Updated runtime parameters for 0.4

pull/11/head
Piero Toffanin 2018-10-22 15:42:55 -04:00 zatwierdzone przez GitHub
rodzic e4dc29de17
commit 519a22cddb
Nie znaleziono w bazie danych klucza dla tego podpisu
ID klucza GPG: 4AEE18F83AFDEB23
1 zmienionych plików z 64 dodań i 71 usunięć

Wyświetl plik

@ -88,20 +88,20 @@ Args::
resizes images by the largest side for opensfm. Set to resizes images by the largest side for opensfm. Set to
-1 to disable. Default: 2048 -1 to disable. Default: 2048
--start-with <string>, -s <string> --start-with <string>, -s <string>
Can be one of: dataset | opensfm | slam | cmvs | pmvs Can be one of: dataset | opensfm | slam | smvs |
| odm_meshing | odm_25dmeshing | mvs_texturing | odm_meshing | odm_25dmeshing | mvs_texturing |
odm_georeferencing | odm_dem | odm_orthophoto odm_georeferencing | odm_dem | odm_orthophoto
--end-with <string>, -e <string> --end-with <string>, -e <string>
Can be one of:dataset | opensfm | slam | cmvs | pmvs | Can be one of:dataset | opensfm | slam | smvs |
odm_meshing | odm_25dmeshing | mvs_texturing | odm_meshing | odm_25dmeshing | mvs_texturing |
odm_georeferencing | odm_dem | odm_orthophoto odm_georeferencing | odm_dem | odm_orthophoto
--rerun <string>, -r <string> --rerun <string>, -r <string>
Can be one of:dataset | opensfm | slam | cmvs | pmvs | Can be one of:dataset | opensfm | slam | smvs |
odm_meshing | odm_25dmeshing | mvs_texturing | odm_meshing | odm_25dmeshing | mvs_texturing |
odm_georeferencing | odm_dem | odm_orthophoto odm_georeferencing | odm_dem | odm_orthophoto
--rerun-all force rerun of all tasks --rerun-all force rerun of all tasks
--rerun-from <string> --rerun-from <string>
Can be one of:dataset | opensfm | slam | cmvs | pmvs | Can be one of:dataset | opensfm | slam | smvs |
odm_meshing | odm_25dmeshing | mvs_texturing | odm_meshing | odm_25dmeshing | mvs_texturing |
odm_georeferencing | odm_dem | odm_orthophoto odm_georeferencing | odm_dem | odm_orthophoto
--video <string> Path to the video file to process --video <string> Path to the video file to process
@ -132,13 +132,15 @@ Args::
0 0
--use-fixed-camera-params --use-fixed-camera-params
Turn off camera parameter optimization during bundler Turn off camera parameter optimization during bundler
--opensfm-processes <positive integer> --max-concurrency <positive integer>
The maximum number of processes to use in dense The maximum number of processes to use in various
reconstruction. Default: <num cpus> processes. Peak memory requirement is ~1GB per thread
--opensfm-depthmap-resolution <positive float> and 2 megapixel image resolution. Default: 4
Resolution of the depthmaps. Higher values take longer --depthmap-resolution <positive float>
to compute but produce denser point clouds. Default: Controls the density of the point cloud by setting the
640 resolution of the depthmap images. Higher values take
longer to compute but produce denser point clouds.
Default: 640
--opensfm-depthmap-min-consistent-views <integer: 2 <= x <= 9> --opensfm-depthmap-min-consistent-views <integer: 2 <= x <= 9>
Minimum number of views that should reconstruct a Minimum number of views that should reconstruct a
point for it to be valid. Use lower values if your point for it to be valid. Use lower values if your
@ -159,40 +161,39 @@ Args::
the reconstruction and a global adjustment every 100 the reconstruction and a global adjustment every 100
images. Speeds up reconstruction for very large images. Speeds up reconstruction for very large
datasets. datasets.
--use-25dmesh Use a 2.5D mesh to compute the orthophoto. This option --use-3dmesh Use a full 3D mesh to compute the orthophoto instead
tends to provide better results for planar surfaces. of a 2.5D mesh. This option is a bit faster and
Experimental. provides similar results in planar areas.
--use-pmvs Use pmvs to compute point cloud alternatively --skip-3dmodel Skip generation of a full 3D model. This can save time
--cmvs-maxImages <integer> if you only need 2D results such as orthophotos and
The maximum number of images per cluster. Default: 500 DEMs.
--pmvs-level <positive integer> --use-opensfm-dense Use opensfm to compute dense point cloud alternatively
The level in the image pyramid that is used for the --ignore-gsd Ignore Ground Sampling Distance (GSD). GSD caps the
computation. see maximum resolution of image outputs and resizes images
http://www.di.ens.fr/pmvs/documentation.html for more when necessary, resulting in faster processing and
pmvs documentation. Default: 1 lower memory usage. Since GSD is an estimate,
--pmvs-csize <positive integer> sometimes ignoring it can result in slightly better
Cell size controls the density of image output quality.
reconstructionsDefault: 2 --smvs-alpha <float> Regularization parameter, a higher alpha leads to
--pmvs-threshold <float: -1.0 <= x <= 1.0> smoother surfaces. Default: 1.0
A patch reconstruction is accepted as a success and --smvs-output-scale <positive integer>
kept if its associated photometric consistency measure The scale of the optimization - the finest resolution
is above this threshold. Default: 0.7 of the bicubic patches will have the size of the
--pmvs-wsize <positive integer> respective power of 2 (e.g. 2 will optimize patches
pmvs samples wsize x wsize pixel colors from each covering down to 4x4 pixels). Default: 2
image to compute photometric consistency score. For --smvs-enable-shading
example, when wsize=7, 7x7=49 pixel colors are sampled Use shading-based optimization. This model cannot
