Pages: 1 [2] :: one page |
|
Author |
Thread Statistics | Show CCP posts - 18 post(s) |
Aeon Amadi
A.N.O.N.Y.M.O.U.S.
3358
|
Posted - 2013.10.18 07:04:00 -
[31] - Quote
LogicLoop, anyway you could pass it on to maybe have the RDVs avoid dropships? We had a PC today in which an invisible RDV took out an entire squad and I remembered I've been meaning to run that by you guys. |
steadyhand amarr
MoIden Heath PoIice Department EoN.
1591
|
Posted - 2013.10.18 10:25:00 -
[32] - Quote
CCP LogicLoop wrote:steadyhand amarr wrote:CCP LogicLoop wrote:steadyhand amarr wrote:A nav mesh. On maps that size it must be one hell of a spider's web u got their, explains why we can't drop inside sockets though. Here was me thinking u used dynamic waypoints :-D Well and just think of this, the sockets changing as well. Compiling a navmesh is time consuming because for each terrain it needs to load every possible socket combination and rotation and build the navmeshs for them then stitch them with the ones in terrain. This is a reason why the RDV some times freaks out (my technical term) on some sockets. I'm don't no the tech terms so I'm going armchair dev words lol. But would not collision boxes be better the rdv simply goes from a to b and if it hits one of the walls it paths around it. I understand I'm vastly over simplifying the problem but nav meshes only really work when u know what ur dealing with and due to sockets and diffrent map types it complicates the problem. Giving the A.I very basic navigation tricks rules instead seems like a better option for dust. But I'm not a dev sssooo.... :-P So navigation points / navmeshes typically take collision and solid walls into account. What is causing problems with our is when the compile has to "stitch" together the navmesh between a socket / component and the main terrain. It in the least technical terms freaks out at times. So this is why I mentioned its a topic being discussed. Valto Nyntus wrote:The idea about the launch pad becoming an installation is simple, you just call a vehicle with the console nearby and it sends it there.is this do able in dust now or at least the near future?(I'm not sure if it solves your problem, but I thought I'd say it) I don't know.
I hate phones limits how much I type lol
My issue with nav mesh is it does not react to changes events which is why u have stitches problem and rdv freak out. My thought was to simply allow the rdv to plot their own paths as they are spawned in rather than rely Oon predefined paths that could end up crashing into each other.
Will make a bigger post when I get home lol
|
Telleth
Molon Labe. RISE of LEGION
162
|
Posted - 2013.10.19 01:36:00 -
[33] - Quote
Was flying around close to the two large towers, right around M and N 14, and I hit an invisible wall/corner. Poof went my dropship. The tops of those towers don't line up with what is there, you can walk around on invisible-ness while the top of the tower sits quite a bit below you. Kinda like when you're on top of the MCC, the structure and visual don't line up. This was during an ambush on the bridge map if that makes a difference. |
|
CCP LogicLoop
C C P C C P Alliance
1125
|
Posted - 2013.10.21 01:29:00 -
[34] - Quote
steadyhand amarr wrote:CCP LogicLoop wrote:steadyhand amarr wrote:CCP LogicLoop wrote:steadyhand amarr wrote:A nav mesh. On maps that size it must be one hell of a spider's web u got their, explains why we can't drop inside sockets though. Here was me thinking u used dynamic waypoints :-D Well and just think of this, the sockets changing as well. Compiling a navmesh is time consuming because for each terrain it needs to load every possible socket combination and rotation and build the navmeshs for them then stitch them with the ones in terrain. This is a reason why the RDV some times freaks out (my technical term) on some sockets. I'm don't no the tech terms so I'm going armchair dev words lol. But would not collision boxes be better the rdv simply goes from a to b and if it hits one of the walls it paths around it. I understand I'm vastly over simplifying the problem but nav meshes only really work when u know what ur dealing with and due to sockets and diffrent map types it complicates the problem. Giving the A.I very basic navigation tricks rules instead seems like a better option for dust. But I'm not a dev sssooo.... :-P So navigation points / navmeshes typically take collision and solid walls into account. What is causing problems with our is when the compile has to "stitch" together the navmesh between a socket / component and the main terrain. It in the least technical terms freaks out at times. So this is why I mentioned its a topic being discussed. Valto Nyntus wrote:The idea about the launch pad becoming an installation is simple, you just call a vehicle with the console nearby and it sends it there.is this do able in dust now or at least the near future?(I'm not sure if it solves your problem, but I thought I'd say it) I don't know. I hate phones limits how much I type lol My issue with nav mesh is it does not react to changes events which is why u have stitches problem and rdv freak out. My thought was to simply allow the rdv to plot their own paths as they are spawned in rather than rely Oon predefined paths that could end up crashing into each other. Will make a bigger post when I get home lol
Technically it does plot its own path. It uses the navmesh to know "where its safe to plot a path". The problem is the navmesh doesn't always create the best options for the RDV because of the stitching. The navmesh looks like a giant web of interconnected points. It's not just lines that we draw. Its job is to tell the RDV where it "can go" and "can not go". The RDV picks its options from all those possible path lines. |
|
Aeon Amadi
A.N.O.N.Y.M.O.U.S.
