Hi Dmitry, I noticed some distortion in the houses, the small harbour (the one with the blue boat) limits are also irregular. The shades that you are referring are also noticeable but not my main concern. The main problem is occasional distortion in random parts of the orthophoto. I cannot see your ortho perfectly (the file is too small) to see if you get the same problems as I did. For example, it seems that the house with the gray roof seems to be distorted in your photo, in the right side, which does not seem to occur in the SenseFly final result. This kind of distortions are my main problem with Photoscan now. It's strange because, most of the times, the limit of the 3D object in the DEM it's very accurate but the correspondant orthophoto limits are not.
Cheers
Hi Monserrat, are you sure about SenseFly service using Orima? Their service is this one
http://pix4d.com/I think that the software approach is very similar to the Photoscan one but I can be wrong, of course :-). The IMU unit from the SenseFly CAM UAV is not as precise as a real IMU unit from classic aerial photography. Mainly is used for the UAV navigation, stabilization and slightly for coarse photo alignment. Even the camera in the UAV is a consumer one. See their site for details. I doubt that it can be used with the workflow you described, common in classical photogrammetry. The reason I bought Photoscan is that I can get great results with a lot less work from images from these UAVS and from Kite aerial photography for instance. Even using classical aerial photography (1/15000 to 1/8000 scales), I've perfected a workflow, using Photoscan in the end, with photos digitized with a Photogrammetric Scanner (or even with photos from Digital Aerial Photography systems like an Intergraph DMC). The results are a lot better, more accurate and, mainly, much much less time consuming that I ever reached with ERDAS. For my investigation purposes, the results are perfectly within the precision that I could expect from this kind of photography. Of course, I always use a very accurate assemblage of ground control points obtained in the field, with precise geodesy data and a very precise Trimble GPS. This is the only way you can get satisfactory data from photogrammetry.
Cheers