rattle winsrattle wrote:The point is it could have been the same pretty ship with half the polycount.
Model Competition 0001 - Small Fighter/gunship/bomber
Moderators: MR.D, Moderators
- LathanStanley
- Posts: 1429
- Joined: 20 Jun 2005, 05:16
Zenka wrote:That's the starship Caydr made a year ago for GEM. And since then he explaind around 14 times that the polycount does NOT ruin the FPS.
(It is a pritty ship though, you should see it textured )
BULLSHIT!
I had to make that obvious.. sorry if its "offensive".... but on my Severnaya map I ran the trees origionally with only ~40 faces and people wanted more detail... I re-modeled them, used the SAME size texturemap and remapped them as well... the result, ~120 polys each, and a drop of ~20% in FPS...
though there are about a thousand trees on the map total, there ARE a thousand units in the game... if you take 1 of every 100 units, say 10 fighters in a game, and crank their polycount ten fold... I doubt you'll notice a difference... but if we remodel EVERY unit to 10x the facet count...
lol... expect a 80% drop in FPS.
polycount matters.. and frankly.. 1000 is WAY too high... its pretty.... SURE... but somewhere around 200 is MUCH MUCH MUCH more reasonable..
- LathanStanley
- Posts: 1429
- Joined: 20 Jun 2005, 05:16
OTA used quads to render, spring uses tris REGARDLESS of how its textured...Das Bruce wrote:If you want to count quads as one then you're going to have to 3do texture it.
since it uses tris, even if you textured it as quads as a 3do, it will still render as slow as a s3o using tris...
its not a matter of "modeltype" its a matter of rendering method and tris...
Re: Model Competition 0001 - Small Fighter/gunship/bomber
Its fine if those are for his own mod, but not this competition.P3374H wrote:Uhh...Rules:
-under 1000 polys
- LathanStanley
- Posts: 1429
- Joined: 20 Jun 2005, 05:16
are you kidding me??LOrDo wrote:snip
(speaking for only the 3 "fighter" craft he has there...)
they are poor creativily...
none of them can be used as an environment bound by normal physics fighter craft.
the first two look like blobbed spaceships with simplistic domes and spheres twisted into collumn-like fuselage fightercraft..
all of them exceed poly count.
none of them have modeled weapons... except the last one thats basically a T-50something whatever fighter from star-wars... (its prolly the best model of them all though, it actually has some modeling thats just not a bubble, on a bubble...)
none of them fit a TA style mod or current AA/XTA game except maybe the last one... and again.. its too much like star wars for my taste...
To be fair, this mod is NOT TA or even TA based, so the units not looking like TA is not th issue here, nothing was specified in the rules either about TA style units. Personally, I like the look of the ships, although the polycount does look a little large, some bits could be reduced with hardly any, if any at all, loss in quality.
- LathanStanley
- Posts: 1429
- Joined: 20 Jun 2005, 05:16
The latter got nothing to do with fitting in TA now does it? :)LathanStanley wrote:meh... I dunno...
I was prolly too harsh... but I wasn't impressed by them at all...
They're neither extraordinary nor really outstanding but they're fine. And probably will fit to the rest of GEM (I suppose it's a GEM model).
- Wolf-In-Exile
- Posts: 497
- Joined: 21 Nov 2005, 13:40
I'd like to put my 2 cents in regarding polycounts.
A note: When I say "polycount" im referring to triangulated models.
I reiterate LathanStanley's statement that model polycounts DO, in fact, have an impact on gameplay.
If not, why don't we see cinematic-quality models being used in any game? Why do we use texturing tricks like bumpmaps, normal maps and so forth to add details if the polycount does not matter?
The reason for this is the higher the polycount a model has, the more processing power, GPU and CPU is required to render, and the toll taken on performance by shadows, reflections, dynamic lighting, HDRI/Bloom effects and any other fancy graphical bells and whistles are directly multiplied by the amount of faces on a model that is being rendered.
Animations also take a fair amount of processing power, and high-poly, animated units will make a noticeable difference in the FPS.
Please do not make such ignorant statements on the basis of "omfg that model looks cool so screw the polycount!"
I and the other people here who have advised to keep polycounts down are speaking from an educated and experienced point of view and certainly not noobs to modding or modelling.
