1920 x 1080 for bus shelter

At the most basic level, pixels are just pixels. Thin lines must be lined up perfectly on the pixel grid. If you have a thin line in a graphic that is only 1 pixel wide, like a stroke on a path, and the line is horizontal or vertical, and the line is black, and the background is white, and the line does not perfectly fit on the grid (think about a checkerboard) theme the line will be 2 pixels wide and neither row of pixels will be black. 

When you scale down a video, the detail is going to be reduced, especially if it has fine detail, the detail level is going to decrease. This is your screenshot:

I think it looks good. I don't see compression artifacts. I see as much detail as I expect. If the top right closeup was cropped from a larger image instead of scaled down, the detail would be about the same. 

Just as in any photograph or video, you can increase the sharpness, vibrance, contrast, saturation and color balance of the image. If you want an overall increase equal in all of the scaled videos you could add an adjustment layer to the entire composition, apply Lumetri and start fiddling with those settings. I might consider that. If you need additional control over each scaled video, add Lumetri or other color and sharpening effects to the individual layers. 

I don't want to get too far into the weeds on compression, but you also need to understand that color in an MP4 file is limited to 8-bits and color is compressed in blocks of 4 pixels. Very thin lines surrounded by vastly different colors, like a pure red 1 pixel thick horizontal red stroke against a white bakground is going to loose some of the red in the red pixels and the white will become a little less white. I don't see fine lines or a lot of graphics in your layout, so this  comp isn't going to suffer very much from that limitation, but you do have to account for the limitations imposed by compression when you design a comp.

I think that an adjustment layer with Lumetri applied and a little tweaking of the sharpness may all that is required. 

Is this for a printed poster or some kind of video display?

If it is for a printed poster then you need the dimensions of the poster in inches, then multiply those dimensions by 150 for a high-quality poster or 72 for a newspaper quality poster.  

For example, if the poster is 36" by 72" then your After Effects comp needs to be about 5400 X 10800 for high-quality printing or 2592 X 5184 for newspaper quality printing. You would save your hero frames as Photoshop Layers using the Composition menu and then send that PSD to the folks preparing the image for printing.

If you are preparing your project for some kind of electronic sign you need the specifications for that specific piece of equipment. Frame size and video format depending on the device. The most common is just standard HD, but I've produced projects for digital displays that are anywhere from 7000 pixels wide and 100 pixels high to 100 X 100.  It all depends on the device. The only thing that is always consistent is that you never want an odd number of pixels because all color compression is always in blocks of 4 pixels. 1251 X 785 is going to resize to 1252 X 786 when it is rendered so make sure you always follow the standards given by the client when creating projects for digital displays.

I would suggest you actually read the online help on the layer quality switches/ interpolation settings or how to use the Detail Preserving Upscale effect if needed. The rest is not really relevant. For rendering you simply pre-compose your vertical composition into a regular HD comp and render it using whatever standard HD MP4 preset is required. Nobody in their right mind would render actual vertical video for such a scenario.

Mylenium