I took a look at the colour values on the underside of the object and the trees below it and there are reasonably consistent.
Below is a shot from Photoshop showing the colour values (highlighted in red at top right of image) for the darkest single pixel on the underside of the object. Here a value of 69,69,69 was found.
When I broadened the sample to a 3x3 pixel area at the position, I got a value of 74,74,74.
I then grabbed a 51x51 pixel sample (to get a reasonably representative average) of the area with trees in it, which gave 68,70,71 as the result. The sample was taken directly below the object at roughly the centre point between the top of the treeline and bottom of the image. Unfortunately Window's screenshots don't include the mouse point, so the sample point isn't shown in these images.
Unfortunately, the large translucent darkened area that reddit's video play controls are placed in, meant that I need to grab a different image to use for sampling the trees area, than for the underside of the object. So I grabbed my tree sampling image from a few seconds earlier in the zoom in portion of the video. That was to avoid the darkening part of reddit's video player GUI, artificially lowering the colour sample of the trees, while still gettting a good sized sample area of the trees colour.
The underside of the object appearing darker than the ground it should be reflecting is an illusion, caused by comparing a large dark area (the trees in this case), with a very small dark region, in the middle of a much larger bright area (the underside of the object in the middle of the sky).
This illusion caused similar responses from some who looked at the
Balwyn, Melbourne photo (1966)
That doesn't mean of course that the object is a physical one in the scene the video is of. It could just be that a person faking this was competent (or just lucky) in their rendering of the "orb", to get the reflection brightness right. Alternatively it could be a real physical object, such as a balloon.