OpenJK General Topic

[mrwonko]

@@mrwonko: What about the Free Software Foundation? Can't they step in where Raven can't?

No, they can only help in cases where the creators made them the copyright holders. So if we have to use the mp3code stuff to enforce a GPL release, we'll need the original creators/copyright holders to do it, it appears.

[eezstreet]

katanamaru: It isn't hard at all to find, just do ctrl+f and search for "transition". Off the top of my head, in SP it assigns ps->saberMove in bg_pmove.cpp and coordinates the actual transition in wp_saber.cpp. There's also a giant saberMoves table which assigns the next transition after an attack, among other things.

 

mrwonko: k. We'd have to see how much of a diplomatic stance Raven takes. There's already a mod that is refusing to adhere to the GPL by not releasing source (it was funny because their response to someone telling them how the GPL works was "k, lets do a code exchange then. My source code for your source code")

[Szico VII]

Moved to coding and scripts>

[eezstreet]

I know this, but it just crashes in MP when Ive done it before!

Said crash has been fixed.

[Syko]

I don't know if anyone else has this issue, but the maximum resolution I can run Jedi Outcast (or Academy for that matter) is always 1024 x 768. When I set it any higher than that it looks horrible - well below that resolution. Is this something that could be fixed with OpenJK?

[eezstreet]

er..define horrible.

[Syko]

I just tried it. It isn't that bad, but the resolution is lower when it is supposed to be higher. I set the game on 1600 x 1200 and in actuality it is set on 640 x 480.

[eezstreet]

Err.. Do r_mode -1, and set the resolution using r_customwidth/r_customheight, then r_fullscreen 0, and finally a vid_restart. What happens?

[Syko]

I know, I fixed it via the cfg file. I set the r_custom height/width correctly and put r_mode -1, and it's at the correct resolution now. I'm just wondering if there's something you could do to OpenJK so you don't have to open a CFG every time.

[Szico VII]

Any chance of being able to implement a map-end MAN VIS system like medal of honor? Dont know which bit of code, q3map , renderer or all of them would need looking into with that though...

 

Basically it allows you to manually set and link vis portals using brushes with a system/vis type shader. You can then manually control which portals can be seen from which areas (instead of it being automated based on geometry, leaf nodes e.t.c). This would be SO helpful, particularly for large open maps but also in pretty much any map.

[mrwonko]

[...]

So just doing the vis manually? I don't see how that would be helpful, wouldn't that just be a source of bugs without any actual benefit? How does it help mappers?

[Szico VII]

Not the WHOLE vis process obviously, that becomes entirely superfluous when you use MANVIS.

 

Think about it - you have a massive outdoor sprawl, or perhaps just a normal sized one? Perhaps a few mountains in the middle or some trees. Or anything. What do you do to block sight through it? Normal vis wont do it, even with ideal structural brush work. Antiportal we all know doesn't work as we'd want it too.

 

Or perhaps just a complicated indoor layout that you cant get to render as you want it too, holes in the floor perhaps, odd shapes, certainly to block portals and mirrors e.t.c.

Here, maybe this tutorial on how it works in moh would be helpful in explaining it.

 

http://homepage.eircom.net/~abyrne/sdk/tutorials/manvis_tutorial.html

 

Honestly, if I could set vis manually (or at least certain parts) you'd be able to design much more natural looking maps without worrying about the engine drawing everything at once or constraining the design so the vis will work. Imagine a new..say modern warfare 3 map in q3, just trying to get the visibility drawings right would be a pain, given how open some of them are.

[eezstreet]

Its a long shot request, and since it really doesn't benefit anyone (especially since 85% of mappers can't even vis properly to begin with) and only in very severe circumstances, I'd say there's a slim chance of it getting implemented anyway. Ask q3map2 devs if they have it implemented, also.

[Szico VII]

Then it would definitely be of use to that arbitrary 85% you just referenced, surely? It would benefit everyone who had improved performance in their maps and for those who played said maps.

Severe circumstances? I cant think of a single map where manvis wouldnt have improved framerates and visibility...

 

Get me a mapper in here to help me out

 

It would in fact be ESPECIALLY useful for those who cant get their head around how to optimise their vis properly for automatic portalling because you can pick and choose where you want things to happen yourself instead of modifying rooms,brushes and designs to get a less effective end result.

[mrwonko]

http://homepage.eircom.net/~abyrne/sdk/tutorials/manvis_tutorial.html

I still don't get it. That page mentions

In essence, what you are doing is taking a load off the CPU’s visibility calculations at runtime by predetermining it yourself.

Q3Map2 already predetermines the visibility during the VIS stage, there's no runtime PVS calculation. If you want to influence the VIS, just add some antiportal/hint brushes?

[eezstreet]

or or or

just use a portal based vis system instead of PVS, which is practically an antique these days.

