Radar LOS

I was wondering if anyone had given thought to the concept of line-of-sight for radar? In other words, terrain affecting radar range and coverage. This is something I saw implemented in some of the Spring games out there, but I'm still wondering at the merits of such a system.

With regards to PA, it adds some interesting gameplay concepts:
First, radar towers would be a lot more valuable and it would be in the player's interest to set up a strong radar network if they want reliable coverage. Now enemy attacks on a radar station are more critical, and since the player can't simply rely on a few radar stations surrounded by a batch of defense. This also means that there's a lot more opportunity for an enemy to land on a planet undetected if they are able to take advantage of a radar blind spot, or take down a radar tower before landing.
It also adds to the value of scouting. Since radar coverage isn't perfect, occasional flybys with a scouting plane would be necessary. This adds to the value of such scouts when used defensively.
As for airplane radar and orbital radar, these units wouldn't be affected by terrain, so they have added value, although plane radar would have a lot smaller range, and orbital radar would be very vulnerable to an enemy
All of this culminates into a more dynamic intel game.

However, there are disadvantages to this concept:
Now there's a bit more complexity with regards to base management, as the player has to think about maintaining the radar grid. They have to manage radar on multiple layers, those being land stations, aircraft, orbitals, etc.
It also means the base is more vulnerable to air attacks, especially in the early game.
There's also the fact that there will be a lot more attacks on radar installations, and players may end up being bogged down by defending their radar grid.

So while this feature has some nice gameplay connotations, it could end up being more of a burden as well, especially on a planetary scale.
I'd be interested to hear everyone's thoughts on this.

This could be a good idea as it actually makes more sense (Especially in sonar), and could give reasoning to put radar in more exposed positions.

But only as long as people who go to build one understand why a mountain blocks their radar.

Sorry to interrupt you both, but that has already be considered and was shot down for a very good reason:

LoS is BROKEN on a sphere. The visual range would be terrible low, to cover a large planet (thinkig about mere 80km diameter!) with radar towers would require over 6.000(!!!) radar towers to cover it, given an height of 200 meters for each tower and perfectly flat terrain. Even when placing all radar towers on mountains which an height of >500 meters, you would still about 1.000 towers. With planes still about 1.000 planes since you can't let planes fligh higher then 3-4 times the height of the larges unit or you won't be able to comprehend the position of the plane relative to the surface.

Thats math. It can't be solved. LoS is broken by design when talking about spherical maps.

Want the actual calculation? Use the search function or ask google how to calculate line of sight on a sphere.

Not even if LoS were a sphere? And besides radar is best for aircraft anyway, sight confirmation would be best for ground targets.

And if Sight can be bended to a planets surface, surly radar can also be 'bent'.

Sight works partly with LoS since the visual radius is much smaller then the radius of radar. Therefor the surface of a sphere is almost flat within such a small area. "Bending" as you described it, would be a terrible complex process and would just break if the celestial body is not exactly spherical, like any type of meteor or planet without liquid core.

Although it's still open if there will be a disctinction between visual and radar and if LoS will be used at all, except for direct fire units.

Looks like you've never heard of over-the-horizon radar.

That's taking it a bit too far in the realism direction. If LOS is broken for a sphere, how will it be accomplished for standard fog of war? I'm fairly certain that will still be in the game, and it's not too hard to extend that to radar LOS. And considering how "realistic" the radar is in TA/Supcom, I don't think it would be too much of a stretch to extend that to PA. After all, even if PA had the standard, circular radar coverage, it would still maintain that unrealistic issue.

I think you're considering true LOS, rather than LOS restricted primarily by terrain, and that's not at all what I'm suggesting.

In any case, I'd be interested in seeing the previous discussion on this, since I couldn't find any previous threads about it. Do you recall what the thread was named?

It was one of the first topics on radar. Consens was, that LoS makes things not only unecesarry complex, but is also kind of broken on anything except flat maps. You can fix it, but that comes with some tremendous overhead in calculations. And still stays completly broken on more complex shaped objects like meteors.

As for the fog of war, many games ignore LoS for fog of war, it's just circular (or spherical) vision and it causes no problems. Especially games which have to deal with rough terrain and don't require LoS for tactical decisions on micro level.

Just because TA had LoS for visual, does NOT mean that it is the optimum solution or that it will make it into PA at all.

Don't forget that in PA, maps for planets are implemented as flat surfaces, then projected on a sphere. (At least, that's what the devs originally stated. I don't think they've changed it though.) This means that LOS can be calculated much in the same way it's been calculated for TA and Supcom. Extending it to radar LOS calculations would be fairly trivial.

Line of sight also gets fun when you start including orbital units.

If you have something in geosync, you have los on a huge part of the map, which seems nice, until you consider the ranges involved.

