Why doesn't anyone USE the terrain features?

by Rambozo on 03/28/2003 06:06, 26 messages, last message: 06/23/2003 23:59, 12565 views, last view: 05/18/2024 15:31

Cube is primarily a terrain engine, right? Why are there no maps that are wide open, outdoor affairs? From what I can tell, Cube is capable of very detailed, curvy and perfect terrain. Why doesn't someone churn outa Tribes 2-ish map with rolling hills and a few temples? Think of the sniper potential!

   Board Index   

#1: Uhhhh

by pushplay on 03/28/2003 08:25

Huh.

The initial design of Stalingrad 2215 was like that. It was a colossal failure.

reply to this message

#2: ..

by Snipa on 03/28/2003 17:03

sniper......sniper.... *drool*
heh, when i get good at mapping ill make oen :D

reply to this message

#3: well

by Aardappel on 03/28/2003 18:17

terrain in cube is brute force, very highly tesselated, and rendered in not very optimal ways for large throughput... meaning that the amount of heighfield stuff you can have in view without fps problems is severely limited. Cube really does best on quake style maps, ironically.

reply to this message

#4: Re: well

by Skaus on 03/28/2003 19:01, refers to #3

Wouldn't it be fairly easy to make some changes to the way heightfields are rendered, adding LOD etc to cut away lots of triangles? Cube is a terrain engine, it just isn't optimized for rendering terrain :)

Someone could do this for a mod if they have the skills. More terrain than you can currently make (not much) has no place in cube's gameplay anyway IMO.

reply to this message

#5: deep

by *spentron* on 03/28/2003 20:58

Wondering if a normal terrain engine is also larger scale.

Cube has some problems with rooms/spaces past a certain size. Maybe can be improved. Also could be improved by fog occlusion, not that that's good.

The current LOD improves framerate but causes nasty artifacts, for example a bit of heightfield turning into a column to the sky. The artifacts could be changed.

Cube editing makes it easy to create lots of polygons. Several of my maps show the current limits of distance and reasonable polycount (tested by crap CPU method ;) ). Space, main room, was started on Cube version 6-24-02 and had to be reduced slightly upon switching to 10-16-02 version. Better performance would be desired for DM. Also could increase minimum hardware specs (map effect on performance doesn't just affect Cube, try Alice or FAKK2, Q3 engine based, on a machine that barely plays Q3A).

reply to this message

#6: Re: deep

by Piglet on 03/28/2003 21:51, refers to #5

a 'normal' terrain engine will be about as detailed as cube is close up, but significantly less detailed further away (where the detail isn't required)

reply to this message

#7: Re: deep

by Skaus on 03/28/2003 22:28, refers to #6

i suppose the height-map wouldn't be as fine-grained either. Making heightmaps store height data for every 8th cube or so would help drasticly.

reply to this message

#8: ..

by Rambozo on 03/28/2003 23:31

Perhaps a new version of Cube will have a better ability for wide open maps....I'd really like to see bigger, open-air arenas in which to kill things. :-)

reply to this message

#9: Re: ..

by pushplay on 03/29/2003 02:17, refers to #8

Try Serious Sam.

reply to this message

#10: Re: ..

by Piglet on 03/29/2003 14:33, refers to #9

indeed, again it's a bit of a 'why cant cube be different' problem, basically it is what it is, and if you dont like that then a different engine might be in order

reply to this message

#11: Re: well

by Aardappel on 03/31/2003 14:26, refers to #4

it already does LOD, that is not the problem. You can have any map running at 30fps, even with insane amounts of detail, just you will get popup up close. I personally think the LOD sucks to the extend that it should be avoided. And it is not easy to change the algorithm to be better, given the current world structure.

There are quite a few tweaks that can be made to make it more suitable for larger scale maps, but it will never compete with a normal landscape engine. Cube works best with quake style maps, sorry.

reply to this message

#12: RE: serious

by *spentron* on 03/31/2003 18:52

Serious Sam's forte does seem to be large spaces without glitches. However, performancewise, Cube eats it for lunch.

reply to this message

#13: Re: serious

by pushplay on 04/01/2003 01:27, refers to #12

That's like saying apples eat oranges for lunch. Performance is measured in many ways.

reply to this message

#14: ..

by Rambozo on 04/01/2003 11:04

But apples do beat oranges. You can eat them without peeling them and they aren't all covered in white leathy skin when you finally do get 'em unwrapped, like oranges are.

reply to this message

#15: Re: well

by Skaus on 04/01/2003 18:16, refers to #11

Well, it might have LOD, but i doubt it's very good for terrain. I might be wrong though. By good, i mean that it would be able to draw a hilltop with roughly the same shape using both 10 and 1000 polies by having algorithms or whatever to ensure that the hilltop isn't completely flattened by losing 99% of it's detail.

anyway. Would it be hard to make another form of heightfields, that's pretty much the same, but uses larger cubes? I think that would result in acceptably good looking small terrain areas :)

reply to this message

#16: Re: well

by *spentron* on 04/01/2003 19:33

"to ensure that the hilltop isn't completely flattened"

Problem is now it may turn into a column when the LOD kicks in, which is far worse.

"Would it be hard to make another form of heightfields"

Interesting idea.

As to the apples vs. Cubes comparison, I really don't know how many polys or anything Serious Sam pushes, but I do know its CPU requirement is higher than even Cube with the polycount way higher than usual. Only one measure of performance there though no doubt.

reply to this message

#17: Re: ..

by pushplay on 04/02/2003 02:55, refers to #14

You can eat an orange without peeling it. I'm not saying I recommend it, but you could if you wanted.

reply to this message

#18: Re: well

by Aardappel on 04/02/2003 10:20, refers to #15

you can try making a "low poly" heightfield making it on 2x2 (or even better, 4x4) cubes at once, and ensuring all the vertices in between lie in line between the surrounding ones. if so, and if the lighting doesn't differ to much, cube will render that heighfield as 2x2 or 4x4 cubes. This is however way too hard work to do by hand, I would recommend writing a function to do this automatically, which can then be a postprocessing on perlin noise, say.

reply to this message

#19: Re: well

by Skaus on 04/02/2003 12:19, refers to #18

cool, so you're saying cube already supports this without any code changes? If i could be arsed, i'd test it :P

reply to this message

#20: Re: #18

by *spentron* on 04/02/2003 23:19

In other words, straight slopes. I figured it might do that but your first reply sounded like it didn't.

Slope command, cut and paste, grouped up/down adjustment ...

reply to this message

   Board Index