Idea: Night vision shorter then day vision?

This.
Robots are robots, autonomous war machines designed to destroy other such automatons. Light would be handy for computer-vision recognition, but that's a human ideal forced onto a robotic concept. It's easier to use infrared to detect non-identified objects against a static background such as a planet.

If anything, light would be an advantage for the player's moral.

You can't imagine a machine to calculate faster and more accurately than a human.

You genuinely cannot think of something that can perform mathematical calculations at a more consistent, accurate, with larger numbers, at a faster speed than a human.

I suggest we stop arguing with him.

It is going quite aggressive here.

Math:
A Planet in PA is about… damn I have forgotten… ok r (meters in Radius). For this calculation we assume it is a perfect sphere.
A Unit is h (meters in Height) and the Unit it wants to spot is also as tall.
We are looking for the direct distance l (in useful SI-Unit meters) ,from head to head, that an unit is capable of seeing another unit.




So let us say the planet is about r=500m and h=5m then l=141.8m
That is not that much how about r=500m and h=20m then l=286m
Still not much but 500m is about a third or so of the slider so let’s go bigger
r=2000 h=20m then l=567m

Conclusion: The Robots can not see so far anyway. At least in a Newtonian physics based Model (No gravitation on light like a black hole).

My personal hope/wish for the game is that they can see as far as that. No silly circles from some silly stats. Just the real line of sight.
Weapons are ok to be limited by a circle.

Edit: Spelling spelling :cry:
Edit2: Wrong formula in first picture corrected that hopefully better now (sorry )

Hopefully NEITHER one will be the case.
Vision based on circles? Thats broken by design, since all units (despite their size or visibility) become visible at the same time, enforcing artificial (and just as broken) hard counter intelligence abilities like "stealth" which makes totally no sense at all. Combine that with tiered intelligence (omni, vision, radar), and things just get worse.
But LoS is just as bad in this case. You need a lot of obstacles and LoS blockers in order to make the terrain appealing from a tactical point of view, but you must be able to see enemies before you can shoot at them. Tying intelligence to LoS is just as broken as tying it to fixed ranges. Especially in PA where LoS is not comprehensible at all due to the curvature of the planet. And you can't just do LoS tests between all units, it's already expensive enough when you do that for shots. It also leads to strange situations where you wouldn't be able to see units which are at the top of a platoon, although both you and the enemy are equipped with indirect fire weapons and you are just standing right in front of it. That places so many limitations upon map design...

Btw.: Your formula is wrong, at least the second one, there is a factor of 2 missing. Besides, 20 meters would be rather big for a bot.

Thank you, yes I was wrong. I corrected it now.

I did a little calculation based on the previous Model on how much of the surface area an Unit could see the same unit in the distance. See calculation:

the Unit is visible at 4% with a small planet and 1% with a larger one (maybe to large?)
Personally that is totally fine with me.
About the heaviness of the problem: To calculate line of sight would be bad yes. But subdividing the Planet would speed that up, this is done in other games too (like Warcraft3)
About the Hill problem: Yes that would be somewhat bad. But radar would be the solution here. Both things are possible (LoS/Circle) it comes down to what is more fun to play

You say you would not like a circle too. What would you then do? Is there something more to consider?

viewtopic.php?p=687594#p687594
viewtopic.php?f=61&t=45594
viewtopic.php?p=721049#p721049

In short: The visibility of the unit is calculated from the combination of:

• Combined sensor resolution (radar, vision, all combined in one abstract value, there is no need to treat them differently)
• Distance
• Size/visibility of the target

No LoS is used (and the intelligence is thereby unaffected by obstacles), but it still acts a lot smarter than hard circles, especially when you compare large to small units. First ones will never be able to sneak under the radar, later ones will.

Also every unit has some type of "radar" and is thereby capable of detecting large units from far away, not as far as real scouts or radar towers, but still from a fair distance. Scout also becomes a legit role in REGULAR army composition (you don't want to miss a barely visible ambush) and you no longer need hard intel counters like Stealth which are always either under- or overpowered, but often enough both at the same time, depending on which unit has such ability...

It's also not that bad performance wise, that theoretically "unlimited radar" each unit has can be calculated very efficient, not in O(n²) (as one might suspect), but actually in O(n log(n)) (absolute worstcase) by using octtrees which means that it is just as costly as the approach with the fixed circles.