in each image. Increasing the value leads to more handle complex scenes. Try to supply linear images to
stable reconstructions, but the program becomes the reconstruction pipeline that are not tone mapped
slower. Default: 7 or altered as this can also have very negative effects
--pmvs-min-images <positive integer> on the reconstruction. If you have simple JPGs with
Each 3D point must be visible in at least minImageNum SRGB gamma correction you can remove it with the
images for being reconstructed. 3 is suggested in --smvs-gamma-srgb option. Default: False
general. Default: 3 --smvs-gamma-srgb Apply inverse SRGB gamma correction. To be used with
--pmvs-num-cores <positive integer> --smvs-enable-shading when you have simple JPGs with
The maximum number of cores to use in dense SRGB gamma correction. Default: False
reconstruction. Default: 16
--mesh-size <positive integer> --mesh-size <positive integer>
The maximum vertex count of the output mesh Default: The maximum vertex count of the output mesh. Default:
100000 100000
--mesh-octree-depth <positive integer> --mesh-octree-depth <positive integer>
Oct-tree depth used in the mesh reconstruction, Oct-tree depth used in the mesh reconstruction,
@ -201,28 +202,13 @@ Args::
--mesh-samples <float >= 1.0> --mesh-samples <float >= 1.0>
Number of points per octree node, recommended and Number of points per octree node, recommended and
default value: 1.0 default value: 1.0
--mesh-solver-divide <positive integer> --mesh-point-weight <interpolation weight>
Oct-tree depth at which the Laplacian equation is This floating point value specifies the importance
solved in the surface reconstruction step. Increasing that interpolation of the point samples is given in
this value increases computation times slightly but the formulation of the screened Poisson equation. The
helps reduce memory usage. Default: 9 results of the original (unscreened) Poisson
--mesh-neighbors <positive integer> Reconstruction can be obtained by setting this value
Number of neighbors to select when estimating the to 0.Default= 4
surface model used to compute the mesh and for
statistical outlier removal. Higher values lead to
smoother meshes but take longer to process. Applies to
2.5D mesh only. Default: 24
--mesh-resolution <positive float>
Size of the interpolated surface model used for
deriving the 2.5D mesh, expressed in pixels per meter.
Higher values work better for complex or urban
terrains. Lower values work better on flat areas.
Resolution has no effect on the number of vertices,
but high values can severely impact runtime speed and
memory usage. When set to zero, the program
automatically attempts to find a good value based on
the point cloud extent and target vertex count.
Applies to 2.5D mesh only. Default: 0
--fast-orthophoto Skips dense reconstruction and 3D model generation. It --fast-orthophoto Skips dense reconstruction and 3D model generation. It
generates an orthophoto directly from the sparse generates an orthophoto directly from the sparse
reconstruction. If you just need an orthophoto and do reconstruction. If you just need an orthophoto and do
@ -239,8 +225,16 @@ Args::
defaults to smrf. You can control the behavior of both defaults to smrf. You can control the behavior of both
smrf and pmf by tweaking the --dem-* parameters. smrf and pmf by tweaking the --dem-* parameters.
Default: none Default: none
--pc-csv Export the georeferenced point cloud in CSV format.
Default: False
--texturing-data-term <string> --texturing-data-term <string>
Data term: [area, gmi]. Default: gmi Data term: [area, gmi]. Default: gmi
--texturing-nadir-weight <integer: 0 <= x <= 32>
Affects orthophotos only. Higher values result in
sharper corners, but can affect color distribution and
blurriness. Use lower values for planar areas and
higher values for urban areas. The default value works
well for most scenarios. Default: 16
--texturing-outlier-removal-type <string> --texturing-outlier-removal-type <string>
Type of photometric outlier removal method: [none, Type of photometric outlier removal method: [none,
gauss_damping, gauss_clamping]. Default: gauss_damping, gauss_clamping]. Default:
@ -279,9 +273,9 @@ Args::
generated with progressively bigger radius using the generated with progressively bigger radius using the
inverse distance weighted (IDW) algorithm and merged inverse distance weighted (IDW) algorithm and merged
together. Remaining gaps are then merged using nearest together. Remaining gaps are then merged using nearest
neighbor interpolation. Default=4 neighbor interpolation. Default=3
--dem-resolution <float> --dem-resolution <float>
Length of raster cell edges in meters. Default: 0.1 DSM/DTM resolution in cm / pixel. Default: 5
--dem-maxangle <positive float> --dem-maxangle <positive float>
Points that are more than maxangle degrees off-nadir Points that are more than maxangle degrees off-nadir
are discarded. Default: 20 are discarded. Default: 20
@ -314,8 +308,7 @@ Args::
vegetation ComplexForest: Varied terrain that is vegetation ComplexForest: Varied terrain that is
forested Default=ComplexForest forested Default=ComplexForest
--orthophoto-resolution <float > 0.0> --orthophoto-resolution <float > 0.0>
Orthophoto ground resolution in pixels/meterDefault: Orthophoto resolution in cm / pixel. Default: 5
20.0
--orthophoto-target-srs <EPSG:XXXX> --orthophoto-target-srs <EPSG:XXXX>
Target spatial reference for orthophoto creation. Not Target spatial reference for orthophoto creation. Not
implemented yet. Default: None implemented yet. Default: None