Could it be the models used ?

General discussion of the Project Reality: BF2 modification.
Cassius
Posts: 3958
Joined: 2008-04-14 17:37

Could it be the models used ?

Post by Cassius »

One of the earliest tricks to boost 3d performance, was to eliminate all polygons and textures that are not visible.

For example an orb used to be rendered filled out on the inside. If you would have a black background and put a cam inside a blue orb you would see blue.

By rendering only the outside shell and putting textures on that only, if you put the cam at the center of a blue orb, I guess you would see the black background color, because only the skin of the orb is rendered and colored.

Now I suppose that thesedays avoiding superflous polygon and texture count this way is something 3d programs do by default, but is there a chance that models have been used which render polygons and textures not visible in the game world ?
Hitperson
Retired PR Developer
Posts: 6733
Joined: 2005-11-08 08:09

Re: Could it be the models used ?

Post by Hitperson »

i'm sure the devs are aware of removing backfaces.
Image
Harrod200:"Fire.exe has committed an illegal operation and has been shut down"
Raniak : "Warning: May crash if fired upon."
M4sherman: "like peter pan but with tanks"
[R-MOD]Eddiereyes909 (on sim tower) "It truly was the game of my childhood and has led to me getting my degree in industrial engineering."
Dragonfire43560
Posts: 157
Joined: 2006-06-30 10:20

Re: Could it be the models used ?

Post by Dragonfire43560 »

Devs, just try to backpedel a bit. What NEW objects did you just recently put in? Maybe a building? A tree? Its gotta be something like this
Ragni<RangersPL>
Posts: 1319
Joined: 2007-08-13 10:44

Re: Could it be the models used ?

Post by Ragni<RangersPL> »

I would say:
It may be the maps.... or new files structure *looking at OBJECTS folder content*... I'm just guessing over here so don't hit me If I'm wrong :)
ImageRANGERS LEAD THE WAY!!!
:29_slaps: Do not post stupid suggestions just because you had a bad round in PR :fryingpan
Triger_Finger
Posts: 5
Joined: 2010-02-06 00:26

Re: Could it be the models used ?

Post by Triger_Finger »

I don't know about you guys, but maps that have alot of buildings (mostly insurgency maps) are the ones that give me trouble with performance, I had this issue back in .87 too with some insurgency maps (fallujah etc.)
Tinfoilhat
Posts: 84
Joined: 2008-11-23 22:06

Re: Could it be the models used ?

Post by Tinfoilhat »

Triger_Finger wrote:I don't know about you guys, but maps that have alot of buildings (mostly insurgency maps) are the ones that give me trouble with performance, I had this issue back in .87 too with some insurgency maps (fallujah etc.)
I seem to be okish on maps with loads of buildings, like Beirut and the other one with the Russians and IDF that the name escapes me right now, however anything with the new grass and vehicles is a disaster area!
At times im not even getting Frames Per Second, its more like Seconds Per Frame! :shock:

That could be my ancient computer, but it would run 0.87 fine most of the time, only if there was half the server population within 50m of me and everyone was shooting and grenadeing etc would it get really crappy.


Although the voices arent real........ :? ??:
They have some pretty cool idea's! 8)
CodeRedFox
Retired PR Developer
Posts: 5919
Joined: 2005-11-08 00:47

Re: Could it be the models used ?

Post by CodeRedFox »

Dev's know all about polygons and the engine will not remove unseen polygons real time. Let me give you a little lesson on game polygons. *teacher voice on*
By rendering only the outside shell and putting textures on that only, if you put the cam at the center of a blue orb, I guess you would see the black background color, because only the skin of the orb is rendered and colored.
All polygons (a surface with four sides) are one sided only, which mean non of them have a backface that renders. Hence, like you said, the camera would see not see the object. In programs like 3dsmax you can switch on backface rendering or use two sided material but this does not happen in the game, ever.

Allot of different issues can crop up when doing game models as you have to look into (lod's, texture sizes, memory space). When the first version of Korengal came out I have used a large texture on the grass fields that got applied wrong. This worked great on my video card but smaller memory cards the level lagged really bad. I went all over that level looking for the issue and never found it until Outlaw? found the issue which instantly raised the FPS.