3400
|
Posted - 2013.10.21 03:23:00 -
[35] - Quote
CCP LogicLoop wrote:
Technically it does plot its own path. It uses the navmesh to know "where its safe to plot a path". The problem is the navmesh doesn't always create the best options for the RDV because of the stitching. The navmesh looks like a giant web of interconnected points. It's not just lines that we draw. Its job is to tell the RDV where it "can go" and "can not go". The RDV picks its options from all those possible path lines.
Given your experience in level design, would you say that it's operating less than, equal to or greater than your expectations? Honestly, considering the complexity of it, from a design standpoint it sounds amazing that it works as well as it does. |
Aero Yassavi
PIE Inc. Praetoria Imperialis Excubitoris
2940
|
Posted - 2013.10.21 05:29:00 -
[36] - Quote
RDV navigation has certainly progressed a long way and is quite impressive how far you have gotten it, but of course it still needs a lot of work. I agree with Aeon, when considering the complexity of it I'm really amazed it works as well as it does. |
crazy space 1
Unkn0wn Killers
1887
|
Posted - 2013.10.21 05:39:00 -
[37] - Quote
um... excuse me.. why not just program RDVs to line up in literal line if another RDV is dropper nearby. Have them wait in line and drop off one at a time. don't mess with the nav net just make them smarter |
steadyhand amarr
MoIden Heath PoIice Department EoN.
1612
|
Posted - 2013.10.21 07:25:00 -
[38] - Quote
Heh like rl holding patters. Thanks for the respones Logic. I guess the big issue is the working on the algrothium to find the best flight path as at the moment rdvs love dropships :-) did my homework and seen navmeshs are basically industry standard.
|
|
CCP LogicLoop
C C P C C P Alliance
1126
|
Posted - 2013.10.22 00:36:00 -
[39] - Quote
Aeon Amadi wrote:CCP LogicLoop wrote:
Technically it does plot its own path. It uses the navmesh to know "where its safe to plot a path". The problem is the navmesh doesn't always create the best options for the RDV because of the stitching. The navmesh looks like a giant web of interconnected points. It's not just lines that we draw. Its job is to tell the RDV where it "can go" and "can not go". The RDV picks its options from all those possible path lines.
Given your experience in level design, would you say that it's operating less than, equal to or greater than your expectations? Honestly, considering the complexity of it, from a design standpoint it sounds amazing that it works as well as it does.
Less than of course. It really should not be bumping into buildings, or dropping vehicles on top of buildings when it should go on the ground. I believe if we stick with the current method the level designers need to be given more control of certain circumstances. I have suggested in the past that we can use volumes that tell the navmesh not to be built in certain places so the RDV just completely ignores the area (like an entire building, or a pit, things like that).
To answer another question I saw, one suggestion that was brought up ( I don't believe any decisions have been made, this was just a suggestion from a Game Designer) is that they just spawn right above you or nearby and drop it straight down rather than flying some distance. However the argument for that was it takes away from the immersion.
Either way, we are clearly aware of some of these issues and are discussing how to best resolve them. |
|
Aeon Amadi
A.N.O.N.Y.M.O.U.S.
3426
|
Posted - 2013.10.22 02:32:00 -
[40] - Quote
CCP LogicLoop wrote:Aeon Amadi wrote:CCP LogicLoop wrote:
Technically it does plot its own path. It uses the navmesh to know "where its safe to plot a path". The problem is the navmesh doesn't always create the best options for the RDV because of the stitching. The navmesh looks like a giant web of interconnected points. It's not just lines that we draw. Its job is to tell the RDV where it "can go" and "can not go". The RDV picks its options from all those possible path lines.
Given your experience in level design, would you say that it's operating less than, equal to or greater than your expectations? Honestly, considering the complexity of it, from a design standpoint it sounds amazing that it works as well as it does. Less than of course. It really should not be bumping into buildings, or dropping vehicles on top of buildings when it should go on the ground. I believe if we stick with the current method the level designers need to be given more control of certain circumstances. I have suggested in the past that we can use volumes that tell the navmesh not to be built in certain places so the RDV just completely ignores the area (like an entire building, or a pit, things like that). To answer another question I saw, one suggestion that was brought up ( I don't believe any decisions have been made, this was just a suggestion from a Game Designer) is that they just spawn right above you or nearby and drop it straight down rather than flying some distance. However the argument for that was it takes away from the immersion. Either way, we are clearly aware of some of these issues and are discussing how to best resolve them.
