Forums » General

VO 1.8.92 - Asteroid occlusion of players on radar.

Sep 26, 2009 incarnate link
*** Vendetta 1.8.92
- Asteroids now occlude player-owned ships from your radar.
- 10 second logoff timer is now changed to 30 seconds for 45
seconds instead of 20 seconds when you receive a private


Ok, a few things about the occlusion:

The basic idea is that, if an asteroid is between you and another person, you won't see them on your radar. This has been requested since.. 2003, at least, and has wide ramifications for everything from piracy to avoiding-pirates to clandestine mining to future skirmish tactics and so on. A fairly big deal. However..

1) It only occludes players for now, not NPCs. It's too CPU-intensive, in the current implementation, so until we optimize it.. it's players only. We're working on optimizations, but I wanted to drop in this early implementation now, so we can get feedback while we're continuing to develop.

2) The occlusion is not hyper-accurate. It uses a raycast->sphere intersection test. Asteroids are generally not spherical, so there's a fairly large area where a player could be visible, but still be occluded by the "sphere" of the given asteroid. The larger the asteroid, the larger the disparity. You can choose to explain this as some sort of "area radar interference" or whatever you like (ie, being near asteroids occludes you). We are going to be trying to improve this, allowing us to define the sort of intersection test for a given object.. so for super-huge objects we can use the (slower, but accurate) OBB test, where for general-case asteroids we use spheres and capsules and things. The general (common) case for smaller asteroids will probably always be inaccurate like this, it's about the best we can do in the near term (without some hardcore, specialized optimization of the OBB test).

3) Occlusion is semi-pointless while the Players-In-Sector list still exists (with distance, even!). We know. Again, this is a first-generation test just to see how it works and get feedback. I've talked about removing the sector list elsewhere, and that is still on the menu, but I'm not rushing into it this second (there's some other stuff I'd like to do). So, if anyone is looking to complain about this, yes, thanks, drive through.. we'll get there.

4) If someone you have selected is occluded by an asteroid (or asteroid-sphere), then you will lose your selection. We have mixed feelings about this, but decided to do it for this first release. It makes perfect sense, but it may be excessively annoying for basic things like.. botting in an asteroid field (clearly, not until we add bots to occlusion, but you get the idea). Especially if it causes the disparity between the spheres and actual occlusion to be so obvious that it becomes.. still more irritating. In a dense field, like the ones coming in the Universe Redux (or, say, Latos M7).. I have a feeling you might lose lock constantly. But, I'm not really sure if that's a bad thing or not. Different is not necessarily bad, and it may make for some interesting tactical adjustments, especially with respect to missiles and the like. So, feedback welcome there.

5) It's a little bit slower, as I mentioned in the #1, but unless you go into a sector with 50 players, it shouldn't matter. In the meantime, we're cooking up ways of making it faster for all cases. Hopefully next week.

That's all for now.
Sep 26, 2009 Thanquol link
This should make Sundays Nation War... interesting. I'll have to see how it works out. Being stuck at work sucks :(
Sep 26, 2009 roguelazer link
Holy... shit...

We've been talking about this feature for, uh, 6 years or so?

I am floored.
Sep 26, 2009 mdaniel link
Wow, finally.
Sep 26, 2009 toshiro link
Sep 26, 2009 Whistler link
Is it Soon now?

That's very cool.

I can understand the ramifications of both losing the lock all the time (potentially annoying) or keeping it persistent (potentially negating the benefits of occlusion).

Perhaps there could be a way to maintain a lock that persists only for a short period of occlusion, and only within a very limited range. Or maybe the target could be automatically be re apprehended if they became targettable after a short amount of time.

Or perhaps there could be a way to visually identify a recently-locked target, Perhaps they could glow or something. This way I could visually pick my target out of other potential targets without having to rely on a lock.
Sep 26, 2009 roguelazer link
Occlusion. It's there, it works, and it's FREAKING awesome.