As example for illustration: GEO on earth is 36k km, we'll round down to Earth being 6k km diameter. Should a satellite 30k km (5x the radius of the planet) have the resolution (radar, visual, whatevs) to see a say 10m unit? Note: spy satellites are at 300 km, not 30k.

If your response is to put radar satellites in low orbits, then you actually get close to the same problem, where you need large numbers to cover the planet effectively. Iridium uses 66 satellites to cover the whole planet, but that is for radio. To do that with radar you still need some super strong radars.

As far as i know, there was no LoS in SupCom. Only direct fire units used LoS like mechanics for their projectiles, but thats all. To many maps had cliffs or mountains which have been placed for tactical reasons (prevent direct fire units from making a straight approach) but would have blocked way to much from visual.
Also LoS would have caused yet another problem: It requires much micro managment to use it properly or to defend against it. It worked in TA because of the fixed zoom and the fact that you had to micro manage a lot, but it wouldn't have worked in SupCom because you just had to many units to make use of LoS. And given that PA is supposed to have even larger armies, the use of LoS becomes even smaller.

And as for the maps projected to spheres:
Yes, that was the original idea. But it had one huge drawback, the poles of each planet would have been inaccesible. They haved switched to a more sophisicated approach wherby each planet or celestial body starts as a cube which is then first morphed into a sphere and then even further until the desired shape is reached.

PS: Here is the post with the calculations i was talking about: viewtopic.php?p=542181#p542181

Problems with LoS radar are discussed (among other things) here.

At first I was for LoS radar, as it works well in Zero-K, but it would just not work on a spherical body. We could use the flat map to calculate LoS, but it would be kind of strange, as, for the player it would be blocked by a mountain but not by the enormous volume of the planet curvature.

And LoS is also a nice way to make visual and radar more different. LoS don't need to have a much smaller effective range than radar, as it would be blocked by terrain and curvature anyway. Satellites would have great visual coverage, which would help differentiate them from the other layers, but may not be able to see the smallest or more discrete units.
It would also bring more variation between map types. In asteroids, for example, visual would be impaired, but radar would be king.

Remember, we're doing this with fancy, futuristic robot tech. If there's a way it can be done, they would have done it. The concern is not about the realism and real world comparisons. The concern is how this system would affect gameplay, and that's the discussion I'm trying to put forth here.

As for engine limitations, I'm fairly certain it wouldn't be too hard to implement. If it were, pathfinding, fog of war, collision detection, etc would all have challenges when implemented on a sphere. In any case, the feasibility from a technical standpoint is for the devs to decide. Whether it's a good idea with regards to gameplay is something we can discuss.

Nah, they ditched that idea; I remember multiple posts indicating another solution was being chosen and finally on the irc channel neutrino confirmed that the terrain would be a deformed cube approach.

However, "flat" LoS can still be implemented; because you are simply deforming a cube (or what is essentially 6 square maps), they should be able to apply a traditional LoS to each side and have it wrap onto other sides.

This means that overall we get the advantages of flat maps, and the advantages of a true 3D planet!

Actually it's going to be much more complex than a cube (although that's where it kind of starts). The main issue was I wanted to be able to support more arbitrary topology. Using a mapping from a flat map to a torus gives you nice properties for scrolling and such but you are stuck with that topology which has some drawbacks. Since I want this game engine to last a long time and be able to extend in different directions we are just going with a real 3d mesh that happens to be a sphere at the moment. So everyone asking for dyson spheres, ringworlds, giant ships or whatever could potentially be supported eventually (not in the first version mind you, concentrating on spheres atm).

This is amazing! Glorious even.

It's like I saw the light, and it blinded me and brought tears to my eyes!
That's one of the best things I've heard about the engine. Arbitrary geometry will make for some absolutely amazing and crazy mods down the line!

As far as the Radar goes, not sure what the best implementation would be. That would probably need some playtesting to see what "felt" right while using it.

Also keep in mind that although the geometry will be unrestricted in some ways we'll still be building in some amount of reliance on the current model. It's more about keeping our options open going forward.

Yeah, that's cool. Just what you said hit me in just the right/wrong area

Kinda like Homeworld did when I first played it. "What, I can warp in under a spaceship and lazors it from below? Why hasn't this been done more?"

Well as far as radar goes you could just mathematically ignore the curvature of the map and just do a quick check if there are mountains in between the detected unit and the radar tower. Maybe it's not super realistic but it might make the gameplay complex enough to be interesting but simple enough for people to work with on the fly.

OK thread derailed, this is way more interesting.

OK yeah I realize this is for later on down the road, and right now it's probably a good idea to get spheres worked out before going nuts.

But I am looking forward to someone fiddling with the code to make a mod that supported M.C. Escher maps.