EDIT: So I played around a bit with ACDSee. I exported the same DNG (the one I linked to above) in all three color spaces (sRGB, AdobeRGB, and ProPhoto RGB) and found the Adobe RGB to be the brightest and closest to the pic displayed in Lightroom. The sRGB was darkest, with the ProPhoto almost identical to the AdobeRGB.
With a completely different DNG, I exported to .jpg in all three color spaces and found that sRGB was the brightest and matched the display in Lightroom exactly.
WTF!?!?
sRGB/Adobe/ProPhoto
OK, it is a gamma issue (with the cat pic). Here is what is up. Standard Windows gamma is 2.5, sRGB gamma is 2.2 (and for reference, standard Mac gamma is called 1.8 but it is actually 1.72, but that doesn't matter right now). I think what is happening is because we are both working on calibrated monitors (with gamma set at 2.2), Photoshop and Lightroom see that the monitor is calibrated down from 2.5 and boost the gamma slightly to compensate when viewing on the calibrated monitor. Windows (and most other applications) don't compare the gamma that is dictated in the profile with that of the monitor, so you are viewing a 2.2 gamma image in what Windows thinks is a 2.5 gamma space, and therefor looks dark. You notice this most in the cat picture because it is already dark to begin with.
With the candies, because they are pretty bright and colorful, you won't notice the gamma issue as much. So when you exported sRGB (which is the only color space that Windows picture viewer displays correctly in), the candies will look correct in sRGB, the colors will be right on, they might be just a touch darker, but right on color ratio wise.
When you export the candies as ProPhoto or Adobe RGB, they are going to look pale and washed out (this is happening with the cat as well, but since there isn't much color in that shot, it just looks lighter). The reason for this is that each of these color spaces can only define a color between 0 and 255. So in all spaces 255, 0, 0 is going to be red, but there is no common agreement on what color red is, or how red is red. So while a given shade of red might be defined as 255, 0, 0 in sRGB, Adobe RGB has a still deeper, more vibrant red because it has a bigger gamut, so it might define what sRGB considers the most possible red as 220, 0, 0 (as opposed to sRGB's definition of that shade as 255, 0, 0) because it needs more room to define that deeper more vibrant red. Now in your file you converted to Adobe RGB, you are calling out that color as 220, 0, 0, which is fine if the application displaying the image knows that Adobe RGB 220, 0, 0 is the same red as sRGB 255, 0, 0. Now if we simply drop the Adobe RGB tag and replace it with the sRGB tag but keep the numbers the same, we are now asking for an sRGB color which is 220, 0, 0 - a red which is a lot less saturated than the 255, 0, 0 red that we really wanted. This is what happens when you view a non sRGB image in a non color aware application, it just throws out the color space tag and assigns it to sRGB.
That was a lot of gabbing, but none of it solves the issue that we are seeing in the cat pic (and really all pics, but it will be more noticeable in the darker pics). I don't have an answer yet, but at least we know what is going on. I want to look at that cat pic on an uncalibrated PC monitor and a Mac monitor tonight to see what it does.
---Edit---
Hmmm, so on an uncalibrated monitor, the sRGB cat looks pretty good, and the dark tones in the ProPhoto image look posterized (which is slightly expected on this crappy monitor). Now I need to wait for the wife's Mac to come home. I'm glad you posed this, gamma is something I haven't given much thought to. Now we just need to figure out how to solve it
That's no good. What about it, the IQ, focus, or just it being a prime?
)