Page MenuHome

visual artifacts in 2.90 on a mesh imported from 2.83 with subdivision
Closed, ResolvedPublic

Description

System Information
Operating system: Windows-10-10.0.18362-SP0 64 Bits
Graphics card: GeForce GTX 750 Ti/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 446.14

Blender Version
Broken: version: 2.90.0 Alpha, branch: master, commit date: 2020-06-05 22:39, hash: rBb74cc23dc478
Worked: 2.83.0

Short description of error
Random lines appear on a mesh with subdivision.


The mesh was imported from 2.83 to 2.90. I could not reproduce with mesh created in 2.90.
Exact steps for others to reproduce the error

  • Open file.

  • Easiest way to reproduce is to select some vertices and hit shift+e (edge crease) and slowly increase value by moving the mouse.
  • They will also sometimes appear when cycling through subdiv levels or cycling between edit and object modes.

Event Timeline

Germano Cavalcante (mano-wii) changed the task status from Needs Triage to Needs Information from User.Jun 9 2020, 3:32 PM

I cannot reproduce the problem.
Has it been fixed in the last build?
https://builder.blender.org/download/

@Germano Cavalcante (mano-wii)
In 9f7d84b656fb-windows64 problem is still there.

Germano Cavalcante (mano-wii) changed the task status from Needs Information from User to Needs Information from Developers.Jun 9 2020, 3:56 PM
Dalai Felinto (dfelinto) changed the task status from Needs Information from Developers to Confirmed.Jun 11 2020, 5:55 PM

I can confirm on 04390941050b
Operating system: Linux-5.4.0-33-generic-x86_64-with-debian-bullseye-sid 64 Bits
Graphics card: GeForce GTX 1080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 440.64

Just noting that I have seen random lines elsewhere (namely T77709):

Can confirm something pretty weird going on with that mesh [looks corrupted to me]

Going to editmode gives me random vertices that dont seem to be in the selection buffer?
Reverting the file and doing this again from a different perspective sometimes makes them disappear, sometimes appear elsewhere

mesh.validate() and
mesh.validate_material_indices() both return False, so mesh should be intact?

Jeroen Bakker (jbakker) closed this task as Resolved.Jun 29 2020, 8:23 AM

I would guess that it has been fixed by one of the two draw manager threading patches that landed last week. Will close the ticket. Thanks for reporting!