I am trying to export a point cloud from RC but the .xyz or the xyzrgb files RC can export are very small (about 700kb) in my project that has 481 images and when I try to open them in another application I see a very sparse point cloud. its not like the dense point cloud I see inside the RC. is there any other way to export the dense point cloud I see inside the RC viewport?
Hi George
you exporting the sparse point cloud (registration )
go to RECONSTRUCTION ribbon = MESH and change the export format from OBJ or PLY to XYZ PointCloud and so can export the model as a PointCloud
that worked perfectly, thank you! 
Hello,
a little question about the point cloud export, is it a true point cloud export or just an export after removing the face index of a previously ‘poisson’ meshed point cloud as some remaining construction vertex suggest?
We are interested by the initial point cloud export without any meshing operation to be sure that any vertex are moved from the initial calculation.
There is a mean to to this in RC?
Thanks.
Pascal
Hi Pascal,
RC does not interpolate polynomial or other surfaces, does not over/under sample scene. So technically if you set “high” mode, you get all the detail which is possible to get given image resolution.
Vertices which come out of the process are correct from the metrological point of view.
For example, given a point-cloud from a laser scanner alone and after applying meshing, you do not get new artificial points, nor repositioned points w.r.t. input. The application can group points which are closer than a given threshold - there is “minimal sampling distance” in reconstruction settings \ geometry \ laser scanners. So for example, if you are ok with 1mm precision, then the system will sub select points such that distance between them is at least 1mm.
If you put there 0, then you get all input points. Of course excluding points which are not consistent across the scene (e.g. parking car which was visible from one scanning position, but it is missing in another 2-3 scanning positions)
If you include photos into a project - the app will preserve laser point-cloud + add/densify points from photogrammetry.
So triangle vertices are always coming from real measurements. This feature is used in industrial applications of RC.
Hello,
thank you for these interesting explanations, for the moment we have not mixed 3D scanner and photo data (but we will test this feature very quickly), I was talking about photogrammetry calculation only.
A mesh exported from CR has a ‘box’ around it with very large triangles that make necessary a tedious cleaning of the mesh (I am sorry but I do not know what to call it other than by talking of ‘interpolation’…).
Is there a way, for example, to fix the maximal triangle edge length or to tell the software not to calculate such artefacts? (in photoscan for example it is possible to disable the … interpolation).
Thanks Pascal
Hi Pascal
to get a idea about combination of laser data and images
https://dl.dropboxusercontent.com/u/150 … r_foto.mp4
for the meshing. for now it creates watertight meshes. we will improve the meshing this year to get more control about its results.
For the triangle removal, there are for sure some good tool that can select and remove the large triangles…
Hello,
thank you for the link and your answer, so we have yet to find how to mix laser and photo data … in case of problem we will ask you …
We are also impatient to discover the future developments …
Best regards.
Pascal
Hi Pascal
If need can offer you a 2 day training especially for the laser + images capture workflow.
If interested send me a email to milos.lukac@capturingreality.com
A PTS output would be much more usuful in the longrun as I assume it will export as a single cloud. A pts file has a header containing the number of points, all software that can read PTS imports them much much quicker than any form or XYZ as it does not need to read the file to know how big it is.
PTS format
12345 number of points
x y z i r g b
x y z i r g b
x y z i r g b
…