Now I know for a few PR is having issues. And we know this and guys are working to figure it out. The main issue we have is while the testers do a great job we do not have a wide range of video cards to test (after all we are all gamers so low end cards are normally not in the mix). The best thing to do is try and see where it happens and when it happens and how often it happen.

Help us help you.
Image
"apcs, like dogs can't look up" - Dr2B Rudd
Warden(Sco)
Posts: 19
Joined: 2009-03-01 22:31

Re: Could it be the models used ?

Post by Warden(Sco) »

[R-DEV]CodeRedFox wrote:Dev's know all about polygons and the engine will not remove unseen polygons real time. Let me give you a little lesson on game polygons. *teacher voice on*



All polygons (a surface with four sides) are one sided only, which mean non of them have a backface that renders. Hence, like you said, the camera would see not see the object. In programs like 3dsmax you can switch on backface rendering or use two sided material but this does not happen in the game, ever.

Allot of different issues can crop up when doing game models as you have to look into (lod's, texture sizes, memory space). When the first version of Korengal came out I have used a large texture on the grass fields that got applied wrong. This worked great on my video card but smaller memory cards the level lagged really bad. I went all over that level looking for the issue and never found it until Outlaw? found the issue which instantly raised the FPS.

Now I know for a few PR is having issues. And we know this and guys are working to figure it out. The main issue we have is while the testers do a great job we do not have a wide range of video cards to test (after all we are all gamers so low end cards are normally not in the mix). The best thing to do is try and see where it happens and when it happens and how often it happen.

Help us help you.
Then perhaps you should include a wider scpectrum of machines in testing which i think is common sense. In fact not doing so is what has caused this problem.

Not everyone can afford to upgrade every year and alienating players is something that cannot be afforded especially when there is only 5/6 Servers full most of the time. (GMT)
Teek
Posts: 3162
Joined: 2006-12-23 02:45

Re: Could it be the models used ?

Post by Teek »

Warden(Sco) wrote:Then perhaps you should include a wider scpectrum of machines in testing which i think is common sense. In fact not doing so is what has caused this problem.

Not everyone can afford to upgrade every year and alienating players is something that cannot be afforded especially when there is only 5/6 Servers full most of the time. (GMT)
the Dev team works for free, they cannot afford to build hundreds of computers, the only computers they can test on is those of the Testing Team.
Image
ff7redxiii
Posts: 22
Joined: 2009-01-26 09:58

Re: Could it be the models used ?

Post by ff7redxiii »

All the new lag is coming from the shrubs, etc that is now drawn much much farther out. Also some of the newer buildings, and the like do not seem to have lower poly models for farther distances, the old problem with QinLing before they fixed it was the game would draw EVER tree on the map, even the ones you cant see behind the next hills, so it lagged. Similar issues seem to be showing up on the new maps because of the new draw distance with shrugs and new models, etc.

I am sure they will fix it soon enough but right now, turn down geometry to low and other settings as you see fit.
CodeRedFox
Retired PR Developer
Posts: 5919
Joined: 2005-11-08 00:47

Re: Could it be the models used ?

Post by CodeRedFox »

Warden(Sco) wrote:Then perhaps you should include a wider scpectrum of machines in testing which i think is common sense. In fact not doing so is what has caused this problem.

Not everyone can afford to upgrade every year and alienating players is something that cannot be afforded especially when there is only 5/6 Servers full most of the time. (GMT)
Send a check to : [email protected] and we will be glad to buy some older computers to run checks on. We do with what we have. We dont have allot of slow computer players signing up for testing roles.
Image
"apcs, like dogs can't look up" - Dr2B Rudd
CanuckCommander
Posts: 431
Joined: 2008-03-19 02:25

Re: Could it be the models used ?

Post by CanuckCommander »

Warden(Sco) wrote:Then perhaps you should include a wider scpectrum of machines in testing which i think is common sense. In fact not doing so is what has caused this problem.

Not everyone can afford to upgrade every year and alienating players is something that cannot be afforded especially when there is only 5/6 Servers full most of the time. (GMT)
That's got to be the stupidest post I've ever read. To refute your point, it is also "COMMON SENSE" to buy a good video card to play a game on a PC, especially a game that is more than 5 years old!!!

