Spherical worlds weeeeeeeeee

[j5mello]

Super Mario Galaxy but thats not even remotely the same genre...

[SinbadEV]

Certain Levels of the second Ratchet and Clank Video game...

I know it would be a pain in the butt to do a real sphere based RTS but can you imagine the popularity we could garner with being the first?

[Google_Frog]

Imagine launching a satellite and waiting for it to orbit over the enemy base.
I don't thing this is a feature you would implement, it's a new engine.

[KDR_11k]

Most RTSes don't have spherical battlefields anyway. Planets are HUGE, you can't see individual units when you want to see a decent chunk of a planet.

[manored]

Make huge units, like Shadow of the Collossum style :)

Just a comment, maybe stupid, but instead of simulating a world that bends upon itself giving the ilusion of a sphere, wouldnt be easier to simulate a 3d space and then simulate gravith to make stuff stick to the planets surface? I think that spore is like that, or at least by the videos I saw I SEENS that you dont "enter" planets to interact with then, the space the cities exist and the planets exist is the same.

[Snipawolf]

It would have to be optional for the map.

If this happened regardless of map/mod, it would break just about every map there is.

[manored]

It would have to be optional for the map.

If this happened regardless of map/mod, it would break just about every map there is.

Since the whole thing is probally not going to happen, I think thats the smaller problem :)

[HoneyFox]

well maybe wrapping edges of map is not as difficult as we thought...
like in The Settlers, the map is wrapped both vertically and horizontally, if spring could support this kind of map settings (usually this should be for those epic big maps), game will be quite interesting. and aircrafts' off-screen problem will be solved maybe...

just

Code: Select all

dx=abs(obj1.x-obj2.x);
dy=abs(obj1.y-obj2.y);
new_dx=-abs(dx-map.width/2)+map.width/2;
new_dy=-abs(dy-map.height/2)+map.height/2;

and then we can continue doing other stuff...

plus: pathfinding should also consider these changes...

[KDR_11k]

Yeah now what about projectiles crossing the edge, range detection, adjusting the pathfinding, ...

[lurker]

Aim routines, broken start positions, los, buildings near the edge, metal extraction, map rendering, whether to try to loop it, whether to try to place units so you can see your corners base in one piece...

[HoneyFox]

i donno how pathfinding work. that maybe the hardest obstacle...

but range calculation would not be a big problem. by using the functions below:
vector2f parseVectorWrapped(x1,y1,x2,y2,map_w,map_h)
{ codes are like those above in my reply... }
float calcDistance(dx,dy)
{return sqrt(dx*dx+dy*dy);}
(z axis could be added into that function, but that doesn't matter much)

projectiles crossing the edges? just set its x or y position to 0 or map_x/map_y, and almost nothing else needs to do.

and i've just thought about another thing.

the opposite edges of the map should be matched.
or some strange thing may happen.

[smoth]

bad idea, been shot down many times, learn to use the damn search feature rather then debated a feature which has been turned down MANY MANY times.

[HoneyFox]

bad idea, been shot down many times, learn to use the damn search feature rather then debated a feature which has been turned down MANY MANY times.

MANY times?
alright. let it be only a dream...

[manored]

I had a dream where all the viruses in my computer were busted and today all viruses in my computer were busted :)

[MadRat]

To do a seamless world in spring a few obstacles would need to be overcome.

1. Minimap - the look of the minimap would have to be dynamically rendered with a centralistic point of view relative to the unit selected. Probably a lot easier done than people would think since it is just a simple shift of the pixels.

2. Pathfinding - issues with the map wrapping back around to itself would have to be solved. Perhaps pathfinding could be done in sectors and all pathfinding just be figured sector to sector.

3. Line of sight in 1st person view - the map would need to be rendered in curved space. A seamless cylinder would be straightforward, however, something spherical would not. I cannot imagine this would be all that easy to accomplish.

4. Line of sight range perspective - theoretically when looking across curved space an object's location will be distorted much different than on a flat plane. The enemy unit, when computed in curved space, might be within your 1000 pixel range when figuring curved space but that same unit is completely out of sight across a flat plane.

Creating cylindrical worlds probably aren't beyond reach. Spherical worlds just seem out of the realm of possibility.

[Das Bruce]

No, just NO.

[jcnossen]

I did a little test with a spherical world with 6 heightmaps mapped similar to cubemapping, and there seem to be even more problems:

5, creating 6 heightmaps that connect on the edges is a pain, AFAIK no heightmap generator (terragen, l3dt) can actually do that.

6, collision detection with the heightmap will be heavier on CPU, because you can't compare Y position, you have to compare distance from the center of the sphere.

[manored]

Well since it seens that spherical worlds are at least years away from us, how about 3d maps? I mean, like, maps that allow stuff such as having a patch of land "float", and bridges that you can pass under, etc...

Wouldnt be fun to see all terrain units such as spiders stick to "ceilings" and bombard enemy units bellow? :)

[imbaczek]

this needs (I think) a total engine rewrite. (actually, this would need merging current spring code with the quake3 engine, or something similar.)

[smoth]

Wouldnt be fun to see all terrain units such as spiders stick to "ceilings" and bombard enemy units bellow? :)

this sort of thing is not good in an rts. Units will be obscured and you will spend too much time fighting with camera angles to see shit.

As much as I would like to see bridges over water, having shit on the bottom of the bridge would be baaaaaaad.