[bug] Depth Map node failure in Meshroom 2025.1.0 (Monstree picture set)

[rodpuckey]

Describe the bug When running the Monstree picture set in Meshroom 2025.1.0, the process proceeds as expected until the Depth Map node. At this stage, the progress bar turns red and the computation stops. The same picture set completes successfully in Meshroom 2023, indicating a regression in the new version.

To Reproduce Steps to reproduce the behavior:

1. Download and install Meshroom 2025.1.0.
2. Load the Monstree picture set.
3. Run the default computation pipeline.
4. Observe that the process halts at the Depth Map node with a red progress bar.

Expected behavior The computation should complete successfully, as it does in Meshroom 2023, without halting at the Depth Map node.

Screenshots N/A

Log N/A (no log provided)

Desktop (please complete the following and other pertinent information):

• OS: [user did not specify]
• Python version: [unspecified]
• Qt/PySide version: [unspecified]
• Meshroom version: 2025.1.0 (release version)
• Binary version: 2025.1.0
• Commit reference: [unspecified]

Additional context

• The Monstree picture set runs successfully in Meshroom 2023.
• This issue may be a regression in the new version.

[cbentejac]

Hi, could you please provide the logs from the DepthMap node?

I just ran the photogrammetry pipeline with that exact dataset and the DepthMap node ran smoothly:

[GMHFerreira]

[11:53:49.214336][warning] Could not determine number of CUDA cards in this system [11:53:49.214336][warning] No CUDA-Enabled GPU. CUDA build version: 12.5 [11:53:49.214336][error] cudaGetDeviceCount failed: CUDA driver version is insufficient for CUDA runtime version [11:53:49.214336][info] Can't find CUDA-Enabled GPU. [11:53:49.215336][error] This program needs a CUDA-Enabled GPU (with at least compute capability 2.0).

[natowi]

@GMHFerreira What GPU do you have and are your drivers up to date? (version?)