These forums have been archived and are now read-only.

The new forums are live and can be found at https://forums.eveonline.com/

Missions & Complexes

 
  • Topic is locked indefinitely.
Previous page12
 

Recent 'issue'

First post
Author
Previous Topic Next Topic
Antony E Stark
The Tory Party
#21 - 2013-07-30 13:51:29 UTC
Alicia Aishai wrote:

A SIMPLE and VERY HELPFUL way to alleviate this issue would be to have a way to display the invisible wall on the screen. It's fine that there are technical limitations to change the system but at least, please let us see the invisible walls (add a navigation toggle) so that we don't get stuck in the middle of no where without clear understand in which direction we should go.


This would work. I know toggling on "show collision boundaries" would look quite yucky graphically, but it would be handy for someone to switch on, check, switch off.

Or

Reduce the size of the collision sphere and allow some of the graphics to be "clipped" as we pass through them. Again, not exactly realistic and a bit yucky but would prevent the issue. It would also require less client/server calls (smaller spheres, less likely hits, less calls) and reduce GM workload. Please have my desk ready in Reykjavik for my arrival tomorrow Cool
Malango
A.D.I
#22 - 2013-07-30 14:09:51 UTC
Alicia Aishai wrote:


I remember commenting on the exact same topic recently.

A SIMPLE and VERY HELPFUL way to alleviate this issue would be to have a way to display the invisible wall on the screen. It's fine that there are technical limitations to change the system but at least, please let us see the invisible walls (add a navigation toggle) so that we don't get stuck in the middle of no where without clear understand in which direction we should go.



This is exactly what we need, I would simply have it turned on when I mission in smaller active propulsion ships and leave it off when I mission in my Rattlesnake and so on. Would have saved my nighthawk.
Vincent Athena
Photosynth
#23 - 2013-07-30 15:41:22 UTC
Sabriz Adoudel wrote:
Vincent Athena wrote:
That said, some collision spheres are really bad. A small object inside a huge sphere, where a small sphere would be quite adequate. Or the other way: objects you can fly through because the spheres are not in the right places or big enough.

Also I wounder how well the math for collision detection is actually done. If the distance between two objects is small enough then the collision takes place. The distance is found by taking the square root of the sum of the squares of the difference in the three coordinates (X, Y, and Z) of the two objects. But for most items on the grid I will be far too far away for a collision to happen, so the test for collision should throw out as many items as possible as fast as possible. One way to do this is to first see if just the X distance is too large for there to be a collision. That is a faster test than doing all the differences, squares and square roots, and for most objects all that is needed to rule them out.

(Actually you can eliminate the need for doing a square root by comparing the difference in the sum of the squares of the Delta X, Y, and Z distances to the square of the critical distance).

Or another way: Say a complex object consists of many collisions spheres. Instead of checking my ship against them all each tick, surround the entire object with a big "area of influence" sphere. First see if I am in that area of influence. If I am, then do all the collision checks. If not, do none of them.



The efficient way to check "is point (x1, y1, z1) within radius r of point (x2, y2, z2)" is as follows:

Step 1: Compute delta(x), delta(y) and delta(z)
Step 2: Are any of those greater than r? If so, report 'no collision'
Step 3: Compute the square of the distance (dx^2 + dy^2 + dz^2) and compare to r^2. If the distance squared is less than r^2 or equal, report 'collision', else report 'no collision'.

Note that when talking about two spherical objects, the radius 'r' is just the sum of the radii of the two spheres. Step 1-2 eliminates all long distance potential collisions without further computation.

That said, because of the overview, it appears that the distance between each player ship and each object in space is constantly computed anyway.

A little faster:

Step 1: Compute delta(x)
Step 2: See if it is greater then r. If so we are done and can report no collision.
Step 3: Compute delta(y)
Step 4: See if it is greater than r. If so, report no collision
Step 5: Compute delta(z)
Step 6: See if it is greater than r. If so, report no collision
Step 7: Compute the square of the distance (dx^2 + dy^2 + dz^2) and compare to r^2. If the distance squared is less than r^2 or equal, report 'collision', else report 'no collision'.

One note: programs evaluate IF statements in their entirety, even if that is not needed. That is

IF (condition A) OR (condition B) THEN...

Even if condition A is false the program will still evaluate condition B and do the OR operation, even though it is not needed to determine the entire statement is false. If you want to prevent the program from doing unneeded work you got to split up the if statement. And when you do you want to make the condition that fails most often be the first one to be evaluated. For example in Eve we players tend to stay in the plane of the ecliptic. So you would not want to test delta(z) first.

