Geomtetry Nodes: Fillet Curve wrong radius with one switch option being 0 #99889

New Issue

D0ppelgaenger · 2022-07-21T15:33:43+02:00

D0ppelgaenger commented

2022-07-21 15:33:43 +02:00

System Information
Operating system: Windows-10-10.0.22000-SP0 64 Bits
Graphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.59

Blender Version
Broken: version: 3.3.0 Alpha, branch: master, commit date: 2022-07-20 22:17, hash: 9f68369247
Worked: 3.3.0 Alpha from yesterday (21/07/2022)

Short description of error
Wrong radius result if one of the options of a switch node is equal to 0.

Even tough the result of the switch is true for all points, sometimes it still does not use the True value, or the other way around (if False >0 and True=0)

It must have to do with Port fillet node to the new data-block , since in yesterday's daily build it worked fine.
Hope I described it in a way that makes sense

Exact steps for others to reproduce the error
Open example file
Connect different values to Viewer Socket of the node output
See that, e.g. radius should be 2. Yet if the fillet node evaluates it, it reads 0 for some points

FilletBug.blend

**System Information** Operating system: Windows-10-10.0.22000-SP0 64 Bits Graphics card: NVIDIA GeForce RTX 3070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 516.59 **Blender Version** Broken: version: 3.3.0 Alpha, branch: master, commit date: 2022-07-20 22:17, hash: `9f68369247` Worked: 3.3.0 Alpha from yesterday (21/07/2022) **Short description of error** Wrong radius result if one of the options of a switch node is equal to 0. Even tough the result of the switch is true for all points, sometimes it still does not use the **True** value, or the other way around (if False >0 and True=0) It must have to do with [Port fillet node to the new data-block ](https://developer.blender.org/rB2551cf908732bab2865654571164d74e8bbad47e), since in yesterday's daily build it worked fine. Hope I described it in a way that makes sense ![RadiusAfterSwitch_FalseEqualZero.png](https://archive.blender.org/developer/F13306127/RadiusAfterSwitch_FalseEqualZero.png) ![RadiusAfterSwitch_FalseNotEqualZero.png](https://archive.blender.org/developer/F13306126/RadiusAfterSwitch_FalseNotEqualZero.png) **Exact steps for others to reproduce the error** Open example file Connect different values to **Viewer** Socket of the node output See that, e.g. radius should be 2. Yet if the fillet node evaluates it, it reads 0 for some points [FilletBug.blend](https://archive.blender.org/developer/F13306688/FilletBug.blend)

D0ppelgaenger commented

2022-07-21 15:33:43 +02:00

Added subscriber: @Leo-Luther

Omar Emara commented

2022-07-21 16:33:25 +02:00

Added subscriber: @OmarEmaraDev

Omar Emara commented

2022-07-21 16:33:25 +02:00

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

Omar Emara commented

2022-07-21 16:33:25 +02:00

I can reproduce the issue. The index of the control point in the pre-beveled curve is 16.
Can you try to simplify the file though to make it easier to debug this?

I can reproduce the issue. The index of the control point in the pre-beveled curve is 16. Can you try to simplify the file though to make it easier to debug this?

Omar Emara changed title from ~~Geomtetrynodes: Fillet Curve wrong radius with one switch option being 0~~ to Geomtetry Nodes: Fillet Curve wrong radius with one switch option being 0

2022-07-21 16:34:11 +02:00

D0ppelgaenger commented

2022-07-21 18:01:33 +02:00

I already removed a tone of nodes and don't know how much more I could take away.

Nodes outside the node group

You can take away the random set cyclic, and it seems that the issue only seems to appear on non-cyclic splines.
It also only happens if the splines have different points and are not uniform, therefore the resample random count is needed.
The nodes that follow the G_Fillet Curve node group can in theory be removed and only serve the purpose of better visualizing the bug

Nodes within the node group

The capture attribute can be removed and was only added to make sure that the issue is definitely not due to a mistake with evaluating the angel at the wrong position
I'd like to remove more nodes inside the group and find another switch condition for which the issue appears, but I haven't found one yet.

