be copied from the splash screen)
Worked: Don't know

Short description of error
When assigning a new material to some meshes (not all), Blender creates hundreds of empty/invalid materials. Possibly one per face? This increases the file size and causes incorrect rendering (both the default/no material and the set material). It's impossible to remove these materials with the material cleanup (Material utilities -> Clean slots only gives me a Python stack trace).

Exact steps for others to reproduce the error
Open the file, there is a "Bad" and a "Good" object. A material "Feet" has been assigned to the good object. Try assigning it to the bad object and it will create many empty materials. I once managed to solve it by recalculating the outer normals, which sometimes but not always works, so I suspect it could be related to that?

This event somehow corrupts the file, so switching over to to the Material Preview-view, after this has happened, even after it has been remedied, causes Eevee to crash as it's populating the materials cache.

I'm a newbie so I could be doing something stupid, but it doesn't seem like the application should segfault at least...

Crash report attached along with the example.

**System Information** Operating system: macOS 14 Graphics card: M1 (Metal, GPU-only) **Blender Version** Broken: f6c52849b502 (would be nice if this could be copied from the splash screen) Worked: Don't know **Short description of error** When assigning a new material to some meshes (not all), Blender creates hundreds of empty/invalid materials. Possibly one per face? This increases the file size and causes incorrect rendering (both the default/no material and the set material). It's impossible to remove these materials with the material cleanup (Material utilities -> Clean slots only gives me a Python stack trace). **Exact steps for others to reproduce the error** Open the file, there is a "Bad" and a "Good" object. A material "Feet" has been assigned to the good object. Try assigning it to the bad object and it will create many empty materials. I once managed to solve it by recalculating the outer normals, which sometimes but not always works, so I suspect it could be related to that? This event somehow corrupts the file, so switching over to to the Material Preview-view, after this has happened, even after it has been remedied, causes Eevee to crash as it's populating the materials cache. I'm a newbie so I could be doing something stupid, but it doesn't seem like the application should segfault at least... Crash report attached along with the example.

crashreport.txt

136 KiB

normals.blend

959 KiB

Josef-Nilsen added the

labels 2023-11-02 20:46:18 +01:00

Iliya Katushenock added the

EEVEE & Viewport

label 2023-11-03 08:48:23 +01:00

Philipp Oeser commented

2023-11-03 10:13:50 +01:00

Member

We have corrupted material indices here

These can easily be corrected:

open a Python Colsole
select the Bad object
run C.active_object.data.validate_material_indices()

Now the question is: how can this corruption be reproduced?
How has the mesh been created? Was this imported (which importer)? Was this modelled with blenders tools (which)?
Please try to find ways to reproduce (otherwise we cannot continue with this report)

We have corrupted material indices here ![image](/attachments/70c42f16-19e6-4488-8f84-d4dc9e6316c0) These can easily be corrected: - open a Python Colsole - select the `Bad` object - run `C.active_object.data.validate_material_indices()` Now the question is: how can this corruption be reproduced? How has the mesh been created? Was this imported (which importer)? Was this modelled with blenders tools (which)? Please try to find ways to reproduce (otherwise we cannot continue with this report)

56 KiB

Philipp Oeser added

Needs Information from User

and removed

labels 2023-11-03 10:13:59 +01:00

Pratik Borhade commented

2023-11-03 10:17:09 +01:00

Member

(would be nice if this could be copied from the splash screen)

Help > Report a Bug 🙂

I can confirm that extra materials slots are generated but unable to identify the reason.

> (would be nice if this could be copied from the splash screen) `Help > Report a Bug` 🙂 I can confirm that extra materials slots are generated but unable to identify the reason.

Pratik Borhade commented

2023-11-03 10:19:05 +01:00

Member

We have corrupted material indices here

oh, nice findings. I did not consider to check the spreadsheet 😅

> We have corrupted material indices here oh, nice findings. I did not consider to check the spreadsheet 😅

Josef-Nilsen commented

2023-11-03 13:51:54 +01:00

Author

I created the geometries in Blender from scratch, with a combination of adjustments in edit mode (manual + bevel), and two boolean difference modifiers. I will try to reproduce...

