Something bugged me when I posted that photo on the weekend of the four different backgrounds for the pasture: why had the barn turned out so much smaller than the aluminum version? As I thought about it, I realized the fence in the foreground had needed a lot of shaping to match the ground profile. So, I realized I must have traced the photo before it was scaled.
Fortunately, the scaling and cutting anew is easy with a machine. So I scaled the tracing up, and produced a new jigsaw puzzle. With the original mock background now in the recycling, I am forced to revert to testing in situ.
Even though the second version is scaled so that the barn is slightly shorter than the aluminum (which was traced from the mockup), the buildings seem too high. Now, the large expanse of plain brown card might be tricking my eyes to see a more imposing scene than they will once the card is painted.
However, the the whole horizon is higher than it was in the mockup. which, let’s face it, was not very scientifically derived. So the mockup was inaccurate, but sometimes you need to enhance reality to make a nice image. I’m inclined to take a half-inch off the height across the board to bring the higher background closer to the vision.
Pembroke:87 
The colors may well be making the ‘rational’ decision harder. As your pictures show you prefer the larger barn, however it may be worth dropping the images into monochromatic format, and editing out the lighting strip. I do it regularly when checking placements and volume, as it literally removes any influence the real colors have. I’m then working with a greyscale image and it’s wysiwyg with no external distractions or influence, the shapes, volume and placement ‘pops’
Edit to above should read ‘I prefer’