Support Website Contact Support Blog

Image distance for Orthplane reconstruction

How much does the distance from subject matter when acquiring images for an orthplane project? My initial thoughts it that I should be back far enough to capture a good version of the facade with like a 24MM  (say 20-30 feet) and get up close with a 8MM. (15-20 feet).

Looking through the image acquisition guidelines there didn’t seem to be much help for the terrestrial distance acquisition, only that the overlap should be as much as possible.

Any thoughts?

Thanks.

The distance to the object will have an impact on the Ground Sampling Distance (GSD) of your project. You can calculate that with the GSD calculator tool. The GSD you select will depend on the accuracy requirements you have for the project. Also note that the GSD should as much as possible be homogeneous throughout the project to improve the matching of keypoints between images. These guidelines apply as well to aerial as to terrestrial images.

For orthoplanes I generally recommend to fly a vertical grid with the camera pointing towards the facade or slightly tilted downwards. The small tilt usually helps to reduce the amount of background when reaching the top of the facade.

Another point that could influence the distance at which you take the images is if there are objects between the camera and the facade, you would select the distance at which the obstacles are not there or where their impact is smaller. 

1 Like

Thank you. I have been testing some of the things you mentioned. I want also to get some thoughts about whether or not the workflow above would work when trying to do a terrestrial scan with cars in the way.

I have a lot of work in more urban areas and parked cars are always going to be a problem. My thought was to use 2 data sets, one up close in front of the cars next to the facade and the other further away to capture a better image of the higher levels. The drone could also be considered a 3rd data set for a higher view of upper floors.

Would this be acceptable, or am I adding complexity?

I am going to be taking a drone out tomorrow for a scan and will test out the vertical grid capture. But I will have problems around the trees. Again being able to utilize a second data set would be great. I can move cars, I can’t move trees, you know?

Thanks for he help.