Re: Giving up? I think so
Quote:
Originally posted by SM
Rather than limiting the battles please implement generic fallback meshes/textures (use vertex shading to get the general colour rather than textures, most people are wearing black/white/red/blue anyway) then trickle the textures from mem to GFX card as you get the oppertunity, and cache everything you can in mem before you log.
Please if anyone in FC thinks I haven't thought this through I'd love to hear their thoughts. I don't want to argue or be akward, I'd just love to hear their side.
I can agree with that :)
In the words of the SWG guy from their boards.
We want the game to have a descent frame-rate no matter what you are doing, I don't care if you all turn into a bunch of little green stick figures, you won't have low framerate!
LOL :)
a small suggestion - continued
I'm not interested in getting into flame wars, so I won't try to defend or argue a point.
However, I did neglect to mention a couple more points about attacker limits and defensive bias.
SIEGES & HUMAN WAVES: In a defensive bias situation, attackers with good tactics still have options. If they place the tower under siege (i.e., prevent more defenders from getting to it), then they can use superior numbers to send in wave after wave of attackers. Each wave wears down the defenders, causing more and more casualties until the attack succeeds. Costly, but effective, as the Chinese proved in the Korean War.
LOGOUT-LOGIN PREVENTIONS: One extremely nasty defensive tactic that should be prohibited is defenders logging out right around the tower. Now that many players have "alts" at level 200, a guild could have dozens and dozens of level-200 alts logged out right around the tower, ready to log-in at the first sign of trouble and defend the tower. This trick has already been used to great effect in DAoC, making most relics uncapturable in prime time (when plenty of those top-level alts can be mustered). It's not uncommon for an attack of 200-300 people see a couple hundred (yikes!) defenders material inside their fort within minutes of the attack starting, as people log-in on their alts, which were logged out inside the fort.
The needed prohibition is that players in a guild which owns a tower cannot log out within X distance of that tower. If you can't prohibit this absolutely, then when they do log-out, their logout position can be reset to their last insurance terminal location. :)
- Anupa's guiding spirit
Re: Giving up? I think so
Quote:
Originally posted by SM
[B]
1. Client issues move order
2. Server initiates zone, removes player from playfield
3. Server issues rez manager command to client: load new playfield + all meshes in area (switching to more generic meshes/lower rez textures if overcrowded)
4. Client thrashes (Screen still blank, or maybe playing a pretty and simple animation)
5. Server prepair to add player to new playfield but doesn't do it yet.
6. Client issues "all done" order
7. Server adds player to playfield
Wouldn't help very much with the streaming of ressource in zone.
The client doesn't knows in advance the list of all of the dynels (ie dynamic object, like players) for the whole zone.
Only the dynel infos for the immediate surrounding of the player are sent, otherwise it would probably take too much bandwidth, and would pose a security problem where one could, for instance, write a small prog capturing the network packets and displaying the complete mission map with all the mobs and the pickup item.
I do think that the streaming from disk takes a lot of time. But I wonder if they use a separate thread to manage the ressource streaming... If not, then the whole game is bound to the randomness of disk accesses, and probably sometime need to wait longer than usual. If that was done by a separate thread with a lower priority, the engine could continue to update and render the game, even whle waiting for some particularly long loading to be done... Althought most probably the ressource manager thread would have a hard time getting some cpu time.
Just some thoughts... I don't pretend to teach the devs their job. It's just that the game seems to have to wait sometimes for a data to be loaded for long times (with lots of HS thrashing - I get it a lot because the partition where I have AO need a defrag), and seems to be completly blocked until the data has been loaded.
Re: Re: Giving up? I think so
Quote:
Originally posted by MORB
Wouldn't help very much with the streaming of ressource in zone.
The client doesn't knows in advance the list of all of the dynels (ie dynamic object, like players) for the whole zone.