I created the geometries in Blender from scratch, with a combination of adjustments in edit mode (manual + bevel), and two boolean difference modifiers. I will try to reproduce...

Josef-Nilsen commented

2023-11-03 16:16:23 +01:00

Author

We have corrupted material indices here

These can easily be corrected:

open a Python Colsole

select the Bad object

run C.active_object.data.validate_material_indices()

I have tried for about 30 minutes to reproduce the issue, but it seems non-trivial. I could spend more time, but I'm not sure that the corruption you mention specifically is the cause, although I don't doubt that there is some form of corruption.

Start by running validate_material_indices, the output returns True, and ensure that it worked in the spreadsheet. So far so good. Both the 3D printing toolbox and the spreadsheet now indicates that the geometry is clean, as far as I can understand. Go back to the Layout view and assign the Feet material, and the problem persists.

I tried to rule out the material itself being the cause, by creating a new material and assigning that, but the issue persists nonetheless.

I suppose it could also be argued that if I did something stupid in edit mode, the fallback could still be more graceful than a Blender segfault, i.e. even if there isn't a clear solution to the problem, one might consider mitigating its consequences with some form of error handling perhaps. This seems Eevee specific however, as Cycles does not segfault, but the Eevee segfault brings all of Blender down along with it.

> We have corrupted material indices here > > These can easily be corrected: > - open a Python Colsole > - select the `Bad` object > - run `C.active_object.data.validate_material_indices()` I have tried for about 30 minutes to reproduce the issue, but it seems non-trivial. I could spend more time, but I'm not sure that the corruption you mention specifically is the cause, although I don't doubt that there is some form of corruption. Start by running `validate_material_indices`, the output returns `True`, and ensure that it worked in the spreadsheet. So far so good. Both the 3D printing toolbox and the spreadsheet now indicates that the geometry is clean, as far as I can understand. Go back to the Layout view and assign the `Feet` material, and the problem persists. I tried to rule out the material itself being the cause, by creating a new material and assigning that, but the issue persists nonetheless. I suppose it could also be argued that if I did something stupid in edit mode, the fallback could still be more graceful than a Blender segfault, i.e. even if there isn't a clear solution to the problem, one might consider mitigating its consequences with some form of error handling perhaps. This seems Eevee specific however, as Cycles does not segfault, but the Eevee segfault brings all of Blender down along with it.

Philipp Oeser added

and removed

Needs Information from User

labels 2023-11-03 16:25:44 +01:00

Philipp Oeser commented

2023-11-06 10:14:18 +01:00

Member

OK, can confirm that doing C.active_object.data.validate_material_indices() cleans up the indices, but assigning the material creates additional material slots.

It would still be good to know how we got to the corrupted state, but cleaning up the indices should really give us a clean plate.
Will put on my desk to investigate more later...

OK, can confirm that doing `C.active_object.data.validate_material_indices()` cleans up the indices, but assigning the material creates additional material slots. It would still be good to know how we got to the corrupted state, but cleaning up the indices should really give us a clean plate. Will put on my desk to investigate more later...

Philipp Oeser self-assigned this 2023-11-06 10:14:25 +01:00

Philipp Oeser added

and removed

labels 2023-11-06 10:14:38 +01:00

Josef-Nilsen commented

2023-11-07 07:00:56 +01:00

Author

I will let you know if it happens again, now that I'm aware I'll be able to detect it earlier and more easily trace back to the operation causing it. Just curious: after assigning the material, causing the extra material slots to be added, can you use Eevee in the viewport or does Blender crash?

I will let you know if it happens again, now that I'm aware I'll be able to detect it earlier and more easily trace back to the operation causing it. Just curious: after assigning the material, causing the extra material slots to be added, can you use Eevee in the viewport or does Blender crash?

