Warzone 2100 Forums

i don't think it's a problem with projectiles per-say. IMO, the problem is lies with players giving orders to units and breaking causality.

now, my question; are orders given to units timestamped? because if you knew the time an order was given and the velocity (just assume it's a move order in this example, for simplicity) of the unit and you should be able to interpolate to the proper position of all units after-the-fact. i do suppose this would end up with the potential for units to jump positions in a rather jarring manner.

the other option is to slow down the game speed so that the period of a game-tick is longer than the worst ping. SupCom does this (they have a rather large 500ms interlock put in so that you won't see any sort of problem if your ping is less than half a second, it also means there's up to that same length of time between when you give an order, and when it's carried out).

BulletMagnet wrote:SupCom does this (they have a rather large 500ms interlock put in so that you won't see any sort of problem if your ping is less than half a second, it also means there's up to that same length of time between when you give an order, and when it's carried out).

with a small test at http://www.speedtest.net/
that would allow me to play against anyone from Europe america south africa russia anda some of asia

ping times from;
europe are about 170~200ms
america/russia are about 200~230ms
south africa about 300~330ms
japan 400~450ms
austrailia 750~900ms

psychopompos wrote:Australia 750~900ms

looks like you won't want to be playing against me then...

psychopompos wrote:that'd be the bit where the hit/miss info isnt being transmitted accurately?

That's the bit where the hit/miss info isn't transmitted at all. Ever.

psychopompos wrote:excluding artillary vs light mobile units.
which are capable of getting out of the way due to the long distance of travel & errors in target prediction.

True.

psychopompos wrote:so work on the transmition of damage in the netcode.
maybe add an extra bit of confirmation.
1)projectiles projected impact location, target + predicted damage sent on firing.
2)impact location, target location, movement & actual damage returned on impact.
3)confirm.

...meh, too complicated. We'll see if another dev wants to do that.

psychopompos wrote:split yer lungs with blood and thunder
how about fixing the collision detection?
dice roling for hits was one of the biggest pit-falls for warzone gameplay its a viciously outdated way of doing anything outside of a turn based game.
in rts its called sabotage.

Because fixing the collision detection is too hard for our ragtag bunch of misfits developers.

psychopompos wrote:im glad you are lucky enough have an army of tanks & cyborgs outside to play with & order about, but for the rest of us computer games are the release for our urges to lead huge armies or viciously dismember enemies.

I guess the main problem is that I find dodging to be far too much micromanagement than should be in Warzone.

psychopompos wrote:i would rather have the realism of knowing i can use fast mobile units as part of a strategy rather then them get hit on the move, by mortar or howitzer fire that landed half a screen behind said unit.
as i just want to turn the game off when i see that happen, rather then having my choices in units & tactics limited by obsolete game mechanics.
its not a bit of wonder that every one depends solely on a swarm of the heaviest units they can build in those situations, because the faster lighter units that could be used for blitz/gurilla tactics get hit even if they evade.

You'll only be able to dodge with fast mobile units if you micromanage them a ton. Warzone has target prediction for all weapons (although it's pretty inaccurate for arty), remember, so just moving quickly isn't enough.

We can abstract this away by having hover units resist artillery damage. It's not even hard to fluff: Hover units have computers that move them out of the way of artillery fire, so they never get hit head-on. Instead, they just get glanced by some of the splash damage (so it doesn't do zero damage, either). But nooooo, no one likes that idea.

psychopompos wrote:i want realism in my games, thats why i play rts, if i wanted dice rolling to decide the success of my action, id play risk

Some level of realism. You also want some level of abstraction, too. You wouldn't want a war in Warzone to last years, during which you gradually develop PTSD, would you?

psychopompos wrote: maybe add an extra bit of confirmation.
1)projectiles projected impact location, target + predicted damage sent on firing.
2)impact location, target location, movement & actual damage returned on impact.
3)confirm.

This would be a case of easier said than done.

Zarel wrote:How's about we just, I dunno, revert to dice rolling for hits, instead of our current buggy collision detection?

split yer lungs with blood and thunder
how about fixing the collision detection?
dice roling for hits was one of the biggest pit-falls for warzone gameplay its a viciously outdated way of doing anything outside of a turn based game.
in rts its called sabotage.