Know a Frozen fan? Check this out

Frozen fanfiction
dexington
Caldari Provisions
Caldari State
#24 - 2013-07-30 17:00:58 UTC  |  Edited by: dexington
Vincent Athena wrote:


One note: programs evaluate IF statements in their entirety, even if that is not needed. That is

IF (condition A) OR (condition B) THEN...

Even if condition A is false the program will still evaluate condition B and do the OR operation, even though it is not needed to determine the entire statement is false.


The | operator will evaluate both condition a and b, the || operator will only evaluate b if a i false, integer bitwise OR vs. conditional OR.

I'm a relatively respectable citizen. Multiple felon perhaps, but certainly not dangerous.
Zor'katar
Matari Recreation
#25 - 2013-07-30 17:24:24 UTC
Vincent Athena wrote:
One note: programs evaluate IF statements in their entirety, even if that is not needed.

Depends on the programming language. Many (including Python, which I understand is used for at least some of Eve) support short circuit evaluation; if the first statement of an AND is False, or the first statement of an OR is True, it won't bother evaluating the second half.

(Edit: Sorry, way off topic. I should know better...)
Vincent Athena
Photosynth
#26 - 2013-07-30 21:56:36 UTC
Zor'katar wrote:
Vincent Athena wrote:
One note: programs evaluate IF statements in their entirety, even if that is not needed.

Depends on the programming language. Many (including Python, which I understand is used for at least some of Eve) support short circuit evaluation; if the first statement of an AND is False, or the first statement of an OR is True, it won't bother evaluating the second half.

(Edit: Sorry, way off topic. I should know better...)

OK, did not know that. But even so, doing it my way saves the time evaluating delta(y) if delta(x) is big enough to insure there is no collision.

Eventually some Dev may pop in and mention a ship's "destiny tunnel"., the volume of space to be swept out by your ship's collision bubble given its future motion.

Know a Frozen fan? Check this out

Frozen fanfiction
Substantia Nigra
Polaris Rising
Goonswarm Federation
#27 - 2013-07-30 22:36:39 UTC
Zhilia Mann wrote:

Also, nice to see you back in this forum. Pretty sure I hadn't mentioned that yet.


TYVM Zhilia. Kind words are always welcome … and all too rare in these forums / fora.

I have been reading this forum section from time to time but not responding a lot, given my main interest of late being in nullsec PvP … and the warfare & tactics thread. My recent re-involvement has been largely prompted by some experiences during our current war against TEST-and-the-rest-of-eve. I’ve had a bundle of people, friend and foe, ping me in local and thank me for my ‘helpful’ forum postings. That really surprised me, since I’d assumed that most of what I’d written was of little use to others. Having enemy PvP-dudes single me out, usually after I make propaganda postings in local, and comment gave me a real boost to try and continue being helpful 

Given that Fountain has now fallen, all over bar the shouting (and mopping up strat work), I’m readying to move on my ‘next thing’ in eve … whatever that may be. Part of my prep is to re-establish a secure homebase for my team … hence the mission running to get faction standings back up over 8 so I can launch some more hisec POSes.

I am not sure of what is next for me. I’m currently looking at options of:
- High-end w-space PvP. I’ve worked C5 PvE solo, but there’s so much more in C5 / C6 systems that I have not explored … especially cap-escalation PvE and dedicated w-space PvP. I’ve been talking with a group, with decent TZ overlap, who do this stuff.
- ‘Pirate’ group PvP. I’ve long been intrigued by Tuskers’ approach to the universe. I do not, however, have the solo PvP sort of background those types of groups usually want.
- Some more solo w-space PvP. Not many kills-per-hour (unlike war against TEST) but I did really enjoy trying to kill things in my little bomber in w-space. For me the planning and preparation is half the fun, so I think I might go back there for a while once I re-establish my industry base.
- W-space PvE, to feel rich again. I’ve been pretty well broke my whole time in nullsec … my own fault for eschewing the PvE opportunities during peacetime and devoting my time to PvP. Possibly a few months in a C3 to get the wallet glowing black not red again.

Lol, anyways, that is way off this topic … which appears to have degenerated into a computer coding fest anyways 
Sorry OP, I wasn’t really planning to hijack, and my earlier comments (^^) stand.

I guess I am almost a 'vet' by now. Hopefully not too bitter and managing to help more than I hinder. I build and sell many things, including large collections of bookmarks.
seth Hendar
I love you miners
#28 - 2013-07-31 09:49:56 UTC  |  Edited by: seth Hendar
Alicia Aishai wrote:
GM Bunyip wrote:
Maybe I can shed a little further light here:

The art models that go with a given item/object in space don't actually depict the areas where the ship will collide with that object. This is defined by an invisible sphere placed around the object - the collision sphere, we call it. My thinking was always that a sphere is used because it creates less client-server calls (relative to a box?) but you'd want someone with technical knowledge to confirm.