Philipp Oeser commented

2023-11-08 15:37:33 +01:00

Member

Eevee doesnt crash for me

but cleaning up the indices should really give us a clean plate.

agree

Eevee doesnt crash for me > but cleaning up the indices should really give us a clean plate. agree

Philipp Oeser commented

2023-11-08 15:50:28 +01:00

Member

Btw. as a workaround you can use https://docs.blender.org/manual/en/4.1/scene_layout/object/editing/cleanup.html#remove-unused-material-slots to get rid of the unwanted material slots

Btw. as a workaround you can use https://docs.blender.org/manual/en/4.1/scene_layout/object/editing/cleanup.html#remove-unused-material-slots to get rid of the unwanted material slots

Josef-Nilsen commented

2023-11-08 15:59:24 +01:00

Author

That didn't work for me either once the slots are there, like I wrote in my initial post:

Material utilities -> Clean slots only gives me a Python stack trace

But one thing that does work for me, is to create a new material for "Bad" rather than assigning a material. Then you can assign another material...

That didn't work for me either once the slots are there, like I wrote in my initial post: > Material utilities -> Clean slots only gives me a Python stack trace But one thing that does work for me, is to create a new material for "Bad" rather than assigning a material. Then you can assign another material...

Philipp Oeser commented

2023-11-08 16:03:42 +01:00

Member

Btw. as a workaround you can use https://docs.blender.org/manual/en/4.1/scene_layout/object/editing/cleanup.html#remove-unused-material-slots to get rid of the unwanted material slots

OR: simply type C.active_object.active_material_index = 0 before assigning the material
(when opening the file, C.active_object.active_material_index is 19976 ... )

> Btw. as a workaround you can use https://docs.blender.org/manual/en/4.1/scene_layout/object/editing/cleanup.html#remove-unused-material-slots to get rid of the unwanted material slots OR: simply type `C.active_object.active_material_index = 0` before assigning the material (when opening the file, C.active_object.active_material_index is 19976 ... )

Philipp Oeser commented

2023-11-08 16:17:43 +01:00

Member

It is actually very easy to break:

From the default startup cube:

type C.active_object.active_material_index = 100
assign a material
100 material slots are created (the first 99 empty), material is assigned to the 100th slot

I find this quite questionable behavior of setting the active_material_index (this should probably be clamped to the number of available slots?)

It is actually very easy to break: From the default startup cube: - type `C.active_object.active_material_index = 100` - assign a material - 100 material slots are created (the first 99 empty), material is assigned to the 100th slot I find this quite questionable behavior of setting the `active_material_index` (this should probably be clamped to the number of available slots?)

Thomas Barlow referenced this issue

2023-12-30 21:28:00 +01:00

Hundreds of empty materials on model #116659

Pratik Borhade referenced this issue

2024-01-01 06:50:57 +01:00

Hundreds of empty materials on model #116659

Philipp Oeser referenced a pull request that will close this issue

2024-04-09 16:26:04 +02:00

Fix #114428: clamp setting object active_material_index #120434

Philipp Oeser referenced this issue from a commit

2024-04-12 17:35:34 +02:00

Fix #114428: clamp setting object active_material_index

Philipp Oeser closed this issue

2024-04-12 17:35:35 +02:00

Blender Bot added

and removed

labels 2024-04-12 17:35:38 +02:00

Sign in to join this conversation.

No Label

Animation & Rigging

Asset Browser Project Overview

Automated Testing

Blender Asset Bundle

Dependency Graph

Development Management

EEVEE & Viewport

Images & Movies

Motion Tracking

Nodes & Physics

Pipeline, Assets & IO

Platforms, Builds & Tests

Render & Cycles

Render Pipeline

Sculpt, Paint & Texture

Video Sequencer

Virtual Reality

Blender 2.8 Project

Milestone 1: Basic, Local Asset Browser

Good First Issue

Animation & Rigging

Development Management

EEVEE & Viewport

Nodes & Physics

Pipeline, Assets & IO

Platforms, Builds & Tests

Render & Cycles

Sculpt, Paint & Texture

Needs Info from Developers

Needs Information from User

No Milestone

No project

No Assignees

3 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#114428

No description provided.