Beautiful footage. I believe you when you say you are getting better quality from downsampled 4K, however, in the Youtube video you posted, the details are soft, so not a good demo of this (look at the trees) there's loss of High Frequency, smearing, compression artifacts. Of course, Youtube is compressing the hell out of the frames. The best way to tell, would be to shoot the same scene with two cameras shooting 4K and 1080 at the same time. These are the kinds of test we do with clients when they are deciding which cameras to go with. As I said in the previous post, depending on a huge number of factors, you can get worse quality out of downsampled 4K. The most obvious would be what software you are using to create the 1080 downsample. We use software most people won't ever touch, it's made specifically for transcoding from one format to another. I don't know if anyone has compared different Prosumer/Professional softwares like PP, FCP, Avid etc, would be a very useful comparison.
One of the questions asked, was whether converting 4:2:2 or 4:2:0 footage to 4:4:4 would help picture quality. The answer is not really. Same as when you are working with 1080 and upressing to 4k, or taking 8 bit material converting to 10 bit.You are having the computer pad the file with data that wasn't there to begin with and making a larger file. Computers can interpret pretty well, they can create new pixels or frames from surrounding material, but in doing so, it can also create artifacts that are worse than leaving the frame in it's native format.
H265 is an up and coming format that has a lot of interest. It can handle resolutions up to 8K and can record in 422 or 444. Better quality video at the same data rates as the current H264 specs. So, I would say that yes, H265 support will get us better looking pictures with a smaller file size.
I would say, the easiest way to get cleaner video in your final result, sharper images, is to color correct your footage. Adjusting contrast makes a great first step. The human eye is more sensitive to changes in Luma/Contrast than Color, there are 120 million rods (light) versus 6 to 7 million cones (color), adding contrast also has the benefit of perceptually adding saturation giving you a nice, rich picture. The first thing I do, is set the black level (shadows/setup), then the white level (highlights/gain), then I ride the middle ranges (mids/gamma). Next I'll cure any color casts (some software has an eyedropper to set the white balance) by finding something white in the frame and adjusting out any tint. Lastly, I'll use secondaries (choosing a range of values or color) to deepen skies, pop the grass, fix skin tones etc.
At the end of the day, unless you're showcasing your work on a 4K monitor from a hard drive setup to play uncompressed files (a horrendous undertaking that even the film studios don't attempt), it's going to end up getting further compressed. Either the networks will compress it to send it over their broadcast, it'll end up on Youtube, Netflix, Vimeo etc, or you take it to a buddies house and play it from your iphone. Don't laugh, but I have clients calling me to complain that their shows don't look good on their iphones, boyfriends tv, ipads etc, even though they loved it in the $40,000 viewing room that we used to create the finished look. I even had one executive complain that he was viewing on his boat in the marina, and the black and white commercial was looking too blue. I had to explain the image had no color and his viewing copy was compressed to hell in order to email, and to please come into the color calibrated room in order to judge.
So final thought, just go out and shoot, have fun, and don't get too hung up on the tech.