The netcode of Quake3 forced modem fans of Quake2 to upgrade to highbandwidth.

Quake3 networking sucked ass at the time. Its about time they fixed it.

It doesn't specify on their site, but from my understanding of q4, both sides would need it, because unlike HL1, and Q3, client and server share the same dll. This was/is a limiting factor for server side bots I believe. If there's an option to disable the dll check(like the q3 pure server option?) it might work as server side only. Otherwise afaik the game verifies your pack file matches, and they wouldn't.

This comment was edited on Nov 29, 21:01.

So if it's just client side, does the person hosting the game and I both have to have it installed to benefit?

I'm waiting for the day when in a multiplayer game the helmet actually protects the player, causing me to have to aim for the eye.

BOOM! EYESHOT!

------
"Oh how awful. Did he at least die peacefully? To shreds you say. Well, how's his wife holding up? To shreds you say."

I think per poly hit detection has potential to be cool, though I didn't get into D3 enough to see or remember if they used any of the cool aspects. If you know what polygon you shot you can look up the material you shot, and the material can determine whether you hit flesh, a section of hard armor, a section of lighter armor, etc. Player probably wouldn't notice much difference unless you decorate all the varying hits with different hit effects and stuff, but ultimately per-poly hit detection is the only good way to do this, then in an individual triangle on a model could potentially have as many hit effects as it has texels.

And Zeph, Doom 3 engine is Client/Server. wtf are you talking about?

And on that day, you'll complain about how sucky the game is, because Doom 5 will have hugely better graphics, Planetside 2 will support 5 times the players, and some third game (because Tribes is dead) will have even faster gameplay.

We're never satisfied with what we have. And no game will ever be able to 'do it all' best, because some OTHER game that concentrated on one thing will do that one thing far better.

I was a big fan of that netcode improvement Valve released, despite the fact that I generally play on servers I get under 50ms ping to (some as good as 12ms). And the 'high pinger gets the win' on a tie was wildly overblown... here's how it worked (possibly how it still works):

I pull the trigger. My 'shot' flies through the Internet, arriving at the server around 20ms later. The server holds that shot while it is calculating the current frame... say, about 15ms on a 30tick server (30 updates per second). If another person's 'shot' arrives during that same tick, and we're shooting each other, then the server looks at the ping of the two people.

I have a ping of about 40ms (20ms to the server, 20ms back), and the other guy has a ping of 200ms (100ms out, 100ms back). If the packet that contains his 'shot' arrives in the same tick that mine does, that means he actually fired it BEFORE I DID.

That is what Valve meant when they said 'high pingers win ties'. Only if a shot is within the same tick will this matter. On a 100tick server, that means the packets have to arrive within 10ms of each other. And the high ping player didn't really tie... he shot FIRST, it's just that his 'shot' took longer to arrive at the server.

if the high-ping players 'shot' arrives in a later tick than mine, even if he shot first, he is still going to lose because the server doesn't retroactively go back and not kill him when it learns he pulled the trigger first. The higher ping player will still lose in most cases when they pull the trigger first, unless their 'shot' arrives in the same tick as the lower-ping player's 'shot'.

Let this horrible rumor die.

Ahhh... I wait for the day when we can have Doom 3 level of detail, Planetside player count, and Tribes 2 speed.

What is really strange is why don't these game companies dissect Tribes 1 and 2 netcode and figure out how they were able to allow 16/16 and 32/32 games with vehicles in full 3d with minimal to no lag? And this was how many years ago?

Some T2 games supported up to 128 people smoothly. To me, that is amazing and I'm dumbfounded as to why it can't be duplicated in today's newer engines...theres some pretty obvious reasons here but there's always workarounds.

If that means that you died, even though you shot a guy with an AWP, heard the gun go off & all that, then that's just too fucking bad for you then. The server says your client is a liar & overrides you & you die.

I thought that was bad enough until I tried Halo 1 MP for the PC...so much bs going on there in every single game I play because of the networking model/Gearbox's shitty port of it. The game's still fun though...:)
This comment was edited on Nov 29, 18:46.

I used to say that CS always flipped a coin on who died. Battlefield does that to this day (Firing a machine gun, full auto at someone's back, 5 feet in front of you, and watching the bullets hit the ground, cement, and a bird flying overhead)

Still, I must say, I am amazed any of this shit works at all!

Per Poly hit detection, while it's nifty in concept, is not really fast or small enough in terms of code. Are you really going to care if you shoot someone or something, and have every shot mapped out perfectly?
Instead of per-poly, they should do it in larger patches. Not hit-boxes, mind you, as those can lead to some goofy reactions in multiplayer, people screaming at each other on who hit who, and hacking.
I'm talking larger patches, like sections of muscle or armor groups. Like the chest, right? You can look at the chest like a box. Divide the box by 4. You've got both pecs, and both sides of the belly. Divide each of those by 4. Now you have 16 possible spots to hit. More resolution up to dividing each of those spots by 4. You have a mesh, that isn't 'per-poly', but clusters of polygons that are close enough together to give the right effect, but 'wide' enough to accept hit data that it will give the appearance of that area being damaged.

