This problem result is a Cloud/Mesh with just a few points or no points at all. The result is not the dense grid of points expected. This can result when:
There were not enough sample points in the region.
Solution:
• MVS: Decrease Min Visible images, decrease Matching threshold, and /or decrease Surface density, though note more noise may result.
• Paired Photos: Decrease the sample size in the DSM Basic settings.
There was little or no overlap in the photos or pairs chosen.
Solution:
• MVS: Works best with many photos with good overlap. Retake photos ensuring suitable overlap and coverage. Increase maximum angle settings.
• Paired Photos: Review the orientation and overlap of the photos to see if they image the same part of the surface. Also review the DSM Trims or the selected surface(s) to see if visible in the chosen pairs of photos. You can use the 3D Surface Photo Visibility as one way of doing this for surfaces. Defined trims are always visible on photos. Ensure that the chosen pairs of photos have the trim(s) or selected surface(s) appearing on both photographs.
Also note that if the extents type is set to 'DSM Trims' and you are executing a multi-photo (3 or more photo) dsm extraction, no 3D data will be generated for a pair of photos that have no enabled ‘trim outside boundary’ type trims marked on either photo (see DSM Trims for a description of trim properties, and DSM Approximate Surfaces for a description of how to use surfaces to define extents).
The depth range setting is too small and does not match the reality of the distance of the real surface from the approximating surface.
Solution:
• This applies to Paired Photos DSM method only. Under the Basic DSM settings there is Depth Range - PhotoModeler sets a default range but it is a guess based on either the 3D points in the project or the selected surface. If the range is too small no good matches will be found within the search distance or bad matches will be found. The stronger the project (lower residuals, non-moving object/cameras, and strong random texture) the larger this range can be and you will still get good results. In a strong project the range can be far larger than it needs to be and good results can still be obtained. Shrinking the range is needed for projects that are not as strong.
The texture of the surface is man-made and consistency repeating (so it is hard to tell which is the correct match).
Solution:
• MVS: Take many photos with good overlap. Possibly reduce the Window Size setting to minimize mismatched features.
• Paired Photos: A repeating texture may give too much choice for matching. Shrinking the depth range may assist. Alternatively decreasing the texture type value to 1 (in DSM Advanced Settings) may increase noise in this case but may give more useful data as well.
The surface is flat or has no random texture.
Solution:
• MVS and Paired Photos: Perhaps the surface can be photographed with different lighting or with a higher resolution camera to improve texture capture. If it is truly a textureless surface, perhaps a random pattern can be projected or painted on before photography.
The approximating surface is too far away from the true surface.
Solution:
• MVS: Ensure existing points (ideally SmartPoints) reasonably represent the surface. If the SmartMatch solution is weak and noisy, understand the source of the SmartMatch problem – if too sparse, there may not be enough texture, or photos may be taken too far from object, or photos may not be high enough in resolution.
• Paired Photos: For the search range to be reasonable the approximating base surface (i.e. defined implicitly by the 3D points in the project, or by a selected surface) needs to be close to the real surface. Some projects will work with the base surface not being exact but it depends on the randomness and how visible the texture is. Make sure the approximating surface is closer to the real surface.
The object moved between photographs.
Solution:
• MVS and Paired Photos: The object needs to be very steady between photos being processed. It should be completely stationary or move less than 2 pixels. If you are modeling an object you know will move then multiple synchronized cameras must be used.
The correlation window is too small compared to natural feature size.
Solution:
• MVS: The default Window size is 7. Check that a typical window of this size contains suitable features to detect and match features.
• Paired Photos: The default search window is 11 by 11 pixels. Look in your images and outline an 11 by 11 region. Does it seem small compared to the elements of the texture? A quick test also is to rerun one of the photo pairs increasing the window size (under DSM Advanced Settings) to see if more data is created.
The texture type is set to too high a value.
Solution:
• This applies to Paired Photos only: The higher the texture type value (under DSM Advanced Settings) the less data will be produced. You use a higher value to trade off noise vs density in surfaces that have textures that are not that globally random (i.e. image patches are good matches for multiple patches in the other image).