Print

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

[Starver]

In all seriousness, inherent compression by maintaining everything down to the seeded background fluctuations until needing direct inspection, maintain deltas/diffs on anything poked and prodded out of consistency of this for offloading from the near cache. Undisturbed (by outside influence) planet need not be stored on anything like a 1:1 mapping to the planetary simulator hardware's piecewise substrate.

In all on-topicality, have we established that spherical geometry is incompatible with a euclidean/manhattan-like simulation/visualisation? Though we needn't tell the residents of the simulation that what they are experiencing isn't 'real', and when they try to develop their own Grand Unified Theory/Theory Of Everything they may well be content that a sphere-less cosmos is natural, however many other problems they have with their GUTs and TOEs.

[KittyTac]

You would still need an awfully big computer for the active parts. Simulating a planet to a significant degree of complexity is not practical.

[Starver]

Random pick (making me annoyed, because that'a one of my long-term development ideas): A universe simulator with photorealistic planet surfaces, apparently, with time-manipulation. And minimum spec is Dual-core 2 GHz, 4 GB RAM, NVidia or AMD/ATI 1 GB VRAM, Windows XP. I might give it a go when I get home.

Spoiler: But this isn't much to do with the OP... (click to show/hide)

No, no immediate sign of interactivity, but I could imagine deltas (saveslots) holding details of player-sites interacted with, their spacial and temporal influence (at least until you wind over beyond the time-cone edge, and maybe further over into prior causality) allowed to influence the next active site with suitable ripples of reality. You could have a procedurally-led landscape down to the levels of grains of sand or less. Should you wish to go poking around drawing patterns in the sand, you can revisit them until the procedural wind blows them away, or observe the simulated beings stumble as they are distracted by your 'nazca line' messages.

Should you encourage entire civilisations towards ultimately developing a stellarly self-destructive possibility, across keypoints of a galactic plane, then sit out somewhere along the galactic axis watching as "Chad was ere" gets picked out in cosmologically-observable death-throws by the noted pinpricks of 'simultaneous' stimulated supernovae, the initial sand patterns themselves having long since become artifact data-points at best, beyond their backups.

I see some philosophical points here, based upon assumptions about perceptions of complexity by those above/below/within potentially multiple layers of simulation. I'm merely positing the counter here because some people seem to think it axiomatic that it can't be done. Maybe our simulation is specifically spawned with the "they overwhelmingly think you cannot sufficiently simulate the real world" tickbox in the LNP launcher, which sets up the configuration circumstances in which this variable is treated as an axiom by the simulants, because it marginally helps FPS or even is just more distracting when trying to Megaproject than even the Economy element... They maybe also commented out the Slooj in the raws, because slooj-workshops are perceived as entirely game-breaking as simulated, making things too easy.

But this is more a lower-forums General discussion (that I've seen/been involved in already, I'm certain) not a practical DF Suggestions matter.

[KittyTac]

Random pick (making me annoyed, because that'a one of my long-term development ideas): A universe simulator with photorealistic planet surfaces, apparently, with time-manipulation. And minimum spec is Dual-core 2 GHz, 4 GB RAM, NVidia or AMD/ATI 1 GB VRAM, Windows XP. I might give it a go when I get home.

The planets' physics are not simulated. They're just textures. And no lifeforms AFAIK. If they somehow collide, they just pass through each other. It's pretty much an interactive screensaver (and a good one at that). You cannot comprehend the processing power needed to simulate our universe at an acceptable scale.

[Starver]

Maybe I cannot comprehend it (or actually you, as I'm maintaining that it might be comprehendible) because our universe is being simulated merely without the element of our being able to easily comprehend it. It still wouldn't rule out it being otherwise simulated.

As I said already. But I shall not repeat myself again again again, just as I'm hearing you tell me what I already said about that space-engine thing.

Round earths/dorfworlds are the intended subject of the thread, not metaphilsophy (<= ok, make that the very last repetition I intend to make) and apologies for perpetuating that tangent. Other than back upon the original subject, you now get free reign to get the last word - however irritatingly point-missing you might do it.

[KittyTac]

I am not denying that our universe is simulated. It's just that our parent universe must be a few orders of magnitude more computationally-powerful than ours (hard to comprehend, huh?). I am just denying that we can simulate an universe that is exactly like ours. You must have misunderstood me.

Thank you for not resorting to walls of meaningless text and/or flaming. Now, let's take this to PM so we do not derail this too hard.

[Shonai_Dweller]

Maybe I cannot comprehend it (or actually you, as I'm maintaining that it might be comprehendible) because our universe is being simulated merely without the element of our being able to easily comprehend it. It still wouldn't rule out it being otherwise simulated.

As I said already. But I shall not repeat myself again again again, just as I'm hearing you tell me what I already said about that space-engine thing.

