Can't produce dense model from pycolmap tutorial because of: Could not fuse any points. This is likely caused by incorrect settings - filtering must be enabled for the last call to patch match stereo. #4280

Describe the bug
When using the https://pypi.org/project/pycolmap/ tutorial section (see code at the end) for "Sparse & Dense Reconstruction" it fails with the following error(s):

I20260321 20:24:57.432746 131780097111552 fusion.cc:283] Fusing image [2/2] with index 1
I20260321 20:24:57.797152 131780097111552 fusion.cc:310] in 0.364s (0 points)
W20260321 20:24:57.797279 131780097111552 fusion.cc:331] Could not fuse any points. This is likely caused by incorrect settings - filtering must be enabled for the last call to patch match stereo.
I20260321 20:24:57.797326 131780097111552 fusion.cc:337] Number of fused points: 0
I20260321 20:24:57.797328 131780097111552 timer.cc:90] Elapsed time: 0.016 [minutes]
I20260321 20:24:57.798541 131780097111552 mvs.cc:341] Writing output: "/home/nkmep/data/colmap/original/pycolmap/mvs/dense.ply"
E20260321 20:24:57.798578 131780097111552 reconstruction.cc:1032] Check failed: colmap::ExistsDir(path_val) Directory "/home/nkmep/data/colmap/original/pycolmap/mvs/dense.ply" does not exist.
Traceback (most recent call last):
File "/home/nkmep/pycolmap/reconstruction_pycolmap.py", line 24, in
pycolmap.stereo_fusion(mvs_path / "dense.ply", mvs_path)
ValueError: [reconstruction.cc:1032] Check failed: colmap::ExistsDir(path_val) Directory "/home/nkmep/data/colmap/original/pycolmap/mvs/dense.ply" does not exist.

It creates the following files/folder at pycolmap: 0; 1; cameras.bin; database.db; frames.bin; images.bin; mvs points3D.bin; rigs.bin

Additionally when using pycolmap with more parameters and other functions its always the same/close errors.

On my personal PC I used colmap GUI automatic_reconstructer with the same images (>80 of "hallway" with machinery on one side) yet on the workstation it always fails.

Expected behavior
Produce dense model(s) like fused.ply in the GUI

Environment:

• OS: Ubuntu 24.04.4 LTS
• COLMAP Version:
  pycolmap 4.0.2; pycolmap-cuda12; python 3.11.15 (in a conda env)
  NVIDIA-SMI 580.126.09; Driver Version: 580.126.09; CUDA Version: 13.0
  NVIDIA RTX PRO 6000 Blackwell Server Edition
• Capture Device: Azure Kinect (only used every fifth image from the video)

The code:

import pycolmap
import sys
import pathlib

if not pycolmap.has_cuda:
print("kein cuda da")
sys.exit(1)

image_dir = pathlib.Path(r"/home/nkmep/data/colmap/original/frames_5")
output_path = pathlib.Path(r"/home/nkmep/data/colmap/original/pycolmap")

output_path.mkdir()
mvs_path = output_path / "mvs"
database_path = output_path / "database.db"

pycolmap.extract_features(database_path, image_dir)
pycolmap.match_exhaustive(database_path)
maps = pycolmap.incremental_mapping(database_path, image_dir, output_path)
maps[0].write(output_path)

pycolmap.undistort_images(mvs_path, output_path, image_dir)
pycolmap.patch_match_stereo(mvs_path) # requires compilation with CUDA
pycolmap.stereo_fusion(mvs_path / "dense.ply", mvs_path)