Looking good, but add some ambient light!

Surely you've just got some bad sector merging, or none at all. I know you can't have that many in that little thing.

I was about to comment that it is refreshing to see a level without any for a change.

Maybe work with the adjoin-turn-off cog and screenshot-textures? The short distances might be a problem, but it could be worth a try...

Just insert a big wall in the middle somewhere. That works wonders for framerates.

It looks wonderful! Even if it wont be playable I say you should finish it so we can all go oooh!

Or he could just arrange his cleaves better.

That too.

Very nice, love the the looks of it so far. Just curious, how many hours have you put into it so far?

[QUOTE=Joseph T]Very nice, love the the looks of it so far. Just curious, how many hours have you put into it so far?[/QUOTE]
Probably something shocking. I'm going to guess at 10-15.

And yes, I'm sure I can arrange the the cleaves better. It does become extremely hard in such a small space though. It's going to require some careful examination of the geometry, and a lot of patience.

I don't think you get out enough. You see, light bounces off surfaces. It helps light up other surfaces. It's called radiosity. JK doesn't do this, so we must fake it with ambient light.

Completely dark shadows like those are totally unnatural.

Actually, I usually use Dynamic Lights in my levels, this is the first time I've tried the Lighter plugin. I might end up ditching it anyway due to the number of surfaces it makes. DLs work beautifully in smaller levels, and I've grown used to using them.

The surface count usually isn't an issue on modern PCs, especially in small levels. Don't worry about it.

texture it !!!! dont waste your time doing lights when its not even textured..

looks nice

Yeah! Texture it! z@nardi! can you texture some ww2 guns and put them in-game for me? You really have to release your level!

This. isn't. real. life. It's a game. I think it looks better this way.

Games mimmic real life. They're simulations of real life. Generally, things that look more realistic are more attractive to the eye. But fair enough.

hey dash long time no talk!

Yeah. Haven't had internet on my computer since I moved. I use WAP to post here I'll try to get some broadband connection or something

Yea man I don't think you get out to the science lab enough. Jeeze what a loser

Oh, yeah, I'm such a loser for knowing how light works. I'm such a loser for being smart. I'm such a loser because I'll have a better job than you in a few years, making a lot more money and having a hotter girlfriend. Yeah, such a loser.

Piss off you little ****.

Emon is sort of right. Radiosity makes things look better and yes, light bounces. If light didn't bounce (if I remember correctly) we wouldn't be able to see anything. Anyway, nice map. I likes it.

Yeah, Reflections. Only a pure black surface absorbs light and doesn't bounce it.

If you could read you'd relise I wasn't calling you are loser. I was sarcastically calling freelance one, eargo the sarcastic face

Then maybe you should have quoted him, not me.

25,000 surfaces later...

:|



:O

Oh my...










Keep it that way!

Cool keep with it... I like that middle step

you;re odd my son.


AWESOME LEVEL>!!!

if(getThingFlags(source) & 0x8){
  do her}
elseif(getThingFlags(source) & 0x4){
  do other babe}
else{
  do a dude}

Ooh, and one more. But I'm still not freaking done with the architecture!

Holy crap, sweet! The outer walls need awnings and windows from unaccessable housing. Maybe even long clothes lines!!!!!!! Eeee! That's serously cool.

Ok let me clear this up, I was sarcastically agreeing with you, therefore I quoted you, else no-one would've known what I was talking about.

Now, that, level, looks, ****ing, awsome. And i've gotta say that is indeed an awsome step. How did you do it?


-a_amazed_person

If that's the case, I apologize.

looking great - but, like so many jk levels, the guy who invented the cell and portal vis det. algorithm must be turning over in his grave.. erm that is, if he's dead. otherwise you're just giving him a terrible loin itch.

considering the fact that todays compys are powerhouses compared to jk's req'd specs its not really a problem. but if i were you, id convert a lot of that into a 3do or two and drop in a couple dynamic lights. you'd lose the shadows, but the performance gained by not running through hundreds of sectors every render could be spent on lighting. but again; its your level and i'm not you. and in any case, it cant be any more devestating on sector traversal than drazen's outdoor areas.


ranting and bashing aside, it looks pretty bad-*** so far. im looking forward to seeing some [more] textures in place - and i hope there are going to be traps!

