Skip to content

Latest commit

 

History

History
310 lines (172 loc) · 9.63 KB

MVEparams.md

File metadata and controls

310 lines (172 loc) · 9.63 KB

Preperation for node implementation

[Convert2MVE]

MVE

https://github.com/simonfuhrmann/mve

https://www.gcc.tu-darmstadt.de/media/gcc/code/mve-20160517-win64.zip

https://github.com/simonfuhrmann/mve/wiki/Build-Instructions-for-Windows

https://github.com/simonfuhrmann/mve/wiki/MVE-Users-Guide

(also https://github.com/SamirAroudj/TSR)

https://peterfalkingham.com/2016/10/14/photogrammetry-testing-5-mve-multiview-environment/

https://informatik.hs-bremerhaven.de/dh2018/doku/02-Erstellung_MVE-Pipeline.html


Convert2MVE --> dmrecon --> scene2pset (PLY) [--> fssrecon --> meshclean]


MVE Bundle to Pointset

This application reads a bundle file and outputs a PLY file with a colored point cloud.

bundle2pset.exe [ OPTIONS ] INPUT_BUNDLE OUTPUT_PLY

Available options:

-s, --spheres=ARG Generates a sphere for every point (radius ARG) [0.0]

MVE Depth Map Reconstruction

dmrecon.exe [ OPTIONS ] SCENEDIR

Available options:

-n, --neighbors=ARG amount of neighbor views (global view selection)

-m, --master-view=ARG reconstructs given master view ID only

-l, --list-view=ARG reconstructs given view IDs (given as string "0-10")

-s, --scale=ARG reconstruction on given scale, 0 is original

--max-pixels=ARG Limit master image size [1500000]

-f, --filter-width=ARG patch size for NCC based comparison [5]

--nocolorscale turn off color scale

-i, --image=ARG specify source image embedding [undistorted]

--local-neighbors=ARG amount of neighbors for local view selection [4]

--keep-dz store dz map into view

--keep-conf store confidence map into view

-p, --writeply use this option to write the ply file

--plydest=ARG path suffix appended to scene dir to write ply files

--bounding-box=ARG Six comma separated values used as AABB [disabled]

--progress=ARG progress output style: 'silent', 'simple' or 'fancy'

--force Reconstruct and overwrite existing depthmaps

Floating Scale Surface Reconstruction

Samples the implicit function defined by the input samples and produces a surface mesh. The input samples must have normals and the "values" PLY attribute (the scale of the samples). Both confidence values and vertex colors are optional. The final surface should be cleaned (sliver triangles, isolated components, low-confidence vertices) afterwards.

fssrecon.exe [ OPTS ] IN_PLY [ IN_PLY ... ] OUT_PLY

Available options:

-s, --scale-factor=ARG Multiply sample scale with factor [1.0]

-r, --refine-octree=ARG Refines octree with N levels [0]

--min-scale=ARG Minimum scale, smaller samples are clamped

--max-scale=ARG Maximum scale, larger samples are ignored

--interpolation=ARG Interpolation: linear, scaling, lsderiv, [cubic]

MVE Makescene

This utility creates MVE scenes by importing from an external SfM software. Supported are Noah's Bundler, Photosynther and VisualSfM's compact .nvm file.

For VisualSfM, makescene expects the .nvm file as INPUT. With VisualSfM, it is not possible to keep invalid views.

For Noah's Bundler, makescene expects the bundle directory as INPUT, a file "list.txt" in INPUT and the bundle file in the "bundle" directory.

For Photosynther, makescene expects as INPUT the directory that contains the "bundle" and the "undistorted" directory. With Photosynther, it is not possible to keep invalid views or import original images.

With the "images-only" option, all images in the INPUT directory are imported without camera information. If "append-images" is specified, images are added to an existing scene.

makescene.exe [ OPTIONS ] INPUT OUT_SCENE Available options: -o, --original Import original images

-b, --bundle-id=ARG Bundle ID (Photosynther and Bundler only) [0]

-k, --keep-invalid Keeps images with invalid cameras

-i, --images-only Imports images from INPUT_DIR only

-a, --append-images Appends images to an existing scene

-m, --max-pixels=ARG Limit image size by iterative half-sizing

MVE FSSR Mesh to Pointset

This app creates a PLY point cloud from the input mesh by stripping the connectivity information. Scale values are computed for each vertex as the average distance to each neighbor (using the connectivity information). Confidence values are computed by down-weighting boundary vertices.

mesh2pset.exe [ OPTS ] IN_MESH OUT_PLY_PSET

Available options:

-s, --scale=ARG Set constant scale for all samples [off]

-a, --adaptive=ARG Average distance to neighbors scale factor [1.0]

-b, --bounding-box=ARG Six comma separated values used as AABB [off]