I know that it therefore might seem like the issue lies somewhere else (point angle node group or so), but it worked before the port to the new data block from yesterday

Also, if you view the values in the spreadsheet they are correct (example: point of the original curve with index 12 shows a value of 2, however the fillet still fillets with a radius of 0)
Also, if you change the False value to even just 0.000001 it suddenly works as expected (which indicates that it is an issue with the switch being set to 0)

Reduced File 2 comments down

I already removed a tone of nodes and don't know how much more I could take away. Nodes outside the node group - You can take away the random set cyclic, and it seems that the issue ***`only seems to appear on non-cyclic splines`***. - It also only happens if the splines have different points and are not uniform, therefore the resample random count is needed. - The nodes that follow the **G_Fillet Curve** node group can in theory be removed and only serve the purpose of better visualizing the bug Nodes within the node group - The capture attribute can be removed and was only added to make sure that the issue is definitely not due to a mistake with evaluating the angel at the wrong position - I'd like to remove more nodes inside the group and find another switch condition for which the issue appears, but I haven't found one yet. ``` I know that it therefore might seem like the issue lies somewhere else (point angle node group or so), but it worked before the port to the new data block from yesterday ``` - Also, if you view the values in the spreadsheet they are correct (example: point of the original curve with index 12 shows a value of 2, however the fillet still fillets with a radius of 0) - Also, if you change the False value to even just 0.000001 it suddenly works as expected (which indicates that it is an issue with the switch being set to 0) Reduced File 2 comments down

Iliya Katushenock commented

2022-07-21 18:04:45 +02:00

Added subscriber: @mod_moder

Iliya Katushenock commented

2022-07-21 18:04:45 +02:00

For example, all of this
Just create new project and do bug by 5-10 nodes

I still don't understand what is wrong

![image.png](https://archive.blender.org/developer/F13306433/image.png) For example, all of this Just create new project and do bug by 5-10 nodes I still don't understand what is wrong

D0ppelgaenger commented

2022-07-21 20:35:08 +02:00

Sry wasn't aware, that all of that old stuff was still there (was too zoomed in and didn't see it)
Here I removed it
FilletBug.blend

The issue is that even though the result for the radius coming from the switch node is 2 (as you can see in the spreadsheet) the actual fillet is done with a value of 0 (resulting in no fillet at all)

Sry wasn't aware, that all of that old stuff was still there (was too zoomed in and didn't see it) Here I removed it [FilletBug.blend](https://archive.blender.org/developer/F13306687/FilletBug.blend) The issue is that even though the result for the radius coming from the switch node is 2 (as you can see in the spreadsheet) the actual fillet is done with a value of 0 (resulting in no fillet at all)

Iliya Katushenock commented

2022-07-21 20:50:52 +02:00

2022-07-21 21-49-18.mp4
Is there an error here?

[2022-07-21 21-49-18.mp4](https://archive.blender.org/developer/F13306701/2022-07-21_21-49-18.mp4) Is there an error here?

D0ppelgaenger commented

2022-07-21 21:16:30 +02:00

Index 6/7 are endpoints, so they are supposed to be 0/False
Index 11/12 however, aren't endpoints, their switch condition is True and the switch therefore outputs 2 as value as seen in this screenshot (so far working as expected). However, the value these two points actually get filleted with is 0.

Also, as soon as you change the False result of the switch (even if just 0.00001) it suddenly fillets with the value you set for True (2 in this case)

Index 6/7 are endpoints, so they are supposed to be 0/False Index 11/12 however, aren't endpoints, their switch condition is True and the switch therefore outputs 2 as value as seen in this screenshot (so far working as expected). However, the value these two points actually get filleted with is 0. ![image.png](https://archive.blender.org/developer/F13306745/image.png) Also, as soon as you change the **False** result of the switch (even if just 0.00001) it suddenly fillets with the value you set for **True** (2 in this case)

blender-admin commented

2022-07-25 12:20:37 +02:00

This issue was referenced by 5feb3541f4

This issue was referenced by 5feb3541f459d916202b705c2e9c62f5778e182b

Jacques Lucke commented

2022-07-25 12:22:05 +02:00

Changed status from 'Confirmed' to: 'Resolved'

Jacques Lucke closed this issue

2022-07-25 12:22:05 +02:00

Jacques Lucke self-assigned this 2022-07-25 12:22:05 +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

Geomtetry Nodes: Fillet Curve wrong radius with one switch option being 0 #99889