last week I finished my lectures in the Master Open Techne about Free Software and 3D data (acquisition and processing). As last year I could spend many times to research new solutions and test some applications. Some months ago, thanks to the friend +Romain Janvier , I was introduce to the use of MicMac, a suite for three-dimensional documentation of reality developed by the Institut national de l’information géographique et forestière (IGN). As Python Photogrammetry Toolbox, MicMac can produce point cloud from set of photos. There are two different ways to acquire images:
- the ground geometry mode (useful for zenithal pictures as a drone data-set or for wall's prospect) = take pictures perpendicularly to the surface (ground or wall) with 60 % of overlapping (between images and lines of image)
- the image geometry mode (useful for any kind of object that has more than one surface) = take a "cross" of images starting from the central one and then up, down, left and right; then take other images moving to the second position (frontal to another surface) and again a "cross" of image; go on in this way for all the surfaces of the object that you need to reconstruct.
The data acquisition is a little bit more complicate than PPT, both in the way to shoot and in the camera settings (keep the same level of zoom, no auto-focus, no stabilization, no flash, ...), but the final point cloud is denser. PPT is more user-friendly (thank to the python scripts and the GUI) but slower in processing data (mostly in the Camera Pose Estimation step of Bundler).
One of the advantages of MicMac is the fast developing that is improving the software and simplifying its usage. I'm waiting for the GUI ;)
Unfortunately Bundler and CMVS/PMVS have not new release since years.