I like this idea, so I did a little GIMP magic and created examples for both the depth-by-darkening effect, and the depth-by-fog effect that a previous poster mentioned. I use a fairly large tileset, so the images are pretty big.
Depth by darkening:
(http://img.photobucket.com/albums/v628/Ace91/df_fog_effect_2.png)
Depth by fog:
(http://img.photobucket.com/albums/v628/Ace91/df_fog_effect.png)
I, too, like Ace91's mockups, particularly the Depth by Darkening one. My idea was to use a slightly different approach, though.
I had thought for a while that a good way to make the display easier to approach, especially with the new graphical tweaks, would be to draw not one layer at a time, but to draw the layer the player is looking at as well as every layer below it(to a certain point. Ten to fifteen layers should be sufficient.
Things such as unmined stone and open areas would be rendered as transparent, so that the next layer is visible below them. The exact render depth should be an Init value.
Every layer is rendered at 80-90% scale of the one above it, with the top layer rendered at 10%. Every layer is centered on the XY cordinate of the tile the player is looking at.
This allows a clear way to show the 3D aspects of the world as well as allowing players to see into those massive pits they've dug out, and in my opinion allows multilayer monitoring better than an isometric viewpoint would allow. I also anticipate that it would lead to more interesting fort design, such as balconies and overpasses, since such structures would now have a visual impact on what the player sees, and it has a minimal impact on the rendering pipeline, because ultimately, it's only an expansion on the viewport settings currently in place.(Render the current layer, the 10-15 layers below them, and have them stacked and scaled in the draw thread, which is almost entirely desynchronized from the game thread)
Another idea I had for this layout was the idea of making the opaque stuff on the current layer semitransparent, with a value in the Init files to let the player set how semitransparent it is. This allows the player to monitor stuff on both the current level and the level immediately below. I wouldn't advise more than one layer of X-ray vision of this sort, as it would get visually confusing, but in my opinion, the DFMA already shows the effect off nicely, so it's been shown that it can work.
An additional idea would be to have fluid layers --that is, magma and water-- be highly(or completely) transparent below the surface, so that a player can see to the bottom of a pool of water or magma at a glance. Perhaps making the surface layer of a fluid normal(or perhaps 30% transparent. It would be another Init value), while the underlying layers are all 90% transparent(another Init value) would produce a nice "haze" effect.
The idea here is to compress as much useful information into the screen as a player can make sense of, in a way that the player can make sense of it.
In short, depth by scaling. I was at first inspired by the way that the DFMA makes the next layer down slightly visible, as well as some playing around in 3Dwarf with a camera pointing straight down and a level slice in a map where I'd built a bridge across the top of a bottomless pit(for the visual appeal i'd anticipated in 3Dwarf. It didn't disappoint)
Through a great deal of thought, I've crafted my suggestion to fit the following requirements.
1: It must be possible to add this hypothetical feature with a minimum of slowdown and, especially, a minimum of development time. This was my number 1 priority, as making it simpler to implement both improves the chances that Toady will actually do it, AND reduces the amount of delay which would be added to overall development if he did decide to. I feel that especially with the new graphics system, this is now a very real possibility.
2: It must be as compatable as possible with present tile and graphics sets. As my suggestion stands, this is absolutely the case, with only the "down a level" graphics being replaced with an invisible square.
3: It must improve the visual effect as much as feasable while adding as little confusing information as possible. This is why I specifically chose scaling. As tested in 3Dwarf, it looks really good, and it adds less confusion than simply changing the colors.(an engraving of a tree in Olivine doesn't look like a tree that's two levels down from the current view, for example) I believe that using scale as the visual depth cue is extremely effective, as it's one of the real-world visual cues we use.
4: It must attempt to compress as much useful information into view as possible, without making it confusing while doing so.
I'll see if I can make mockups later.