StrikeAthius: turning it into 3dos would be a bad idea

Firstly, it isn't possible to convert a level to 3dos without remaking it from scratch. Secondly, dynamic lighting is awful. I can't fully describe how much more effort it requires than regular lights, but let me just say the fact that intensity and range aren't exclusively controllable is a pain in making appealing dynamic lighting. Thirdly, the collision-detection of 3dos is a piece of ****--the ability to pass through 3do surfaces is not uncommon. Finally, I'm fairly certain occlusion culling doesn't work on 3dos, or at least that's the only thing I can figure, since levels I made out of 3dos still had terrible framerate.

Theoretically it seems like a good way to cut down on surfaces, but, to summarize, its suckiness outweighs the minimal benefits.

Anyway, awesome work, that's some of the best lighting I've ever seen in JK.

It's the lighter plugin.

fixed

where there's a will, there's a way.

a JKL describes all the visible surfaces in a level. so does a 3do model. it wouldn't be very difficult to write a tool that re-organizes the way those surfaces are represented; a 3do equivelant of a jkl STILL DESCRIBES the same exact visible/collideable surfaces. same polygons with the same vertices and the same texture coordinates.




i won't entirely disagree with you there..




probably true, but i've never personally seen evidence of this.

one problem with a '3do level,' though, is the massive number of sphere/plane collision checks that normally do not occur in even 'highly adjoined' JKLs such as this one.

to fix the collision-detection, as well as speed up the collision-detection, a non-collidable 3do version of the level could be inserted in the ORIGINAL level; with all its original sectors intact.

then set the geomode of each of the sector's surfaces to 0. turn off visibility traversal through the sectors by changing their adjoin flags. you STILL GET the superfast collision detection of the level geometry itself, and you STILL GET the big performance boost from not traversing the sectors during rendering



it is fact that jk does not support occlusion culling. this is evident because the original LEC door cogs manually turn off adjoin visibility through doorway sectors. also, occlusion culling wouldn't buy a dang thing if the entire level were just one 3do. a 3do cannot cull parts of itself.

if you were implying a type of 'surface culling' i can tell you it would give no performance increase, since a polygon must be fully rasterized in order to see if it is occluded or not.




the goal is not to cut down on the number of surfaces, but the number of sectors that must be traversed. if it's going to end up rendering a huge proportion of the level's surfaces ANYWAY, you might as well not do a computationally expensive test before rendering them all. you're trading off overdraw and algorithm efficiacy.

making a 'level 3do' its not 3do's intended use, but neither is disabling the 'visible adjoins limit' so that levels like this are renderable. there's a reason that there is a maximum number of visible adjoins hard-coded in: performance.

I know how unfashionable it is to dissect posts to provide an outlet for wit-filled rebuttals, but I'm afraid I have to lower myself to such depths.

Certainly, but until that will actualizes the aforementioned way, my point stands.

I'd question your sanity otherwise. ;P

I actually have a level with this problem, so if you really want evidence I can send it to you.

Err, I'm fairly sure changing the sector's surfaces to 0 would give you a HOM--assuming, of course, you mean the non-3do's sector's surfaces. I guess I'm not following you.

Hmm, I'm afraid I am again sent into a tizzy. Maybe my OpenGL books or literacy have misled me, but isn't occlusion culling by definition surface culling? Perhaps you thought I meant portal clipping.

Semantics aside, that 3dos are not culled in any beneficial manner is my point. Indeed, what's more, non-3do levels are.

You assume that a majority of surfaces are non-occluded, which in this case is quite a rash conclusion considering Daft's level obviously has quite a few interior spaces (ergo, occluded surfaces).

Whether a portal engine or a pathetic excuse for culling creates the better performance is ultimately down to the particular level. In my experience with levels that have a large number of occluded surfaces, the performance favors the former.