Created attachment 262 [details]
Examples of varying focal length and imaging distance.
Preface: My calculated pointing angles to a particular spot on the ground were not relating to the resulting DIRSIG imagery. Therefore, I was working with nadir viewing geometries when I stumbled upon this behavior.
Using the ms2010 scene, I set my focal length to 40 mm, my platform position in scene ENU coordinates to (2860m, 1760m, 22000m) with pointing angles in Platform ENU coordinates to 0,0,0. The result is the far left image in the attachment.
I want to reduce my GSD, so I increase my focal length. I increased my focal length to 134 mm. Preview mode provides me the middle image of the attachment.
I increase the altitude of my platform to 34000m, maintaining the focal length at 134 mm and my preview results "go off the reservation" (shown on far right of attachment). I only clip the bottom part of the actual scene. Now, I did not change my x and y coordinates. I didn't change my angles. What could be happening here? Is there an upper bound on the altitude of the platform?
I'm using only a simple atmosphere (no MODTRAN) and I'm running DIRSIG 18.104.22.16886.
I can provide the .ppd and .platform files to any who are interested.
Furthermore, I have no angle or position offsets in the platform mount and camera. Simply put, why do my scene X and Y coordinates change with Z? They shouldn't be changing because I'm working in scene ENU coordinates.
Created attachment 264 [details]
Platform file with mounts correcting for the ms2010 scene angle bias.
There is an angular bias in the ms2010 scene. I've corrected it with a coarse correction to the platform mount and a fine correction to the camera mount in the .platform file (attached). I'm within 1 meter accuracy at altitude of 400 km. Now, when your angles are 0,0,0, you'll be pointed nadir at whatever x and y of your platform position.
This is a workaround. I hope this bug just affects the ms2010 scene and not micro or megascenes, but I haven't tested it.
Created attachment 265 [details]
.scene file of ms2010
This sounds super fishy. You have it marked as using 4.4.1, is that correct? I put a preview build of 4.4.2 out two weeks ago. I just want to make sure I test correctly.
(In reply to comment #5)
> This sounds super fishy. You have it marked as using 4.4.1, is that correct?
> I put a preview build of 4.4.2 out two weeks ago. I just want to make sure I
> test correctly.
Correct, I am using 4.4.1. I worked this out on the phone last week with Mike Gartley, and it was user error. I was playing with jitter some time ago in the ppd file and forgot to remove it. Hence, the random pointing location on the ground. Embarrassing, yes, but also a good case study in what to look for when you think the sky is falling.