 

And yes, what mrwonko said. Vis is already calculated prior to runtime in the vis stage, and all the server has to do is figure out if they're in the same PVS. You still have to do this regardless of whether you're using manually defined vis or not, because it's the same exact data ergo, handled the same way. This is why I think its much more relevant to talk to q3map2 developers about this, since it's mostly their job to handle MANVIS.

 

btw, if 85% of people can't vis properly, why would a change in vis help them? It'd only hurt them really, since most JK2/A oriented sites focus on "auto" vis, or the current vis system.

 

Manusl vis also sounds more error prone and more usage of brushes/more time consuming in general for a mapper to use.

[Delta_135]

Get me a mapper in here to help me out

 

i'd love to have some more control over VIS too

no matter how many tutorials i read on vis portals it just never work out the way i want it too

so i usually just end putting down a bunch of doors with arena portals in them

although that never seems to work as intended nether...

 

and since i'm a coding newbie i have no idea how one would even code something like that.

 

EDIT: can't seem to get the quote thing to work probably...

[Szico VII]

I still don't think you guys get what I mean - the benefits of being able to MANUALLY set what can be seen and where would be totally invaluable. Hints and portals never give you the kind of specific control that man vis would. Perhaps we're getting confused over terminology but thats the esseence of what appeals about manvis.

 

As for that CPU comment in post - yeah that saves only up to 5% and is already auto calculated. Thats not where the majority of the fps saving comes from, it comes from effectively blocking or causing areas not to be drawn which would otherwise have been using the q3map2 vis tracing. Is that a better way of putting it.

 

And @@MoonDog...Um... not so much? Thats still creating portals automatically or with hint, which usually only works well in corridors or interior areas. Although Id take that scissoring and stuff over the current system.

[mrwonko]

Maybe we should make a new VIS help thread where people can send me their maps and I can show them how to do proper VIS...

 

Yes, having a PVS at all makes some things inherently worse, and adding an optional additional portal/antiportal system might help, especially for large outdoor stuff. But that would also benefit from some LOD'd terrain, for example. In general the lack of LOD on anything but playermodels makes huge levels impractical anyway... Yeah, there are a lot of things we could improve about Jedi Academy. But I don't see how manually doing the PVS work Q3Map2 usually does for you helps.

[eezstreet]

 Perhaps we're getting confused over terminology but thats the esseence of what appeals about manvis.

I wat'd

 

 

As for that CPU comment in post - yeah that saves only up to 5% and is already auto calculated. Thats not where the majority of the fps saving comes from, it comes from effectively blocking or causing areas not to be drawn which would otherwise have been using the q3map2 vis tracing. Is that a better way of putting it.

Are you talking about CPU usage in q3map2 or ingame? I've read up on this fairly briefly, and MOH and JKA are essentially the same when it comes to the way it works ingame, it's the actual compiler itself that's different, as far as I can tell. The CPU usage you're referring to is probably the improved vis or whatever overall, and Manvis would not contribute to that at all, if you got the exact same groupings of tris to show up. Everything you've described so far sounds like it'd be better off being implemented on the compiler.

 

 

And @@MoonDog...Um... not so much? Thats still creating portals automatically or with hint, which usually only works well in corridors or interior areas. Although Id take that scissoring and stuff over the current system.

No it isn't. It's portal-based vis, instead of vis based on Potentially Visible Sets.

Furthermore, you contradicted yourself here by saying you'd "take scissoring and stuff over the current system", despite dismissing his suggestion as being the same as the current system.

[Szico VII]

Perhaps I should just have written - can we improve the vis system?
I'm not dismissing any system - if there's a better way to get a better visibility algorithm than what we have then fine.

Didn't dismiss MoonDogs link, sorry if it came across that way - but from what I took it's still a system that would only work well in indoor corridor type-areas. (Doom 3 )

What i really want to do is to be able to manually specify what bits of the map are drawn and when - to have complete control over it. Perhaps that's not a possibility, but it was stated that 'anything is possible' so I thought Id put the request out there!

 

P.S. @@mrwonko

 

Maybe we should make a new VIS help thread where people can send me their maps and I can show them how to do proper VIS...

 

That sounds like an excellent idea...I'll send you nightfall ;D

[eezstreet]

I'd look into how modern engines handle vis. I believe they use volumes and buffer zones similar to how Doom 3 handles it. Vis shouldn't be a hassle, it should be automatically handled by the engine.

[Szico VII]

Reading skills perhaps...? lD

[eezstreet]

Yeah, I was like "wtf?"

[Szico VII]

Sorry, didn't think I was.... everyone could stand to work on that though really... *cough*

ACCEPT MY HUMBLE APOLOGIES *waves hand*

 

I still would like a new vis system where you can manually control what is seen from where if you'd like to - whatever method used to implement it. Hopefully it will be considered.