Page MenuHome
Create Task
Maniphest T79322

Minkowski F2 in the Voronoi node yields very different results when set to 4D, compared to 3D.
Closed, ArchivedPublicBUG
Actions

Description

System Information
Operating system: Windows 10
Graphics card: GeForce 980

Blender Version
Broken: 2.83+

I don't think this has ever worked. It's been an issue since the 4D option for procedural textures got introduced.

Short description of error
Switching Voronoi texture node with distance metric Minkowski from 3D to 4D makes it behave completely different. Works fine with all the other distance metrics. In Cycles as well as Eevee.

Exact steps for others to reproduce the error
Add a material to the cube. Use The Voronoi Texture set it to Minkowski, F2, exponent: in the screenshot it's 0.3 the smaller the more apparent the differences.
The position output illustrates it best, but it's happening for the other outputs as well.
Going through different values of w sometimes makes the little stars appear, but far fewer and smaller than with 3D.

VoronoiMinkowski.blend627 KBDownload

Event Timeline

Frederik Steinmetz (SpiderguyF) created this task.Jul 27 2020, 3:56 PM
Sybren A. Stüvel (sybren) changed the task status from Needs Triage to Needs Information from User.Jul 28 2020, 2:28 PM
Sybren A. Stüvel (sybren) added a subscriber: Sybren A. Stüvel (sybren).

Please try older versions of Blender as well, to see if this was a recently introduced thing, or whether Blender has worked like this for a long time already. This can considerably help in classifying this issue and finding the root cause.

Campbell Barton (campbellbarton) added a subscriber: Campbell Barton (campbellbarton).EditedAug 11 2020, 2:45 PM

This report is missing:

  • Last version that worked.
  • A blend file.
Frederik Steinmetz (SpiderguyF) updated the task description. (Show Details)Aug 11 2020, 2:51 PM
Ankit Meel (ankitm) changed the task status from Needs Information from User to Needs Triage.Sep 8 2020, 12:14 AM
Ankit Meel (ankitm) added a subscriber: Ankit Meel (ankitm).

The report is correct (the difference is visible at lower values of exponent ), but I don't know if that's expected or if it's a limitation.

Philipp Oeser (lichtwerk) changed the task status from Needs Triage to Confirmed.Oct 14 2020, 7:38 PM
Philipp Oeser (lichtwerk) added a project: Nodes.
Philipp Oeser (lichtwerk) added a subscriber: Philipp Oeser (lichtwerk).

Can at least confirm it (dont know how to classify either without checking code).

Bastien Montagne (mont29) changed the subtype of this task from "Report" to "Bug".Oct 15 2020, 12:56 PM
Bastien Montagne (mont29) edited projects, added Render & Cycles; removed Nodes.
Brecht Van Lommel (brecht) added subscribers: Omar Emara (OmarSquircleArt), Brecht Van Lommel (brecht).Nov 5 2020, 4:52 PM

@Omar Emara (OmarSquircleArt), maybe you want to look at this?

Omar Emara (OmarSquircleArt) added a comment.EditedNov 5 2020, 5:39 PM

@Brecht Van Lommel (brecht) This is expected as far as I know. As the exponent goes to zero, the unit n-sphere volume becomes smaller, and as the number of dimensions increase, the rate of decreasing of the volume increase. So if you want to get the same visual result as you change the number of dimensions, changing the exponent is necessary. That also applies when going from 1D to 2D and from 2D to 3D.

Brecht Van Lommel (brecht) closed this task as Archived.Nov 5 2020, 5:48 PM
Brecht Van Lommel (brecht) claimed this task.

Ok, that's fine then, would have been nice if the results matched but no bug here.