One thing I do know, is that the more collision spheres you add to an object, the greater the technical cost in terms of server load, and client-server calls. For this reason the "fidelity" (or accuracy) of the spheres is sometimes quite a ways off and quite simple. These actually used to be more accurate, but then we ran some numbers, reduced the fidelity, and noticed a significant performance improvement (a.k.a. part of the "Need for Speed" initiative, of some years ago). When you think about how many sites are active at a given time, how many objects are in them, and how many interactions there are between those collision spheres and players, this makes some sense - it would add up over thousands of sites.

So, yup, that's why there are invisible walls in odd places sometimes. It's not so much a bug as it is an unfortunate byproduct of technical limitations.


I remember commenting on the exact same topic recently.

A SIMPLE and VERY HELPFUL way to alleviate this issue would be to have a way to display the invisible wall on the screen. It's fine that there are technical limitations to change the system but at least, please let us see the invisible walls (add a navigation toggle) so that we don't get stuck in the middle of no where without clear understand in which direction we should go.


would never happen, then you would actually see how bad are some of them

i remember that a year or so ago, the sphere were way more accurate, now you get bumped all over the place, in some asteroid belts, it' is so bad that you actually have to log off to get out of those sphere, even if visually there are only a few little roids (sometimes, only ewarp is able to get you out).

and i you happen to loose a ship to this BUG (cause yes, it IS a bug), the CCP's answer is:


Quote:
we can see you were stuck, with no possibility to move because of the environment, but since the actual fight show nothing abnormal, we cannot refund your {ship}


???? Shocked

cause yes, not being able to move has NO IMPACT AT ALL on a fight right?
Aurette
962348 Corporation
#29 - 2013-08-01 07:32:19 UTC  |  Edited by: Aurette
Substantia Nigra wrote:

My advice, having just been refamiliarised with this issue, is:
- Small combat drones … always, always, always. When you messup and your shields are melting is usually the (Murphy’s Law) time that you discover there is one more frig on the field, and it’s a point/scram frig to boot, and it is now too close or too small for your main weapons to do significant damage to. Most any drone will kill them, warriors just get there faster and start applying damage faster, and do not take up much drone bay space.


Yes good advice. Warrior I or Warrior II, anything else is just junk. About all I want a drone for is that pesky frigate that is under my guns. The warrior's speed makes it the best drone in my opinion, it will save ships when you really need to bug out.
BrutalButFair
Fleet of the Damned
#30 - 2013-08-01 09:10:19 UTC
GM Bunyip wrote:
Maybe I can shed a little further light here:

The art models that go with a given item/object in space don't actually depict the areas where the ship will collide with that object. This is defined by an invisible sphere placed around the object - the collision sphere, we call it. My thinking was always that a sphere is used because it creates less client-server calls (relative to a box?) but you'd want someone with technical knowledge to confirm.

One thing I do know, is that the more collision spheres you add to an object, the greater the technical cost in terms of server load, and client-server calls. For this reason the "fidelity" (or accuracy) of the spheres is sometimes quite a ways off and quite simple. These actually used to be more accurate, but then we ran some numbers, reduced the fidelity, and noticed a significant performance improvement (a.k.a. part of the "Need for Speed" initiative, of some years ago). When you think about how many sites are active at a given time, how many objects are in them, and how many interactions there are between those collision spheres and players, this makes some sense - it would add up over thousands of sites.

So, yup, that's why there are invisible walls in odd places sometimes. It's not so much a bug as it is an unfortunate byproduct of technical limitations.


So the guy should be reimbursed because he didn't know this. I think noone knows this and should be made obvious somehow.
Malango
A.D.I
#31 - 2013-08-01 15:10:16 UTC
seth Hendar wrote:
Alicia Aishai wrote:
[quote=GM Bunyip]

and i you happen to loose a ship to this BUG (cause yes, it IS a bug), the CCP's answer is:


[quote]we can see you were stuck, with no possibility to move because of the environment, but since the actual fight show nothing abnormal, we cannot refund your {ship}


???? Shocked

cause yes, not being able to move has NO IMPACT AT ALL on a fight right?


Yup it's a bug in my opinion but I got no ship back. Didn't play for 2-3 weeks after. As I was rather upset that visually I was completely in space with nothing near me. And most of us had no idea that spheres were so badly placed.
Zoltan Lazar
#32 - 2013-08-01 22:20:13 UTC
If you've watched star trek or read ender's game or almost any other space combat sci fi, the first lesson is to think in 3d. The clutter is often on a plane, just because you warp in on that plane doesn't mean you have to stay on that plane.
Previous page12