Open full view…

flight path does not cover the area of interest when camera at 45 deg angle

Sat, 28 Apr 2018 11:24:15 GMT

I am trying to capture images with the objective to create a 3D model of an area. With just top-down images and a circle around image sequence, this doesn't work well enough. What is needed I think is to capture the area [Flight path does not cover the area of interest when camera at 45 deg angle. When I set the angle to 45 degrees and test it out in 3D view, you can see that a large area is missed in the attached screenshot. (flightpathdoesnotcovertheareaofinterestwhencameraat45degangle.jpg) So the flight path and capture moments needs to be adjusted according to the pitch angle (and probably the yaw angle if this isn't 0 deg). In general, what would be good is to create an extra option whereby the indicated area is scanned topdown plus also at a pitch angle of 45 degrees (or another angle, which doesn't show the sky in the image, to avoid color differences) with the topdown images and to maximise the available image pixels). Not only 45 deg forward view, but also 45 deg backwards, sideways (each side, 90 degree yaw angle, although this could also be achieved by flyi ng a second zig-zag pattern perpendicular to the main one), and maybe even at 45 and -45 deg yaw angle. The 3D creation software (https://www.capturingreality.com/) needs all these angles, and has no limit to the number of images.

Tue, 01 May 2018 10:27:36 GMT

Hi pjotrvs, Thanks for your question. You are right! At the moment when you do a Top Down flight and you change the gimbal angle from 90 degrees to say 45 degrees, part the area might not be covered. We are working on improving this aspect of the Top Down plan. What can work for you in the meanwhile is to create a slightly larger area (by how much depends on the altitude and the gimbal angle) that contains the area of interest and plan a Top Down flight for it. This is just a workaround for now, in case you are interested to try it out. More generally to the points you raised. It seems that you are mapping a rather complex (from the 3D structure point of view) area. This usually requires flights in several directions with gimbal angles that in the range of 30 to 60 (e.g., as you mentioned 0- 45 degrees). We are currently working on flight plans that will be suited for exactly this task. These might be cross-hatch type of missions, as you referred to, where one flies in two orthogonal directions with a gimbal angle in the needed range, or it may be a set of orbits that cover the region with appropriate overlap. We are working with Bentley Systems' Reality Modelling team on this topic, as we are partners and they are experts in this sort of model generation. I hope we will have a product that we can release in the near future going along these lines. In the meanwhile, in case you did not find the option, it is possible to manually set the direction of the scan lines in the Top Down flight, in case you want to generate a cross-hatch flight as you referred to before. From our experience, it generates good results. Best, David

Tue, 01 May 2018 13:34:09 GMT

Thanks, David, I will do that. I am comparing several software products and services for the automatic 3D model creation: Bentley, ReCap, RealityCapture, Bentley, Pix4D, and DroneDeploy.Com. So far, Bently's ContextCapture has the worst results, and RealityCapture the best and most editable, and DroneDeploy.Com the fastest & easiest. I know you have a partnership with Bentley, and they have been in this space for a long time, but to me it looks like their technology is getting a bit outdated compared to these newcomers. So please pay some attention to them as well. As soon as the wind drops, I will do new runs and processing tests. Thanks again, keep up the good work!

Thu, 03 May 2018 09:12:51 GMT

Thanks for this information! From our experience, it is hard to choose an overall winner in all aspects. For better quality it seems to be a good idea to consult with the vendor what image acquisition methods are best to use with their photogrammetry software as the results might vary dramatically based on the acquisition method. It might also depend on whether you are interested in only a point cloud, or a mesh of the scene. In any case, thanks a lot for sharing your findings!