A note: When I say "polycount" im referring to triangulated models.
I reiterate LathanStanley's statement that model polycounts DO, in fact, have an impact on gameplay.
If not, why don't we see cinematic-quality models being used in any game? Why do we use texturing tricks like bumpmaps, normal maps and so forth to add details if the polycount does not matter?
The reason for this is the higher the polycount a model has, the more processing power, GPU and CPU is required to render, and the toll taken on performance by shadows, reflections, dynamic lighting, HDRI/Bloom effects and any other fancy graphical bells and whistles are directly multiplied by the amount of faces on a model that is being rendered.
Animations also take a fair amount of processing power, and high-poly, animated units will make a noticeable difference in the FPS.
Are you a modeller? Have you tested the engine's tolerance for high-poly models?Caydr, screw the polycount phobia noobs, those are fucking pwn.
Please do not make such ignorant statements on the basis of "omfg that model looks cool so screw the polycount!"
I and the other people here who have advised to keep polycounts down are speaking from an educated and experienced point of view and certainly not noobs to modding or modelling.
Just a quick point... Has anyone tested the engine's tolerance for high-poly models?
Because the only person I know of that has tested and posted his results on this forum is infact Caydr, and as His resurch showed, in the case of his ships (which are probable a few peices max) there was very little drop in frame rate with his several 1000 pollie models compared to just normal couple of 100 poli stuff. And I seem to remeber his machine isn't some sort of super computer. So, seeing as his tested his moddels in spring and they work fine at the poli count there at why are people telling him they are to many pollies?
aGorm
::although yes, they are to many for the comp, im talking about people saying that in genral there to high pollie. And yes, Im sure some of you uber moddelers could do just sa well with half as many pollie. but thats also nothing to do with it.
Because the only person I know of that has tested and posted his results on this forum is infact Caydr, and as His resurch showed, in the case of his ships (which are probable a few peices max) there was very little drop in frame rate with his several 1000 pollie models compared to just normal couple of 100 poli stuff. And I seem to remeber his machine isn't some sort of super computer. So, seeing as his tested his moddels in spring and they work fine at the poli count there at why are people telling him they are to many pollies?
aGorm
::although yes, they are to many for the comp, im talking about people saying that in genral there to high pollie. And yes, Im sure some of you uber moddelers could do just sa well with half as many pollie. but thats also nothing to do with it.
I sort of did, spammed 50 flying what-evers with a polycount of roughly ~1000 using 256x256 textures. They were using the explosion generator as well so that's another 4 particles or 4*6 triangles on the clock (plus textures). It was running half decent when all were moving here but my comp is really a bad example (friggin old)...
Anyway the same on 50x 2500 polygon monsters using 512x512 texures with only two particles each model had quite some loss in FPS as result. It's either the polycount or textures or both. Though I'm not sure yet if spring renders hidden geometry as well. Technically only 1500 tris were visible at all time.
Anyway the same on 50x 2500 polygon monsters using 512x512 texures with only two particles each model had quite some loss in FPS as result. It's either the polycount or textures or both. Though I'm not sure yet if spring renders hidden geometry as well. Technically only 1500 tris were visible at all time.
GAHHHHH...easy people...
Yes, the point of this is to generate models for new gpl mods.
..And yes, I did state 1000 polys (which is rather high), which has been clarified to triangles. There will be no more arguing over that fact.
I'll wrap this comp up fairly soon, after a couple more entries probably.
Yes, the point of this is to generate models for new gpl mods.
..And yes, I did state 1000 polys (which is rather high), which has been clarified to triangles. There will be no more arguing over that fact.
I'll wrap this comp up fairly soon, after a couple more entries probably.
huge polycount lovers, check this out: http://en.wikipedia.org/wiki/Image:Norm ... xample.png
- Wolf-In-Exile
- Posts: 497
- Joined: 21 Nov 2005, 13:40
^ Yes, normal mapping is awesome but it isn't supported by Spring afaik.
If only it were.....
See "Normal Map Generation" section:
http://www.unrealtechnology.com/html/te ... ue30.shtml
If only it were.....
See "Normal Map Generation" section:
http://www.unrealtechnology.com/html/te ... ue30.shtml