Round earths/dorfworlds are the intended subject of the thread, not metaphilsophy (<= ok, make that the very last repetition I intend to make) and apologies for perpetuating that tangent. Other than back upon the original subject, you now get free reign to get the last word - however irritatingly point-missing you might do it.

And as has been pointed out (a few times).
1) That's what the map rewrite will accomplish (wraparound where necessary - obviously not for flat worlds riding on the back of giant toads)

2) They won't be round, but donuts (Toady says his maps won't make round worlds).

So there's nothing really to discuss. It's happening during the Big Wait. The end. Back to simulated universes.

[voliol]

Maybe I cannot comprehend it (or actually you, as I'm maintaining that it might be comprehendible) because our universe is being simulated merely without the element of our being able to easily comprehend it. It still wouldn't rule out it being otherwise simulated.

As I said already. But I shall not repeat myself again again again, just as I'm hearing you tell me what I already said about that space-engine thing.

Round earths/dorfworlds are the intended subject of the thread, not metaphilsophy (<= ok, make that the very last repetition I intend to make) and apologies for perpetuating that tangent. Other than back upon the original subject, you now get free reign to get the last word - however irritatingly point-missing you might do it.

And as has been pointed out (a few times).
1) That's what the map rewrite will accomplish (wraparound where necessary - obviously not for flat worlds riding on the back of giant toads)

2) They won't be round, but donuts (Toady says his maps won't make round worlds).

So there's nothing really to discuss. It's happening during the Big Wait. The end. Back to simulated universes.

Heh, if the simulated universes derail moves to PM (or another thread not in the suggestions board) that's probably better; makes this thread it easier to bump in the future if someone finds something interesting to add to the actual topic of this thread.

[Shonai_Dweller]

Maybe I cannot comprehend it (or actually you, as I'm maintaining that it might be comprehendible) because our universe is being simulated merely without the element of our being able to easily comprehend it. It still wouldn't rule out it being otherwise simulated.

As I said already. But I shall not repeat myself again again again, just as I'm hearing you tell me what I already said about that space-engine thing.

Round earths/dorfworlds are the intended subject of the thread, not metaphilsophy (<= ok, make that the very last repetition I intend to make) and apologies for perpetuating that tangent. Other than back upon the original subject, you now get free reign to get the last word - however irritatingly point-missing you might do it.

And as has been pointed out (a few times).
1) That's what the map rewrite will accomplish (wraparound where necessary - obviously not for flat worlds riding on the back of giant toads)

2) They won't be round, but donuts (Toady says his maps won't make round worlds).

So there's nothing really to discuss. It's happening during the Big Wait. The end. Back to simulated universes.

Heh, if the simulated universes derail moves to PM (or another thread not in the suggestions board) that's probably better; makes this thread it easier to bump in the future if someone finds something interesting to add to the actual topic of this thread.

Ha, well yeah. That would be better.
I guess the one thing left to solve is a clever way that Toady's new tiles could actually be put together into a globe without having to make the whole game 3D or something silly.

[Starver]

Graticule. Requires vectorisation (or variably resizing transformation of the raster data) to deal with the effectively trapezoidal or (upon meeting the poles) triangular shape. And that skews neighbouring tiles round, obviously, asking for increasingly noted twists.

It goes beyond TWBT/proportional-font support to rendering projections, though may be a currently unused subset of functions in the libraries used for that.

And then you have to decide if you deal with tiles as probably one of three optional setups:

• Essentially equal-size tiles (disco ball) that disjoint between latitudes (fudging the gaps as much as possible),
• Stick with fixed lat and lon for notably thinner tiles with less effective substance to them as you start to near the poles,
• Choose equal-area so that they get thinner and taller but at least contain a 'unit' quantity of surface area (and are very distorted at more extreme latitudes, even if distorted the other way at the equator to mitigate this.

If going for the "blown cube" method, you maybe use six distinct gnomic (or similar) projections for four equatorial faces and two polar ones, but the extremes of each "square" face have highly distorted square tiles that at the corners exhibit 120° internal angle to match the two adjacent corner-meeting 'square'-maps. Again needs a whole graphics/display rewrite, and a choice of whether/how to maintain an equal-area quality across all tiles.

Personally I favour starting from scratch with an icosahedral supergrid.

Spoiler: Details here (click to show/hide)

The twenty triangular faces are subdivided into 'blown' triangular/shark-tooth subdivisions. The twelve supergrid corners require the triangle corner involved morphed from the natural 60° to 72° (noticeable, but transitionally easy to ignore) as the subdivided hex-grid (of sub-triangles) format becomes, at each of those points only, pentagridded at the centre with hexes radiating. But with the general distortions approaching that point.