Think of it like normal mapping, but for hit boxes.

This mod is a server side only mod that should make network play smoother in these games, right? The author doesn't make much clear on his site...

No, it is a user side only mod.

Of course it is a user-side mod, or to be more exact it is a client-side mod. The reason being is because Doom3, just like Doom2 before it is a Client-to-Client networking game. The reason that it is Client-to-Client is because it's easier & more accurate to share data that way than Client-Server-Client networking.

In a game like Counter-Strike, a client connects to the server & everything he does is sent to the server & the same for everyone else, so there's a constant decision-making process going on. If you & some guy shoot eachother at the exact same millisecond, then the server makes a decision & it is passed onto the clients. If that means that you died, even though you shot a guy with an AWP, heard the gun go off & all that, then that's just too fucking bad for you then. The server says your client is a liar & overrides you & you die.

In fact, with games like Counter-Strike a client *CANNOT*, repeat *CANNOT* ever make your ping higher just because he has a high ping. I'm sure you've seen this. The people will say, "Kick the high pingers!!!"

But on a server-side game like Counter-Strike, BF2, etc., THE SERVER IS IN CONTROL NOT THE PLAYERS!!! So if someone had a 4,000ms ping, it would ****NOT**** affect you in the least bit at all.

(This rumor came from people paying big fucking bucks for high-bandwidth in the late 90's & getting pissed when Valve released "Net-Code" which gave the lowly modem-users inititive, that is the Server would favor "ties" when the HPB shot a LPB. It became wildly popular because the people who were pwning in servers because of their high-bandwidth lost a lot of their advantage and were pissed off. The myth became legend. But it was never true.)

This comment was edited on Nov 29, 17:24.

No it doesn't, or rather it shouldn't. Player hit detection means jack in any commercial game engine i've worked with. it doesn't effect the networking much. Server typically has authority over hit detection for inflicting damage, and clients do it client side for visual effect.

It should be expected that that animations would be kept in sync enough where it's accurate enough visually. It must have been stupidly implemented for that to be one of the limiting factors of their networking. Can't say I'd be surprised if that were the case though.

You have some source to this? Or are you assuming that's the reason they removed it? I'd be more inclined to remove it because its completely unnecessary, and likely stupidly expensive for what it provides.

This comment was edited on Nov 29, 16:27.

and per poly hit detection means jack shit to the networking aspect.

Sure it does. That is why Raven removed it for Q4, as it allowed them to go from 4 players to 8 and use the same resources.

Live Chat with other BluesNews forum members (beta; GreaseMonkey script):
http://www.ualr.edu/szsullivan/scripts_/BluesChat.user.js

Alternating Logo (GreaseMonkey script):
http://www.ualr.edu/szsullivan/scripts_/BluesNewslogo.user.js

and per poly hit detection means jack shit to the networking aspect.

per poly hit detection doesnt mean shit when nobody is playing the game online

________________________
music from space captain:
http://www.soundclick.com/bands/6/errantways_music.htm
http://www.soundclick.com/bands/4/invisibleacropolis_music.htm

Here we get Doom 3 out of the box with 8 player support? This was in 2005 right?

The difference with Doom 3 and all the others you listed is that Doom3 does per-polygon hit detection. This puts it in a whole different class then those, and trying to compare the number of players allowed is not accurate. To detect 4 players in a MP session in Doom3 requires a few factors more of processing power and network data to achieve to detect a hit or miss.

I think the idea of per-polygon hit detection is good; it was however too far ahead of its time. If they went with traditional hit boxes MP would have been much nicer (if technically less accurate and inferior).

This mod is a server side only mod that should make network play smoother in these games, right? The author doesn't make much clear on his site...

No, it is a user side only mod.


Live Chat with other BluesNews forum members (beta; GreaseMonkey script):
http://www.ualr.edu/szsullivan/scripts_/BluesChat.user.js

Alternating Logo (GreaseMonkey script):
http://www.ualr.edu/szsullivan/scripts_/BluesNewslogo.user.js

Didn't anyone play Counter-Strike?

Nah. Was pretty much a niche thing, and interest fizzled pretty quick



-----
It may be that one day a young man will adore a Pinata.

They are being very quiet lately too.

What is really strange is why don't these game companies dissect Tribes 1 and 2 netcode and figure out how they were able to allow 16/16 and 32/32 games with vehicles in full 3d with minimal to no lag? And this was how many years ago?

Here we get Doom 3 out of the box with 8 player support? This was in 2005 right?

It's like the Battlefield series and their front end server list, which is utter crap, same goes for Counter Strikes server list, total bullshit. Load up Tribes 2 and you had a complete community, favorites, history, buddy lists and a website maker all right there, in the game, out of the box.

Guess these guys (ID and others) are too busy playing their own shit to realize it has all been done before, WAY before, and imitation is the greatest form of flatery.

Why reinvent the wheel every time?


Kind of like today's music:
Tribes 1 and 2 = Black Sabbath/Heavy Metal

Doom 3, Q4 = Korn/Nu-Metal when there really is nothing "nu" about it and it sucks.