Hmm, I see why more developers are moving to consoles...less people b1tching about upgrading.
ma21212
Posts: 2551
Joined: 2007-11-17 01:12

Re: Could it be the models used ?

Post by ma21212 »

I was thinking the ALT-TAB hotfix..?
Also my computer is pritty old and slow, (P4 2GHz, ATI 9600, 1 RAM), would love to be a tester.
Image
Image
hiberNative
Posts: 7305
Joined: 2008-08-08 19:36

Re: Could it be the models used ?

Post by hiberNative »

i just dunno why karbala flows like before but fallujah stutters just because the main base got moved?
-Image
Cassius
Posts: 3958
Joined: 2008-04-14 17:37

Re: Could it be the models used ?

Post by Cassius »

Well I can see how the devs want to take advantage of better grafic cards reaching the consumer, even if an radeon 5770 running the latest PR with 40 FPS does not look as good as the same card running Crysis maxed out with the same FPS.

It just had a suspicion seen as some people had a dramatic drop when facing certain objects, but I guess that drawing only the visible parts of the objects and not filling it out with unseen polygons, bitmaps is pretty much default by now on any modelling software.

I have a radeon 3200 HD which is pretty dated. In some spots i have 40 fps in others more like 17 and sometimes even less when zoomed in, I guess I will have to bite the bullet and set textures to medium until I upgrade my card.

Also Antialising does not increase performance for me, so you might want to try and switch that off.
Last edited by Cassius on 2010-02-07 04:23, edited 1 time in total.
BloodBane611
Posts: 6576
Joined: 2007-11-14 23:31

Re: Could it be the models used ?

Post by BloodBane611 »

The thing that will help the DEVs solve this the fastest is data on your computers and where you are having performance hits, posted in this thread: https://www.realitymod.com/forum/f10-pr ... plate.html
[R-CON]creepin - "because on the internet 0=1"
Spinkyone
Posts: 200
Joined: 2008-07-02 22:40

Re: Could it be the models used ?

Post by Spinkyone »

It should be said though that the older Nvidia G80/G92 series, which is pretty much every 8800/9800, are AK47 common and now have major issues with PR from my friends experience. Capable of maxing out the FPS in one direction and then slowing to an unplayable <30 FPS in another indicates something is very wrong indeed.

The question that should be asked is why I'm a looking at a blank wall and nothing else and still have <30 FPS and yet in another area I can have a screen full of vehicles and have max FPS? The mind boggles.
Dragonfire43560
Posts: 157
Joined: 2006-06-30 10:20

Re: Could it be the models used ?

Post by Dragonfire43560 »

Spinkyone wrote:It should be said though that the older Nvidia G80/G92 series, which is pretty much every 8800/9800, are AK47 common and now have major issues with PR from my friends experience. Capable of maxing out the FPS in one direction and then slowing to an unplayable <30 FPS in another indicates something is very wrong indeed.

The question that should be asked is why I'm a looking at a blank wall and nothing else and still have <30 FPS and yet in another area I can have a screen full of vehicles and have max FPS? The mind boggles.
This exactly describes my situation and I have a ....wait for it





8800 GTS 328mb
M_Striker
Posts: 513
Joined: 2008-05-31 00:36

Re: Could it be the models used ?

Post by M_Striker »

CanuckCommander wrote:That's got to be the stupidest post I've ever read. To refute your point, it is also "COMMON SENSE" to buy a good video card to play a game on a PC, especially a game that is more than 5 years old!!!

Hmm, I see why more developers are moving to consoles...less people b1tching about upgrading.
May be so, common sense, but I hope you realize a lot of people here are kids and don't have the money to buy 100 dollars worth of computer parts.
amazing_retard
Posts: 376
Joined: 2008-10-01 03:13

Re: Could it be the models used ?

Post by amazing_retard »

M_Striker wrote:May be so, common sense, but I hope you realize a lot of people here are kids and don't have the money to buy 100 dollars worth of computer parts.
Whats the point of all this arguing? I have a shitty comp, is there a place I can sign up to be a tester?
Post Reply

Return to “PR:BF2 General Discussion”