Only the dynel infos for the immediate surrounding of the player are sent,
All True, however the terrain objects that the player would "see" can be loaded before arrival (stopping post-whompa building popping) minimising the post-zone thrash. If the meshes are done properly you should need 1 mesh for each breed (not each body type) regardless of wether they are wearing cloaks or not and a few extra meshes for helmets/weapons/tank armour. Have a few generic textures in memory (use vertex shaders to colour the armour) or just straight vertex colour. Then monitor free CPU and trickle in "important" textures and meshes from disk to mem and from mem to GFX card as and when you can.
On a system with a reasonable (or unreasonable) amount of memory there should be nor resource based reason why there cant be hundreds of meshes wandering about (or sliding about a-la Quake2)
AO does look pretty, I'll give it that, but do you really want to maintain pretty at the expense of gameplay?
There are still other things that you can do.
I have to say that the solution is the lesser of 2 evils. Notum wars probly will suck if this issue was left unchecked.
There are other avenues that could help make this problem less of a blatent issue, but they would require planning and crap like that. indeed im writing the post because of the problems you descriped (i keep crashing so now im here at the forums to rant).
I'm sure that my couple hours a day are spent doing the same thing that most other do. I log on, hang out in omni-ent and chat with people or shop, I head to nlc to find a team, and we all go to bs to do a mission, and then if the ocassion arizes, we all head off to camalot to ***** and complain about the lag. All punctuated with a crash here and there.
One problem is definately the congestion. There are zones that no doubt developers spent many many hours developing that are practically deserted and some that are over crowded and causing problems.
I think that one of the reasons that this congestion occurs is because traveling in the game is just too damn convienent. the other is the lack of anything more proffitable to do.
Amendment #1 that I would suggest would be to try and devide the player base over the entire world. You have to turn the players from the deeply beaten paths. When an experienced player starts a new character, he pretty much goes to places pre-ordained by the playerbase experiences. This would not be so bad except for the fact that there are places where the people tend to pile up for many levels (like 20k, omni-ent, newland, tir). Where they sit around looking for a team to do a mission as quickly as possible, do as little thinking as possible, and level up as quick as possible, with the smallest possible risk. And then they'res the dayly mess of camalot, where the player base has piled up the worst. This is my phase 1 of what I'm trying to convey to help the situation. Make it more profitable to travel and try and force the people at different levels to be in different places in order to accomplish their goals. This would mean designing certain area's of a game for a certain level range.
But before you would do that you would have to consider this.
No one is going to agree with me on this one but:
when I outline my tipical game session that I share with so many of the same faces every time, I realize that the 3-4 places i usually go are very very far away, but they can be reached in less than a minute. The greatest danger in traveling in this game is crashing. I don't know why moving around this big world needs to be so damn easy. They wompas and grid are convienent but they make things just too efficient. You can get a yalm at lvl 20 or less and fly the entire world with no real danger. Why? These things are nifty but they serve to flame the lazy mission frenzy and ultimately the player pile up.
Try these 3 things
1. Limit the whompa network, you can go almost anywhere at any level right now. Make them less convenient
2. Limit your ability to use the grid like 12 hours or so or make it cost a bunch of money to use it.
3. Yalms......just get rid of them or have them time limited and will vanish in 2 hours.
If you decided to limit the transportation, you could easily control where the people should be at what level and spread them out. Make it impossible for a level 2-50 to get to camalot, he would surely perish by some nasty mob because he couldnt wompa around or fly there with no danger.
Also if I'm down in DAV doing something profitable besides bird watching and site seeing and I recieve a "come to camalot clanners have the box room!" theres no damn way I could make it there in time. (of course to keep people from just piling up in camalot, you'd have to make other big similar places for people to hang out at, make them choose which area they are gonna try and defend against the other side/loot they want the most)
Well thats my solution for lag reduction.
*limit transportation
*design playfields to be more accomidating at certain levels to try and get the players seperated out.
Now this combined with your idea of limiting the amount of players in a zone I think would help alot. Even though its a lot of work I think that idea is something to strive for.
There is one small flaw with these things and that is the leveling out of the player base at a higher level, I guess the solution to that would be to make many places for high levels to hang out at.
those are my ideas, now wheres my paycheck
Vash.