Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Mike Mayday]

I'm well aware that the moment when a projection different than top-down is even considered is a song of a distant future. Nevertheless I'd like to explore the mechanics and benefits of employing such a method of display.
After many discussions, we have agreed that for a functional display, oblique/cavalier projection works best.
To start gently, I've devised a mix of oblique and ASCII that requires no new graphics except for a few bases which I've already made (floor tile, wall tile, ramps and stairs):



The height of the wall could be dynamically modified. In an optimal situation, the tip of a wall tile would only reach as high as the middle of the floor tile behind it. This way, a single wall tile would only obstruct at most 1/4 of any object.

The greatest benefit of a 3d projection is being able to view multi-level constructions and many levels at a time. To clearly display which level you're currently viewing, the lower levels could get less saturated colours. Here we're looking at the top of a tower at night, the lower dwarf also carries a light source apparently?


Floating/flying/falling units would have a spot of shadow on the ground below them, so that you can tell where exactly they are. This shadow would be visible even if the layer where the creature is was sliced off.

I've been doing some tests on how that display may then be used for "prettiness" (since it's bound to start happening) and basically a welcome possibility would be to have tiles that have graphics larger than the space they occupy:



The original post is located here:

Spoiler (click to show/hide)

Heyo
I've been making some mockups to see how the cavalier projection would actually work. So far I can see two methods of applying it.

1.The "clever" method - in which a wall fits perfectly in the tile it occupies, but graphics are used depending on the surrounding walls. In other words this:

translates into this:

Take a closer look at how the tiles translate to iso. This method has a great advantage- walls will never obscure an object behind them. Unfortunately, it has its flaws. For example, a one-tile opening looks like it's almost two tiles wide (See the image above). Another problem appears with multi-level constructions:

Look at the tile SE of the upper chest- this is actually a wall tile, but it looks not right in conjunction with the lower wall.
Just in case- here's the view of the lower level (I'm using rotated 32px tiles BTW).


The second method is the "clean" method. Everything is properly placed and there are no 3d or size problems. However, in order to prevent the walls from obscuring the view, they have to be kept rather low (in the "clever" method their height is more than enough). Here's a quick mockup comparison (wall tiles highlighted red):

The alternative on the right would not obscure any more than the middle method.

Thoughts? Ideas? Feelings?

[Duke 2.0]

 I applaud you on bring this back up again with more mockups, as this issue is a fun one to mess around with.

I feel no matter what method we use, there will always be problems. We should just focus on a system whose problems don't impact gameplay as much.

 I rather like the clever method. We already have to view the levels individually, so this system just makes things prettier. And we all get used to the quirks of various tilesets, a simple thing like opening looking a little wide won't be a problem.

 I would like to see these mockups with height shading(Lower displayed lavels are darker/lighter to show depth) and the clean and alternate methods used with a multi-z-level construction. Preferably showcasing a small tower and a small pit.

[AncientEnemy]

I prefer the clever method, I wouldn't mind only viewing one z level at a time anyway. plus, you could have a multi-z level view where higher z levels show up transparent, or simply use 'xray vision' where anything that would obscure a lower level isn't displayed at all

[Neonivek]

I am probably the only person but I like Alter the most.

[Jadael]

Alter most closely matches the current setup, with each wall section being a tile.

[Sunken]

I'm not saying it's better, but what about the version where a "thin" wall runs throught the middle of a tile, rather than along one side? Maybe the geometry gets more complicated. It'd be somewhere in between "alter" and the other two, I guess.

[Grigorij.Gluhov]

hahaha! Ultima 7? 

Much better to use simple 3d-engine than old tile isometric. It will be simple to make and use.

Forget to say... Good pics! 

[Tack]

just one quiery. will the tiles interact with a door to create more wall to fill the gap?

[Grimlocke]

I like these drafts! quite well made 

And I dont quite agree with it that making a 3d engine would be better then an isometric one. Isometric suits DF much better if you ask me.

Anyways, back onto the actual purpose of the topic, I like the clever method the best. The height of the walls looks better, and it wouldnt obscure anything.

Wouldnt you be able to solve the wall conjuction issue by moving the z-level on top downwards by the walls diameter? That way the base of the top wall over overlap the top of the base wall nicely.

Though the alternative one might be the most realistic in its depiction (not showing corridors wider then they are), I would preferr it over 'clean'.

EDIT: typos

[Tormy]

I am probably the only person but I like Alter the most.

Nop, you are not the only one.  I also like Alter the most....however I still say that: 2D top-down > isometric in DF.

[Volfram]

I'm still a bigger fan of the "Depth by Darkening" ideas(particularly mine, "Depth by Scale") largely because I think they would be much faster to impliment than anything approaching isometric, and they do go a long way towards improving the appearance of the game.

[edit]I didn't say it, but I was thinking it when I posted this, those mockups are impressive, and if implementation time weren't an issue(which it is, for me), I'd definitely be more in favor of them.  I prefer Alter, due to improved clarity and reduced number of possible variations the computer may have to go through depending on the adjacent tiles.[/edit]

[Randominality]

another vote for lter here

[Bricks]

Clever seems nice for single-layer depictions.  The one-tile pillars look awful with the first two methods, as they don't really depict what is going on.  I think alter is the most honest depiction - could you possibly mock-up alter w/ transparency for obscuring tiles?

I don't think a 3D depiction would outclass isometric.  Iso is simple, clean, in most cases, rotatable, and its more of a standard for this type of game (I can't speak for roguelikes, but it can certainly be said about other building/design games (Roller Coaster Tycoon, Age of Empires, Zeus/Pharaoh/Caesar).  Iso also allows for local texture on each tile without having to simulate each bump.  One clear downside with Iso will be when lighting is overhauled, as that is VERY tough to depict without planning hunderds of tiles, or sacrificing appearance.

[Lap]

2D top-down > isometric in DF.

I always go with 2D in roguelikes, but the huge amount of Z-level switching is the sole reason I might actually prefer a well done isometric version. Is there really a good way to show multiple z levels in 2D?

[Neonivek]

I should say... if Alter loses my vote is for Clever