As a metaphor, the horizon carries many connotations. We see it as a point of reference, a place to aspire to; or something to move beyond to satisfy an explorer’s curiousity. As a reference point it helps us orient to a location. In school and in art classes, students are taught the horizon is always at eye level. In layout design, this presents a conundrum, whose eye level?

Two primary schools of thought have arisen over the years. One places the imagined horizon line at the eye level of a scale figure, the second puts it at the builder’s standing eye level. I’ve illustrated this second idea in the photo below. While not precisely at my standing eye level, the horizon in the photo (which I consider to be at the base of the distant woodland) placed on the backdrop is close.

To be quite fair, this photo isn’t the best choice for many reasons. Obviously the grasses in the foreground of the photo are out of scale with the 3D scenery and the colors of the two don’t match at all. That white barn is floating in space way above the roofline of the mill in the foreground. Hmmm. Let’s look beyond the flaws of the photo at some basic principles that are being violated here.

Principle No.1
The land in this scene is supposed to be flat. The field in the photo and this area of the layout both represent flat ground, yet the disconnect between the two should be clear to see. Why? Because with the low angle that I shot the scene from, the horizon isn’t where we expect it to be; it’s too high on the backdrop. Over the years I’ve learned to filter out other visual distractions when evaluating things like this and even if I move back to look at the scene, it still doesn’t read correctly to my eyes (below).

When we look at a modeled scene, we automatically search for the visual clues that we’re used to seeing outside. When those clues are missing or handled wrongly, the illusion of realism we’re trying to recreate suffers. One of the strongest visual clues is the placement of the horizon line on the backdrop. Obviously, mountainous terrain deserves a different treatment than flatter portions. Regardless of the modeling scale though, for most applications, it should be located close to or at a scale figure’s eye level and not the builder’s. The only way it would work at the builder’s eye level is if the layout is also quite high to match, therefore keeping the relationship of the horizon in the scene consistent with what we’re used to in the full-size world.

Principle No. 2
There are visual relationships that have to be maintained.
The reason my sample photo doesn’t work is that there is to much space to fill between the horizon in the photo and the surface of the layout. The grasses in the foreground of the photo are out of scale because they were close to the camera, yet the overall 3D scenery has an implied distance from the viewer (nearly 100 scale feet from front to rear). Outside at that distance, you would not see that much detail in the grasses, nor would they be so tall in relation to the scenery in front of them. Even if the colors were a better match, the realism would still suffer because an important visual relationship has been violated; namely that objects get visually smaller and less distinct as they recede into the distance.

In this photo I lowered the horizon closer to where we expect it to be. Even though the edges of the photos are clearly seen, and there’s a bit of mismatch in the tree line, you should be able to “see” the flat nature of the scene consistently going from front to rear and continuing onto the backdrop. Let’s add the white barn from the first photo back in and see how it looks.

Much better, don’t you think? I’ve lowered the horizon a little more and visually, the barn is down on the ground where it belongs and appears to be some distance away behind the mill, adding that sense of depth and volume so sorely lacking in model scenery.

Once again, even though the photos are clumsily pieced together (the horizon has a slight tilt to it) and mismatched in terms of color with the 3D foreground scenery, the overall effect of the scene comes though: our eyes read the open field behind the mill and want to go “back there” to the woods in the far distance. Let’s push the visual trickery farther.

I replanted the big foreground maple tree next to the mill building. Now the eye has multiple points of reference. Along with the two buildings, you can compare the actual foreground tree to the one in the background and the visual clues are what you expect to find. The scene is convincing even with all the raw edges clearly visible.

Principle No. 3
Colors and details of the layout and backdrop need to hold together, but respect the laws of visual perspective.

Another contributing factor to the realism is that the level of detail on the layout and backdrop are in the correct relationship to each other. Hopefully, all the relevant details you would expect to see in the outside world are present in the modeled scene. There are no oversized blobs of plastic passing for spikes on the track or clunky rail joiners and ugly feeder wires to destroy the illusion I tried to create. In addition, I hope the textures of the scenery read accurately as the eye moves around the modeled area.

There’s a debate over the use of high resolution photos for the backdrops. One camp feels they distract because they feature too much detail, the other feels they contribute greatly for the same reason. I’m in favor of using hi-res photos. When handled properly, they add more than they distract. In spite of my art skills, I never really considered painting the scenery on the backdrop. I’ve seen some examples that were well done, but my view is that unless the artist is very skilled, it’s easy to ruin the effect. In the case of painted background scenery, simple shapes and careful coloring are more than enough. For my take, the level of detail on the layout should guide the amount of detail included on the backdrop after taking the laws of visual perspective into account.

I’ve only covered the tip of a big iceberg in this short post. We’ll visit this subject again.

Regards,
Mike