Hellow guys,
I was reading in the file below about the statistics of the generaion of points (number of points) based on image size and quantity. The first example sites 500 images. I have been working with not more then 30 images (12MP - 14MP). I am wondering if I am not shooting enough images.
What is the “typical” image count you are seeing from datasets with good resolution and accuracy and information (points)
Hi, if you shoot 30 images, you will have a big percentage of images in the “borders” of the project, so in absolute numbers, the amount of 3D points obtained will be lower, but it does not means that the project results are bad.
What you should do is always fly a larger areas than the area to study, then the area of interest have always images in x-y and therefore more points are found.
The different graphics of the quality report indicates you if the overlap was good and if the amount of keypoints for the project is good, at the same time, just process step 2 and check if the results are enough good for you or not, if not, fly with more overlap and lower altitude.
That makes sense and thank you for pointing that out. (border images not generation many points.
It is typical/realistic to fly the same areas more than once in terms of workflow? I am thinking you want to do it “right the first time” in an “ideal” situation - is that truly achievable ?
Hi, if for any reason you did nt get good amount of points in the borders, then you can fly again, you can then process with both datasets or just the last one.
If you use good geolocation, then you can try to fly only the borders, but I would not do it.
Just for this time is needed to fly again, as a general work flow, always fly larger area so you do not need to fly again or two flights. So the ideal situation can be achieved with one flight.
Thanks, I have been noticing that is the best way to do things. I appreciate you input.
I tested different scenarios to find the optimal ‘flight height vs image spacing’ for eg 100m above ground (about every 5 seconds). however, I would like some software suggested selection if I have way too many images in a folder and have the software decide what is the best selection of images to still give an accurate surface. my folder system is getting full ito Gb with too many images.
I also have to use thirdparty software to first reduce the size of my images before I start the pix4d processing. i.e. typical 6Mb images are reduced to 1Mb and gives me the same resolution Mosaic. I think the chase for higher res images is sometimes not worth the waiting/processing time for what the result gives you.