I made up a video of some testing I did to figure out what flight/processing parameters gave me the best results for my drone photogrammetry.
I was surprised that nadir images didn't give better results for an orthographic image. I'm thinking the better point cloud and therefor DEM that was created from oblique images was the reason for this.
I was also very surprised at how much processing settings affected final results.
?ÿ
Any feedback or observations you've seen over the years from your personal experience is always appreciated.
?ÿ
BC,
Been working on the same type of process with our missions so appreciate the time and effort you put into this work. Very helpful.
Question; how do you perform your circular or oval mission around the building?
Thanks,
John
BC,
Been working on the same type of process with our missions so appreciate the time and effort you put into this work. Very helpful.
Question; how do you perform your circular or oval mission around the building?
Thanks,
John
My pleasure,
Pix4DCapture has that option built right in. The only issue is Pix4DCapture doesn't support terrain follow so if you're trying to capture an area with significant topographic relief, keep that in mind. A work around would be to fly a double grid in map pilot pro and manually adjust your gimble to the desired angle once the drone is in the air.
Pix4DCapture also supports free flight mode that will capture a photo every X meters horizontal and Y meters vertical. This gives you total control to really get into some tight spots to get extra images. Just make sure they have enough overlap with your typical oblique/nadir images so that your photogrammetry software will be able to align everything and stitch your images to each other.
I just read a paper where they explored this issue in depth and the concluded more or less the same thing. The best was a nadir/oblique combination, but just off nadir was about the same. Adding in non grid images made the results even better.
Frankly, I stopped doing circles and just accompish nearly the same with free flight instead. The thing in both grid and free (and circles) is to fly one low and in, and another up and out from the first.
I've found that I really don't need multiple grid flights at different levels on most cases. It all depends on the topography, size and scope of the project though.
Any chance you have a link to that paper? I'd love to read it.
https://www.mdpi.com/2072-4292/11/3/239
Long and detailed, but worth it and the conclusion are pretty clearly put.