-c, --no-confidences Do not compute vertex confidences

-x, --no-scale-values Do not compute sample scale

-n, --no-normals Do not compute sample normals

MVE FSSR Mesh Alignment

Generates a combined mesh from Stanford or Meshlab alignment data. The combined mesh contains all triangulated range images in a global, consistent coordinate system. Stanford alignments are .conf files with global camera and a list of meshes with alignment information. Meshlab alignment are .aln files with a list of meshes and a camera to world transformation matrix.

meshalign.exe [ OPTS ] ALIGNMENT_FILE [...] OUT_MESH

MVE FSSR Mesh Cleaning

The application cleans degenerated faces resulting from MC-like algorithms. Vertices below a confidence threshold and vertices in small isolated components are deleted as well.

meshclean.exe [ OPTS ] IN_MESH OUT_MESH

Available options:

-t, --threshold=ARG Threshold on the geometry confidence [1.0]

-p, --percentile=ARG Use the nth percentile (0 - 100) as confidence threshold [disabled]

-c, --component-size=ARG Minimum number of vertices per component [1000]

-n, --no-clean Prevents cleanup of degenerated faces

--delete-scale Delete scale attribute from mesh

--delete-conf Delete confidence attribute from mesh

--delete-color Delete color attribute from mesh

MVE Mesh Conversion

Converts the mesh given by IN_MESH to the output file OUT_MESH. The format of the input and output mesh are detected by extension. Supported file formats are .off, .ply (Stanford), .npts or .bnpts (Poisson Surface Reconstruction) and .pbrt.

meshconvert.exe [ OPTS ] IN_MESH OUT_MESH

Available options:

-n, --normals Compute vertex normals

MVE SfM Prebundle

Statistics generator for prebundle files. The graph mode outputs the matching graph in Graphviz DOT format.

prebundle.exe [ MODES ] PREBUNDLE_FILE

Available options:

-g, --graph-mode=ARG Graph mode: Output matching graph file for DOT

MVE Scene to Pointset

Generates a pointset from the scene by projecting reconstructed depth values in the world coordinate system.

scene2pset [ OPTS ] SCENE_DIR MESH_OUT

Available options:

-d, --depthmap=ARG Name of depth map to use [depth-L0]

-i, --image=ARG Name of color image to use [undistorted]

-n, --with-normals Write points with normals (PLY only)

-s, --with-scale Write points with scale values (PLY only)

-c, --with-conf Write points with confidence (PLY only)

-m, --mask=ARG Name of mask/silhouette image to clip 3D points []

-v, --views=ARG View IDs to use for reconstruction [all]

-b, --bounding-box=ARG Six comma separated values used as AABB.

-f, --min-fraction=ARG Minimum fraction of valid depth values [0.0]

-p, --poisson-normals Scale normals according to confidence

-S, --scale-factor=ARG Factor for computing scale values [2.5]

-F, --fssr=ARG FSSR output, sets -nsc and -di with scale ARG

MVE Scene Upgrade

This utility upgrades an MVE view, a prebundle file, or an MVE scene to the new format. See the Github wiki for more details about the new formats. INPUT can either be a single .mve view, a single .sfm prebundle file, or a scene directory. In the latter case, all views and the prebundle.sfm are upgraded.

sceneupgrade.exe [ OPTIONS ] INPUT

Available options:

-k, --keep-original Keep original files

MVE SfM Reconstruction

Reconstruction of camera parameters for MVE scenes using Structure from Motion. Note: The prebundle and the log file are relative to the scene directory.

sfmrecon.exe [ OPTIONS ] SCENE

Available options:

-o, --original=ARG Original image embedding [original]

-e, --exif=ARG EXIF data embedding [exif]

-m, --max-pixels=ARG Limit image size by iterative half-sizing [6000000]

-u, --undistorted=ARG Undistorted image embedding [undistorted]

--prebundle=ARG Load/store pre-bundle file [prebundle.sfm]

--log-file=ARG Log some timings to file []

--no-prediction Disable matchability prediction

--normalize Normalize scene after reconstruction

--skip-sfm Compute prebundle, skip SfM reconstruction

--always-full-ba Run full bundle adjustment after every view

--video-matching=ARG Only match to ARG previous frames [0]

--fixed-intrinsics Do not optimize camera intrinsics

--shared-intrinsics Share intrinsics between all cameras

--intrinsics-from-views Use intrinsics from MVE views [use EXIF]

--track-error-thres=ARG Error threshold for new tracks [10]

--track-thres-factor=ARG Error threshold factor for tracks [25]

--use-2cam-tracks Triangulate tracks from only two cameras

--initial-pair=ARG Manually specify initial pair IDs [-1,-1]

--cascade-hashing Use cascade hashing for matching [false]