Print

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

[Drunken]

Forgotten beast rendering in 3D...
It would have to first search through the entire archive history to figure out what it needs.
The engine would have to load a base mesh, add or subtract parts from it, based on physical limbs/head/internal organs, etc. It would then have to do skin/materials, so either give a mesh the right texture, or else completely convert it into mass particles (for things like steam), or run a shader all over it for something like water (which would then require advanced lighting calculating to account for distortion, direction, etc. It would then have to generate tertiary features, such as 'covered with/patchy feathers (More 2D planes, extra eyes, teeth, etc. Then it would have to generate potential places for various additional behaviours (spit, powder, etc) and work out how to render them, the physics behind them, etc.
After that it would have attribute and run scripts for characteristics, randomly generated effects.
It would then have to load animations.
It would then have to apply physics to everything.
It would then have to run AI scripts.

That alone would take a fairly significant amount of processing.

Did you even read my post? It would first have to look up the beasts name, then print it on a white block. Done. I could process that with a pencil and paper. Sure a more complex and pretty system could be devised, but that was not my suggestion and if it is an obstacle to 3d then lets just not bother.

[MasterMorality]

Forgotten beast rendering in 3D...
It would have to first search through the entire archive history to figure out what it needs.
The engine would have to load a base mesh, add or subtract parts from it, based on physical limbs/head/internal organs, etc. It would then have to do skin/materials, so either give a mesh the right texture, or else completely convert it into mass particles (for things like steam), or run a shader all over it for something like water (which would then require advanced lighting calculating to account for distortion, direction, etc. It would then have to generate tertiary features, such as 'covered with/patchy feathers (More 2D planes, extra eyes, teeth, etc. Then it would have to generate potential places for various additional behaviours (spit, powder, etc) and work out how to render them, the physics behind them, etc.
After that it would have attribute and run scripts for characteristics, randomly generated effects.
It would then have to load animations.
It would then have to apply physics to everything.
It would then have to run AI scripts.

That alone would take a fairly significant amount of processing.

Did you even read my post?

Nope.

[Niseg]

As far as I understood rendering is kinda separate from the game itself . Multi threading might be useful but it's not that necessary if you got non-blocking rendering   .  I think isometric view would be easy to add.

I'm not really sure why people think they need a supercomputer to do this. All DF needs is a few algorithm optimizations and it should run fine on anything.

[Reelyanoob]

a simple 3D ascii would also be nice if you ask me(floating characters for mobs, everything else flat/stuck to ground).
No need for cheesy animations!

I'd go stick-figures based on body plan (i.e. tube arms, ovoid head/body, with some new body part tags for size shape, orientation), and overlay textures.

Go look at Spore creature rendering. Something like that would obviously work for DF. EDIT: Just came across the Spore reference a few pages back.

Really the main draw for 3d for me would not be for animations or any such thing, but rather so I can view my fortress in all its glory in realtime!

Not having to scroll up and down through z-levels would be glorious.

Yeah gloriously uncontrollable! How are you going to select anything? Or see what is in a room? Maybe there could be a true roaming 3D camera, but that would make gameplay almost impossible in my view, and be worse than z-levels

My vote is for Isometric cut-away at the "current" z-level, with lower levels shown but greyed out (with the grey becoming black deep below ground). If you want to see the whole thing, you just go up in z-levels, and have a switch for the greying out.

[peterix]

Go play Armok I. I'm sure you'll find the experience enlightening. This is what the game was before Toady decided to scrap the graphics.
/snarky

IMHO, we don't need 'just' 3D. We need 3D tailored to the game in a way that it enhances the experience and gives us more power over the game.

Now, I'm not saying even simple 3D wouldn't help. You can, after all, do the same kind of z-level view, maybe with a visible grid to help the player see what actually is on their current z-level. This would give us the same control scheme we already have in the game (good control over a single z-level) and ability to see more than one z-level (solving the problem of finding those sneaky goblins on a hill). Good enough for the overworld and possibly some simple constructions.

Now, let's take a look where this would have problems. Say you wanted to build a pyramid within a pyramid. You'd never be able to see the whole inner pyramid, because when you look from above, you see the outer one and looking at the lower z-levels gives you only a cross-section. My idea here is to introduce some kind of view-depth-peeling mechanism, where z-levels are abstracted away when looking at complex structures and you rather switch between peel-layers. It would need a lot of development, testing and usability work to get it right, but in the end, it could solve the worst usability problems DF currently has (or at least the problems I see as worst).

Also, creature models, item models or any of that is fluff. Simple sprites with letter textures rendered on them would be just about right as a start.

[Cromage]

Now, let's take a look where this would have problems. Say you wanted to build a pyramid within a pyramid. You'd never be able to see the whole inner pyramid, because when you look from above, you see the outer one and looking at the lower z-levels gives you only a cross-section. My idea here is to introduce some kind of view-depth-peeling mechanism, where z-levels are abstracted away when looking at complex structures and you rather switch between peel-layers. It would need a lot of development, testing and usability work to get it right, but in the end, it could solve the worst usability problems DF currently has (or at least the problems I see as worst).

Now that sounds interesting. You could segment the map and identify peels with a flood-fill algorithm. It doesn't necessarily sound difficult... at least for 95% of constructions and small enough segments. The remaining 5% could be tricky though. I wonder if Stonesense could implement something like this...

[lackofgrace]

tl;dr ( its late, so im just going to bed. therefore i dont know if this is relevant )

ive actually started a experiment for a 3d df esque game ( the setting is futuristic, but that dosent matter )
rendering the terrain isnt a problem, i can render a 256 * 256 * 256 cube landscape with ~500fps
creatures wont be a problem in my implementation, but im faking it sine i use 2d billboards. ( see ragnarok online for example )

the huge bottleneck has so far been liquid simulation.
even getting good speeds with a single liquid type has been hard. since liquid moves around so much i pretty much constantly have to update the water chunks, resulting in over stressing the communication line between cpu and gpu.
just simulating the liquid in a 256^3 world takes so much time i need to do it in a different thread just to get viable realtime fps.

[randyshipp]

Put me down as someone who thinks graphics exactly like Stonesense currently has, plus mouse/windowing support, would be awesome.