Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[bug] MR2023.1.0 on Windows: LSCM & ABF unwrap do not compute / generate the right textures #1965

Open
JeS24 opened this issue Apr 7, 2023 · 1 comment
Labels
bug for actual bugs (unsure? use type:question)

Comments

@JeS24
Copy link

JeS24 commented Apr 7, 2023

Describe the bug

Setting the unwrap method to LSCM or ABF in the Texturing node in MR 2023.1.0 on Windows, does not produce the correct texture images, even though the logs / status say that the node was "successful". For large datasets, the node generally does not compute. But on the toy monstree_mini6 dataset, the node computes for about 15 minutes (on my system), before producing a completely dark texture image, as shown below. Basic unwrap works fine. I have attached the verbose tracebacks & corresponding texture images below.

Steps to reproduce

  1. Load monstree_mini6 into the default photogrammetry pipeline.
  2. Set the unwrap method to LSCM / ABF in the Texturing node.
  3. Compute.

Expected behavior

Unwrap procedure should yield the correct texture files in a reasonable amount of time.

Logs

(In these cases, the textureSide parameter was set to 4096 instead of the default 8192.

Desktop environment

  • OS: Windows 11 22H2
  • Python version: 3.10.5 (System ver)
  • Qt/PySide version: Not installed (System ver).
  • Meshroom version:
    • Binary version: 2023.1.0

Additional context

  • The same behaviour is present in MR 2021.1.0 on Linux and Windows.
  • For LSCM & ABF, the following is mentioned for all triangles: No visibility for triangle * in texture atlas * after scoring!!
  • In the case of LSCM, the following error is present:
[12:28:28.819603][info] Start mesh atlasing (using Geogram LSCM).
OpenNL error in nlEigenSolve_ARPACK(): Could not factorize matrix
OpenNL error in nlEigenSolve_ARPACK(): Could not factorize matrix

I am aware, there are other issues on this repo mentioning this, but unlike in those cases, there is no segfault here.

  • That error does not occur (or is not logged) in the case of ABF.
@JeS24 JeS24 added the bug for actual bugs (unsure? use type:question) label Apr 7, 2023
@guizy67
Copy link

guizy67 commented May 19, 2023

LSCM and ABF are also not working as expected for me on Linux Ubuntu 22.04.
Taking the default pipeline, and after a successful Texturing using Basic, changing the unwrap method to LSCM does procedure a textured mesh.
The result is a mesh that is rotated by 180 deg compared to the input mesh and no texture.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug for actual bugs (unsure? use type:question)
Projects
None yet
Development

No branches or pull requests

2 participants