How it is working

PhotoModeler Scanner 2014.0 - What's New

The SmartMatch and Dense Surface Modeling capabilities of PhotoModeler Scanner 2014 and PhotoModeler Motion 2014 received some signifcant improvements.

The four signficant improvements are:

  1. New Multi-View Stereo (MVS)
  2. New SmartMatch Auto-calibration
  3. Revamp of SmartMatch Orientation
  4. Improved Meshing

New Multi-View Stereo (MVS)

Previous: Dense Surface Modeling (DSM) capability of PhotoModeler Scanner is a pair-of-photos-based image correlation system. While it produces excellent results, when there are a large number of photographs choosing which photo pairs to run and to merge can be tedious.

New-2014: MVS is a new technique that studies all photographs in the project to generate the dense surface mesh. In addition MVS rarely requires trimming. In these ways MVS is far more automated and for large projects can produce better surface results. The pair-based DSM capability is still available as an option on the DSM dialog.

Here is a screen shot of a 360 degree scan of a tree trunk done with the new MVS that would be difficult with the older style DSM (and in-fact with any scanning device!):

Tree Trunk 360 deg scan

In addition you may want to watch the Introduction to MVS Tutorial Video.

New SmartMatch Auto-calibration

Previous: To get the best results in some SmartMatch projects the camera being used should be calibrated. PhotoModeler often gets good results with just the EXIF data (camera parameters written by the digital camera into image header), but better results and less noise result if the camera is calibrated. In addition if you have already calibrated the camera, sometimes it is at a focus distance different than the project (e.g. 2m focus calibration but camera is on a UAV with focus at infinity or 10's or 100's of meters). In this case it is useful to fine tune the camera calibration to produce a better final results.

New-2014: The new SmartMatch Auto-calibration method will, by default, perform field calibration as the orientation stage runs (if the camera is not calibrated or was setup by EXIF headers). For cameras already calibrated, the auto-calibration can be turned on as an option.

Revamp of SmartMatch Orientation

Previous: Projects with larger numbers of photos and those with weaker textures and weaker overlap would sometimes not complete orientation in previous versions, or the orientation would be incorrect.

New-2014: A revamp of the Orientation sub-system in SmartMatch means that many more projects will orient correctly now.

Improved Meshing

Previous: The new triangulation system introduced in 2013 produced more complete (fewer holes) meshes but often extended the surface far beyond where the point data support was. In addition the meshes were often smoothed even though the underlying point data would support more triangles and more detail.

New-2014: The improved meshing means that trimming of the final triangulated surface is more accurate and useful. Smoothing is also reduced, creating much higher surface detail if the photo data supports it.

An additional improvement is the compute of RGB values for mesh points is threaded (speeding it up)

Below is a sample of a wood carving and the improvement in detail in the new meshing. Top is the previous version's trianguled mesh, bottom is the new one at high detail setting (you can even see the wood carving tool marks!).

Detail Improvement in Carving Surface



With these changes more SmartMatch / DSM projects will complete automatically from start to finish with great results and with little to no user interaction. A recent blog post show a couple of examples done with the new version.

Video of Project Example

Or watch on YouTube.

© 2014 Eos Systems Inc.