With further subdivisional trickery (introduce a procedural sideways morphing that jiggles the skewing function around sideways by subtle degrees to disguise the obvious trend towards a corner, unless you stumble upon the exact spot?) you can make for a blended vectorised landscape that can be further subdivided with triangles within triangles for as low as you wish to go and remove most evidence of tiling across rolling landscapes and allow for cliff-edges/riverbanks/shores etc at fairly arbitrary angles not relying upon the privilidged coordinate system in use but possibly more natural and player-moldable into entirely unManhattanlike layouts.

But that'd be a ground-up rewrite that I'd never expect to happen with DF. It'd be a far bigger change than the transition from 2d to 3d, and potentially a 'second lifetime project' for Tarn to start upon, with not enough obvious advantages. Plus it definitely sends it towards 3d-like calculations to maintain (even if you stop the interface becoming more than 2d slices of the local map).


(I may have missed the point of the need to point out to me that it needed the graphics rewrite or that dumb wrapping leads to donuts (except they should then still be trapezoidally tiled, please note!), but it was on-topic so I read through it anyway)

[KittyTac]

Just ignore the gaps. It's not like you can actually fall into them, so let's have a disco-ball world.

[Ninjabread]

The closest to a true sphere I can think could work would be something like this

Spoiler (click to show/hide)

         □□□
      □□□□□
    □□□□□□
  □□□□□□□
  □□□□□□□
□□□□□□□□
□□□□□□□□
□□□□□□□□
□□□□□□□□
□□□□□□□□
  □□□□□□□
  □□□□□□□
    □□□□□□
      □□□□□
         □□□□□
             □□□□□
             □□□□□□
             □□□□□□□
             □□□□□□□
             □□□□□□□□
             □□□□□□□□
             □□□□□□□□
             □□□□□□□□
             □□□□□□□□
             □□□□□□□
             □□□□□□□
             □□□□□□
             □□□□□
             □□□

Walking off of one semicircle horizontally sends you to the corresponding side of the other, flat side to flat side, "curved" side to "curved" side.

Would probably seriously mess with directions though, because A) special wrapping rules, and B) north in relation to the tiles and north in relation to the poles are not quite the same thing, and the latter can change based on your position, so much so that it can be the direct opposite of the former.

[Starver]

Just ignore the gaps. It's not like you can actually fall into them, so let's have a disco-ball world.

This is how the world map might be. A meridian line for reference (north-south travels directly up/down it), the rest #s where north/south is increasingly disjointed. The arrows indicate "go to the other side for west from west edge or east from east edge" and north/south from any leftmost or rightmost edge sends you to the leftmost/rightmost of the next row (however many columns it is away) with other non-meridian norths/souths being proportonally between the two.

The N and S are polar world-tiles which are respectively North/South from any of the adjacent row of eight (nicely tuned to be eight tiles so that the eight cardinal/and semicardinal directions from the pole send you to just one neighbour-latitude tile. But other attempts to go (say) 1N,1S might round you back onto a different tile from where you started this double-hop.

Code: (Example world) [Select]

                     N
                 <###|####>
             <#######|#######>
          <##########|##########>
       <#############|#############>
     <###############|################>
   <#################|#################>
 <###################|###################>
 <###################|####################>
<####################|####################>
<####################|###################>
 <###################|###################>
  <##################|#################>
    <################|###############>
       <#############|#############>
          <##########|##########>
             <#######|#######>
                <####|###>
                     S(That's mathematically tuned to 10° of latitude per tile but slightly under 8° of longitude per tile on the equator (the easiest way to end up with the convenient 8 pole-adjacent tiles!), increasing as the absolute latitude increases. And rounded, so really there's fractional tiles missing/extra to get them to meet exactly half way round in an integer number (gaps and overlaps, shared around the ring). No easy way tune that out without using dissimilarly-sized tiles according to latitude.)

Obviously a real-world would have more than 479 tiles and not take up the entire world. This is just a zoom-out. And if you were cursoring around the globe you'd probably spin the latitudes around (pop #s off one side, push them onto the other, at the right time to make sense to keep the |-row vertical) so that direct north-south lies upon the meridian at all times and it's only distorted at the new fringes.

Zoomed in, you'd see a subset of the map, which could be rendered 'flatter' and with less obvious disjoints (you may still happen upon 1N+1S≠OriginalPosition problem, periodically) until you start to get quite close to the pole. At and over the pole you get far more than eight neighbouring tiles around the pole (technically). So maybe falling back upon the 'pumped-up cube' model to map polar regions in a better grid format, as already discussed.

Also check out alternate map projections with various solutions/new-problems to mapping your grid (discoball or otherwise) to the sphere.

(Slightly ninjaed?)

[Rufflikerex]

What the hell happened to this thread? Poor op just wanted something simple, and you damned elves had to start complicating things. 

[Putnam]

Wait, since this doesn't seem to have been focused on enough:
OP does not want something simple. It is an actual mathematical fact that it is impossible to make a flat projection of a sphere without distortion. It takes highly complicated math or strange abstractions to make a proper spherical world built out of flat tiles.