I assume that these complications are the reason(s) that recalled vehicles aren't transported out of the area by an RDV and simply fade away, or is that simply being saved for a later date as necessary to get other priorities knocked off the board? |
|
crazy space 1
Unkn0wn Killers
1893
|
Posted - 2013.10.22 06:30:00 -
[41] - Quote
Another question for feedback. Right now we call them in but don't really know where they will land to be honest. We have zero control over it. Would it be possible to throw a flare and have the RDV drop off at that exact location? Or at least close nearby to it?
So could space out the drop-off spots to be more careful, at least as a temp ban-aid fix.
I suppose that doesn't fit into the current system of picking it from a menu... you could have the flare appear in your hand once you choose what you want called in? |
|
CCP LogicLoop
C C P C C P Alliance
1130
|
Posted - 2013.10.23 00:25:00 -
[42] - Quote
crazy space 1 wrote:Another question for feedback. Right now we call them in but don't really know where they will land to be honest. We have zero control over it. Would it be possible to throw a flare and have the RDV drop off at that exact location? Or at least close nearby to it?
So could space out the drop-off spots to be more careful, at least as a temp ban-aid fix.
I suppose that doesn't fit into the current system of picking it from a menu... you could have the flare appear in your hand once you choose what you want called in?
That is basically the kind of idea that was brought up here. |
|
Evicer
THE HECATONCHIRES
25
|
Posted - 2013.10.23 02:06:00 -
[43] - Quote
CCP LogicLoop wrote:Soraya Xel wrote:CCP LogicLoop wrote:Research Facility (Large Socket)-Top of both sets of stairs under the main research building is just a dead end. (Funny thing this. Artists forgot to put fake doors up there, we also intended to block those stairs off with infamous crates) I was drastically fond of this being a sniper point. Please don't block it off. Most likely, the biggest change would be addition of some doors at the top so it doesn't look silly. This wouldn't be published until we push a content update again either. I dont see a problem with it currently.Its a good spot for uplinks as well.
|
|
CCP LogicLoop
C C P C C P Alliance
1130
|
Posted - 2013.10.23 02:36:00 -
[44] - Quote
Evicer wrote:CCP LogicLoop wrote:Soraya Xel wrote:CCP LogicLoop wrote:Research Facility (Large Socket)-Top of both sets of stairs under the main research building is just a dead end. (Funny thing this. Artists forgot to put fake doors up there, we also intended to block those stairs off with infamous crates) I was drastically fond of this being a sniper point. Please don't block it off. Most likely, the biggest change would be addition of some doors at the top so it doesn't look silly. This wouldn't be published until we push a content update again either. I dont see a problem with it currently.Its a good spot for uplinks as well.
I don't think it will be blocked off. It will probably just get the fake doors added at the top. However I don't even know if they will do that right now due to time constraints. |
|
Soraya Xel
New Eden's Most Wanted Top Men.
686
|
Posted - 2013.10.23 14:26:00 -
[45] - Quote
Yeah, I was surprised the turret on that small socket wasn't fixed last patch, but I figured it was cuz y'all were busy getting us a Caldari socket set ready. |
|
CCP LogicLoop
C C P C C P Alliance
1133
|
Posted - 2013.10.24 04:19:00 -
[46] - Quote
Soraya Xel wrote:Yeah, I was surprised the turret on that small socket wasn't fixed last patch, but I figured it was cuz y'all were busy getting us a Caldari socket set ready.
Well the turret fix better be in on next content release (1.7). Doors at top of research stairs, that one is a low priority, so it may not happen.
|
|
Sylwester Dziewiecki
Beyond Hypothetical Box
197
|
Posted - 2013.11.14 11:51:00 -
[47] - Quote
CCP LogicLoop wrote:I want to get a list of the known issues consolidated here. Please feel free to have me add to the list if it is relevant. This includes all the supporting sockets and the large outpost. We want actual bugs. Research Facility (Large Socket)-Top of both sets of stairs under the main research building is just a dead end. (Funny thing this. Artists forgot to put fake doors up there, we also intended to block those stairs off with infamous crates) -Road posts / bumpers are popping in too late. -Gate structure you can walk through the railing in a corner. (Fixed internally) On skirmish game mode null-cannon terminal 'A' on E5 have a very annoying spawn point with completely prevent hacking that terminal.. spawn point is 2m at front of terminal - everyone that is killed during the attack on point can spawn and kill everyone that is trying to hack terminal. This spawn point rewards lame gameplay.
|
|
|
|
Pages: 1 [2] :: one page |
First page | Previous page | Next page | Last page |