Hi, I’m trying to create the best possible models of a number of high rise towers and I’m looking for some pointers on where I am going wrong. Outputs required are the pointcloud, and some good quality ortho images, I don’t need mm accuracy but the results so far are just not good enough, particularly the orthos.
The example below of 2 towers is around 2500 images, I’ve added around 30 CPs and I just can’t tighten up the model to produce neat lines and edges.
What do I need? More images, or better processing? Any basic tips will be gratefully received
Taking images from at least two different distances from the object is recommended to improve alignment. This can help reduce alignment issues.
In your case, the probable culprit is the shiny surfaces such as windows and loggias. If possible, masking them out would be a great thing to do.
You can try adjusting the position accuracy values found in the alignment settings. By default, they are set to 10 meters. On the second alignment, decrease that value (e.g. from 10 to 6). Repeat this process until you reach a value in centimeters (e.g. 0.05). This can be done in 5-6 alignment. Don’t worry, after the initial alignment, all other alignments should be quicker.
Any tips on suitable distances? I am finding the small drones far too susceptible to wind and GPS wobbles to trust flying too close or using the apps with manual facade flights (eg, dronelink) It’s also not always possible to vary the distances due to the proximity of other buildings, but in other cases I am able to fly an oblique grid over the top so that may help?
I will try adjusting the position accuracy, any other tips will be gratefully received
