POINT CLOUDComing soon
Stream Capture turns overlapping photos into a dense, georeferenced point cloud and classifies it into ground, vegetation, buildings and noise. The depth work runs on your GPU, the cloud is written in the formats survey software reads, and you can correct classes in 3D before anything is exported.
Stream Capture is in development and not released yet. At launch it will come in the same download as its sister program Stream Terrain; today's download contains Stream Terrain.
First the photos are matched to each other and the camera positions are solved; that gives a sparse cloud of tie points. Then, for every photo, depth is estimated against its neighbours and the estimates are fused into a dense cloud. A point is kept only when several views agree on it.
A photogrammetric cloud is not a laser scan. It records the surface the cameras saw — canopy, roofs, stockpile tops — and ground under dense vegetation has to be estimated from the ground that was visible.
Dense depth is computed on the GPU. The Maximum preset adds a slanted-plane PatchMatch refinement for sharper detail; the points kept are those that at least two views agree on.
Points carry ASPRS class codes: ground, low, medium and high vegetation, building and noise, with an optional water class. Statistical outlier removal cleans the cloud, and a ground filter separates the terrain.
Select points with a rectangle, a free-form lasso or a brush, assign the target class, and undo or redo class edits. Saved view bookmarks let you return to the same spot.
The default is LAS 1.4 with point format 7 (RGB and GPS time) and the coordinate system embedded; LAS 1.2 is available for older tools. Add COPC for region-by-region reading, CSV with coordinates in the output system, or a 3D Tiles tileset for web viewers.
A density multiplier and a noise-cleaning multiplier scale the cloud up or down, and a sparse-cloud option keeps the unfiltered alignment points.
The dense stage works out of core in tiles and uses an automatic or fixed RAM budget, so a cloud larger than memory is still processed.
An open text model of cameras and points is available for research pipelines, and the LAZ, COPC and CSV files keep the output coordinate system, so GIS and point cloud tools place them correctly.
The cloud is written in the output coordinate system you confirm in the wizard.
It is in the products list of the Output step; the dense cloud is the Dense Cloud stage in the Process section.
Fast Draft gives a quick, lighter preview cloud; Balanced is the everyday choice; Maximum adds the refinement and the densest grid.
Switch the view to the dense cloud (key 1). The overlap map in the Quality Report shows weak areas, and holes and noise are visible here.
Select points in 3D and mark them with the target class.
LAZ is written with the project; the Export Center adds COPC, CSV and 3D Tiles.
Photograph the subject or site with plenty of overlap, then process the photos in photogrammetry software. In Stream Capture you add the folders, tick Point Cloud and process; the dense cloud comes out classified, in LAZ, in your coordinate system.
Yes, that is its main use. The cloud is georeferenced from the photo positions, and from ground control points when you add them.
It depends on photo count, overlap and preset: more overlap and a higher preset give a denser cloud.
They measure different things. Photogrammetry gives dense colour and a precise surface where the camera sees it, but cannot see through canopy; LiDAR can. For open ground and structures photogrammetry is usually sufficient and far cheaper.
LAZ is LAS compressed without loss, typically a small fraction of the size. Stream Capture writes LAZ by default; most current GIS and point cloud tools read it, and LAS 1.2 point format output exists for older readers.
Yes. The quality report gives the cloud's height scatter around surveyed points, and the overlap map shows how many photos see each cell.
Stream Capture is in development and not released yet. At launch it will come in the same download as its sister program Stream Terrain; today's download contains Stream Terrain.
About Stream Capture