For the DSM methods to work, an approximate version of the shape needs to be established to form the basis of the point cloud. This approximate surface in most cases is done by PhotoModeler automatically in the background.
The Paired Photos method uses the approximate surface in conjunction with a depth range setting. So the closer this approximate shape is to the real world shape, the smaller the depth range search can be (and a smaller depth range search lowers the possibility of an incorrect match producing noise data).
The MVS method uses existing points (often SmartPoints) to roughly define the shape at the outset but since it uses multiple photos to help confirm a match, it does not need a depth range setting.
You can define this approximate surface shape in two ways:
• Create and select one or more surfaces that roughly match the object (e.g. for a pile of dirt this approximate shape might be a rough pyramid). These surfaces can be path surfaces, NURBS surfaces, surface draw connected surfaces, or even cylinders.
or, a method requiring less manual work:
• Run DSM without selecting any surfaces. In this case an 'implicit' approximate surface will be created using the existing points in the project (and if using SmartMatch, the number of existing SmartPoints in the project may be quite high and a very close approximate surface will be generated implicitly).
• A different implicit surface is created for each photo pair being processed using the Paired Photos approach. Only those 3D points that are marked and referenced on both photos of the pair are used and if there are DSM Trims on the photos, only the 3D points that 2D project to be inside the enabled trims will be used. A best fit plane is created from these points and then they are triangulated and extended and trimmed as needed to match the desired extents.
• For MVS, an implicit surface is created from SmartPoints.
With certain point sets, method 2 may produce a best fit plane that is not suitable and is not perpendicular to the camera plane. In these cases you may need to use method 1, manually creating a suitable approximate shape.
An approximate surface (i.e. a Surface defined in PhotoModeler using any of the surfacing tools) can be used to define the extents of a DSM or it can be used as a ‘base’ surface to set the depth range.
Note 1:When an approximate surface(s) is selected before opening the DSM Options Dialog, the ‘selected surface(s) extents’ option becomes available in the ‘Extents from’ setting. If no surface is selected, ‘DSM Trims’ or 'DSM Masks' (if any are defined) or ‘whole photo boundary’ will only be available. The Extents setting is only available for Paired Photos, but if there are DSM Trims or DSM Masks defined, MVS will use them to limit the extents of the Cloud/Mesh.
Note 2:When an approximate surface(s) is selected before opening the DSM Options Dialog, the depth range values will be based on the selected surface. If no surface is selected, the depth range will be based on a best plane fit implicitly defined by PhotoModeler based on the 3D points in the project.