5
$\begingroup$

The BBC News article UK satellite makes HD colour movies of Earth has led me to a series of YouTube videos as well recorded by the Carbonite-2 spacecraft built by Surrey Satellite Technology to be operated by Earth-i.

The first video of Buenos Aires is quite striking, it looks like the viewpoint is not moving, even though the spacecraft is in a nearly circular LEO at about 500 km. https://www.n2yo.com/satellite/?s=43115

How is this possible?

Related: Can image-stacking allow this 0.25 m satellite telescope achieve 0.65m resolution?

Open the videos in a new tab/window and set to 1080p resolution and full screens if you've got 'em.

$\endgroup$
  • 1
    $\begingroup$ Someone should do the math on how rapidly the angles would change based on altitude, and perhaps how long some of these clips are based on vehicle time between known landmarks. A few clips seem to show some perspective shifts, I suspect these clips are all simply short enough that the apparent motion is minimized. $\endgroup$ – Saiboogu Apr 16 '18 at 16:13
  • $\begingroup$ @Saiboogu but some clips don't. The Buenos Aires video is 17 seconds of what seems to be pixel-perfect alignment at 1080p, that's the reason I put it in there first. $\endgroup$ – uhoh Apr 16 '18 at 16:19
  • 3
    $\begingroup$ The two tall buildings near the upper left show some shifting when you flip between the beginning and end of the video. The blue-gray building on the left of the river also shows a difference. $\endgroup$ – BowlOfRed Apr 16 '18 at 17:06
  • $\begingroup$ @BowlOfRed yep you are right! look at that, beautiful! i.stack.imgur.com/vkpGw.gif $\endgroup$ – uhoh Apr 16 '18 at 17:53
  • $\begingroup$ @BowlOfRed I've added an answer based on your suggestion. $\endgroup$ – uhoh Apr 17 '18 at 3:31
7
$\begingroup$

@BowlOfRed's comment nailed it.

I'd thought about making an optical flow version but time does not allow, so I've just used screenshots and imageio found here.

It looks like the spacecraft was overhead, and in fact did move many degrees (as seen from the ground) during the 17 second video of Buenos Aires.

They are clearly remapping frame-by-frame using some reference in the image, and not the calculated surface of the Earth. If you look at the shore of the river which should be relatively constant elevation, the shore oscillates in areas that contain tall buildings, and in the opposite direction than the tops of the buildings. In areas with no tall structures such as the extreme right of the video frame, the shore is motionless.

See this GIF for that: https://i.stack.imgur.com/vkpGw.gif

This suggests that they are nulling something like the average motion in each region of the image, rather than using a calculated ground.

The best place to look to see what's happening is the pair of bridges at the top left corner of the first GIF. The tall buildings at the bottom right corner of the second GIF are also a giveaway.

Sorry for the vertigo-inducing GIFs!

enter image description here

enter image description here

below x2: The pair of (very tall) bridges in the first GIF, screen shot from Google Maps.

enter image description here

enter image description here

$\endgroup$
3
$\begingroup$

The videos are all very short The first one is 17 seconds, while the others don't seem to be any bigger. The altitude is just over 500 km. The distance moved during that time will be about 1250 km. Assuming the pointing is constant, the angle should be at most 15 degrees or so. This is reduced if the image is at an angle. If the satellite is pointed at a 45 degree angle, for instance, then the view angle difference will only be a few degrees.

I suspect that what is happening is the images are processed on the ground to not show any field of view distortion, although the perspective will change a bit. It's hard to know for sure, but I think I see a few slight distortions in the videos shown.

$\endgroup$
  • $\begingroup$ +1 I suspect you are right! ;-) In addition to the distance being farther, the direction of motion is no longer perpendicular to the direction of view, so the angular change per second drops dramatically away from the zenith. $\endgroup$ – uhoh Apr 16 '18 at 17:48
  • $\begingroup$ However, upon closer inspection, the view seems to be fairly close to top-down. It seems that it is not so hard to know for sure. $\endgroup$ – uhoh Apr 17 '18 at 3:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.