For completeness:
The horizontal/vertical values are simple the raw hex values of the numbers input in the horizontal/vertical boxes. So if you put 1920 in the vertical box, you'll see the hex value 8007.
Going backwards...
80 07 == 07 80 (Converted to big endian... Normal left to right view that us humans are used to
)
0-----7----8--0
4096 256 16 1 (Notice these values... They're base 16. (1*16 = 16), (16*16 = 256), (256*16 = 4096), etc.)
(7 * 256) + (8 * 16) = 1792 + 128 = 1920
And that's all there is to it.
The combined hex field is a bit more complicated. Unfortunately, I can't explain it so briefly. There is an industry standard that some genius figured out called IEEE 754. It uses something called two's complement. The important thing you need to know is if you take 1920/1080, you get 1.77777...8. The hex value of 1.7777...8, according to the IEEE 754 standard, is 39 8E E3 3F. That's where the sign, exponent, and mantissa come into play. If I'm being honest, I learned IEEE 754 at some point in college, but I don't remember how it's calculated and I don't particularly care. It's one of those things that I can always look up in a book or online.
To show how irrelevant it actually is to know, I wrote this program, which perfectly represents IEEE 754, without referencing the specification.