creating a environment where pretty much everything is dim and smoothed out in a sort of middle ground of light level
In your video example everything is compressed into the "lower ground" of light level, ie: if you looked at a histogram you would likely see a single "mountain" to the left (dark) end of the histogram. If you point your camera at a dark scene, and make it look dark in the viewer panel, the histogram will show a narrow (steep) mountain to the left. As you open the lens or increase the shutter angle or increase the light you will see the mountain not only move to the right, but it will expand, left to right, as it moves over. You are now using more of the camera's dynamic range to describe the same information. That will mean more separation between tones in that area.
Of course you can expand the range later in software like Resolve, but depending on what codec you shoot and the amount of underexposure, you can "break" the image by trying to do too much in software. This results in banding and accentuating noise, that is then further accentuated by the encode going to H264-H265, which is then re-encoded going into YouTube. Starting with underexposure, trying to fix it in post, then "up resing" it, then going to an end user codec like H-264, then having YouTube encode for their platform compounds errors exponentially as you go along. The best pipeline is good lighting, full exposure, darken in post, output at the shooting resolution to a quality codec like ProResHQ, or ProRes 444, and then upload. Vimeo is a superior platform to show your work over YouTube visually, and don't forget that everything you upload to YouTube becomes property of Google. That's right, your copyright goes bye bye. They own it.
Creativity is the ability to accept ambiguity.