Hi everybody
In the FAQ, it is said:
"Note that the two algorithms can also be combined by first running the Delaunay meshing to robustly filter outliers from the sparse or dense point cloud and then, in the second step, performing Poisson surface reconstruction to obtain a smooth surface."
I'm very interessed by this feature and I tried to use it like this:
colmap feature_extractor \
--database_path colmap/database.db \
--image_path images \
--SiftExtraction.num_threads 8
colmap exhaustive_matcher \
--database_path colmap/database.db \
--SiftMatching.num_threads 8
mkdir colmap/sparse
colmap mapper \
--database_path colmap/database.db \
--image_path images \
--output_path colmap/sparse \
--Mapper.num_threads 8
mkdir colmap/dense
colmap image_undistorter \
--image_path images \
--input_path colmap/sparse/0 \
--output_path colmap/dense \
--output_type COLMAP \
--max_image_size 2000
sudo colmap patch_match_stereo \
--workspace_path colmap/dense \
--workspace_format COLMAP \
--PatchMatchStereo.cache_size 25 \
--PatchMatchStereo.geom_consistency true \
--PatchMatchStereo.num_iterations 2
colmap stereo_fusion \
--workspace_path colmap/dense \
--workspace_format COLMAP \
--input_type geometric \
--output_path colmap/dense/fused.ply \
--StereoFusion.cache_size 512
colmap delaunay_mesher \
--input_path colmap/dense \
--output_path delaunay-output.ply
colmap poisson_mesher \
--input_path colmap/dense/fused.ply \
--output_path poisson-output.ply \
But the mesh I get from Poisson mesher is the same if I use the Delaunay one before or not.
Could you please help me?
Maxime