Holey roids are now the coolest things ever.
Sep 26, 2009 Kierky link
Very cool update, however, I haven't seen much opposition towards the concept of removing the players in sector list.. Just different ideas. I guess we would need to rate each of the ideas in a sperate thread, and would need to give community feedback on each of the ideas in question.
Sep 26, 2009 genka link
Without trying it out, I'm guessing I don't like this, but I suppose I'll wait until I have a chance to play before I decide for sure.
Sep 26, 2009 Kierky link
You don't like anything, genka.
Sep 26, 2009 FatStrat85 link
I'm excited to see how this changes gameplay. Nice update, devs. It reminds me of some of the radar ideas in the following thread that would also be really cool.
Sep 26, 2009 zak.wilson link
Regarding the loss of lock, I think you should not lose lock if you're within a certain distance, say 250m of your target. There's some realism to that (radar is more sensitive at a short distance), and it would make it a bit harder to get out of combat (running is already too easy).
Sep 26, 2009 toshiro link
How is it more realistic to maintain lock-on with a target that dips behind a roid than to lose it? If an object is occluded by another object, it matters not how close it is, unless the object were to magically bend the electromagnetic field.

Besides, realism is not the central argument here...
Sep 26, 2009 DivisionByZero link
Holy cow... and my computer is still borked!

Now, all you have to do is make a group share radar contacts and you'll have implemented some of most requested items in the game!
Sep 26, 2009 Professor Chaos link
I think realism is an eventual goal, toshiro, but you're right. Losing lock makes more sense. Maybe since running is too easy, it can be balanced by more powerful close-range weapons or something. At super-close range, if you keep losing lock, having to rely on visual target acquisition makes things very interesting. As with all things, this will be much more fun as the player base increases. I haven't seen an update in a long time that makes it more tempting to resubscribe!

@incarnate: Are you familiar with geodesy? Is it possible for non-spherical 'roids to have a server-side algorithm that calculates a geoid for each large asteroid and assigns a best-fit ellipsoid for the intersection? It still won't be perfect, but may be the best compromise between spheres and perfect intersection tests.

EDIT: Division, if group radar contacts were shared, that could be the most compelling incentive for group missions/battles yet. I can imagine a team with a medic, a radar specialist, and three to five fighter/interceptors for quite a few interesting combat purposes. Wow, I'm really excited about this, and I can't play right now!
Sep 26, 2009 Phaserlight link
Wow, what a giant leap forward, I still can't believe this until I log in. I bet players will have a much better reason to hang out in asteroid fields now. Are asteroids really best described by capsules and spheres? I think most fit the shape of an ellipsoid as Professor Chaos suggests. Either way, the less than perfect fit could be explained as radar echoing from the edges of asteroids.

Nice work, devs. This is yet another reason to love VO.
Sep 26, 2009 zak.wilson link
It's reasonable to assume that radar, given enough power can penetrate a little bit of rock. It's also reasonable to assume that computers might have some hope of triangulating the position of a previously known target based on echoes off other asteroids.

Realistically simulating that triangulation and short-range rock penetration would require that Guild Software spend a whole lot of money on faster servers. Keeping lock at a short distance would not.

Another option would be to have a function that determines the probability of losing radar contact and lock based on distance, how long the target has been occluded and the angle between the direction of the target and the nearest edge of the occluding object.
Sep 26, 2009 DivisionByZero link
Actually, with the way radar works, being near another large object is not unrealistic to lose ID.

All a pure radar system sends back is echoes off the environment. Computer hardware and software has to distinguish between echoes to actually give you information on what it is. When the target is close to a roid, it could well be a bumpy roid, or a ship, but if the radar system doesn't have the resolution to distinguish, it wouldn't know any better. Of course, this could be very close on the front side of a roid too, not just occluded situations. So, from a realism point of view, you could lose a lock just by having your target "lay down" on top of the roid, for instance, even if you're facing it.

(radar is a little more complicated than this, especially with differences in reflectivity, but that's something else that has to get sorted out by the radar system)

There are other signals that typically help identify targets for radar systems. Transponders and electronic noise/emissions all aid in the determination of a target. But a complete loss of lock is still within the bounds of realism, IMO.

[EDIT] Just read Zak.Wilson's post: Radar waves hardly penetrate anything. The fact that they can use radar and reflect it off of water droplets when you look at a weather forecast should tip you off on that. If a radar wave is penetrating a material, it's being absorbed and probably won't be coming back out.
Sep 26, 2009 missioncreek2 link
Awesome! Remove the sector list.

The lock loss problem can be solved with a new version of TargetLs that auto re-selects the current target when it reappears.
Sep 26, 2009 FatStrat85 link
Nice idea, missioncreek2! I'm sure drazed can get that working soon.