Is it not the point that they do not have anyone to fix this? If this is the case then they should go back to the old code until someone can fix it correctly.
With all these little hacks to go around the way the projectiles work now it is just going to be a bigger and bigger nightmare to maintain.

-Kosh- wrote:Is it not the point that they do not have anyone to fix this? If this is the case then they should go back to the old code until someone can fix it correctly.

I do not think you understand just how old the old code is. It would probably be more work to go back to it now than to rewrite it all.

Per wrote:

-Kosh- wrote:Is it not the point that they do not have anyone to fix this? If this is the case then they should go back to the old code until someone can fix it correctly.

I do not think you understand just how old the old code is. It would probably be more work to go back to it now than to rewrite it all.

I looked at the svn logs to see and it is old but I do not really see an alternative unless that devurandom person comes back and fixes it. They changed too many things compared to the original source files.
It is too bad this went on so long without people testing or saying anything about it.

On one hand you had working code that was not accurate since it used the dice concept but on the other hand you now have code that works sometimes but it requires lots of subtle hacks to fix the code. Those little hacks will have to be reverted or else people will wonder why things still do not work when the code finally does work.

Zarel wrote: That's the bit where the hit/miss info isn't transmitted at all. Ever.

same thing?

zarel wrote:

psychopompos wrote:so work on the transmition of damage in the netcode.
maybe add an extra bit of confirmation.
1)projectiles projected impact location, target + predicted damage sent on firing.
2)impact location, target location, movement & actual damage returned on impact.
3)confirm.

...meh, too complicated. We'll see if another dev wants to do that.

i think im starting to understand the 'targeting/collision' bug now

Zarel wrote:Because fixing the collision detection is too hard for our ragtag bunch of misfits developers.

""?

Zarel wrote:

psychopompos wrote:im glad you are lucky enough have an army of tanks & cyborgs outside to play with & order about, but for the rest of us computer games are the release for our urges to lead huge armies or viciously dismember enemies.

I guess the main problem is that I find dodging to be far too much micromanagement than should be in Warzone.

i have no issue with the implementation of evasive/balanced/aggressive behaviors like those you suggest for hovers.

Zarel wrote:. Warzone has target prediction for all weapons (although it's pretty inaccurate for arty), remember, so just moving quickly isn't enough.

speed alone still helps for simple close to target.
also, teraign height & simple unit navigation help too even without evasive maneuvers.

Zarel wrote:We can abstract this away by having hover units resist artillery damage. It's not even hard to fluff: Hover units have computers that move them out of the way of artillery fire, so they never get hit head-on. Instead, they just get glanced by some of the splash damage (so it doesn't do zero damage, either). But nooooo, no one likes that idea.

dont see why you need a computer, irl, units under fire would evade naturally.
though i guess such a system would be interesting if the effect was activated by proximity of a CB sensor.
and i am still waiting for the day i can have THEL units to shoot arty rounds out of the sky

Zarel wrote:Some level of realism. You also want some level of abstraction, too. You wouldn't want a war in Warzone to last years, during which you gradually develop PTSD, would you?

if you can make the gfx that good, i would give it a go

-Kosh- wrote:This would be a case of easier said than done.

i would not deny that myelf.
nor would i wish to undermine what the devs actually do accomplish.
and i see movement toward accurate hit/miss detection for hits and away from dice probilities for hits as a step forward.
id would rather see that kept then have it retarded by failings elsewhere.

Dodging may not be realistic, but it is fun. The new projectile code improves on the old in other ways too. For example, inaccurate shots still kill if shot into a mass of units.

Increasing projectile speed is nothing more than a hack that avoids the real problems. I mean, collision comparison being done on an FPS level?? Yes, this project doesn't have any professional devs, but copping out never solved any problems.

zoid wrote: Yes, this project doesn't have any professional devs

How do you know this?

zoid wrote:Dodging may not be realistic, but it is fun. The new projectile code improves on the old in other ways too. For example, inaccurate shots still kill if shot into a mass of units.

Maybe it's fun for some people, but not me. :[

zoid wrote:Increasing projectile speed is nothing more than a hack that avoids the real problems. I mean, collision comparison being done on an FPS level?? Yes, this project doesn't have any professional devs, but copping out never solved any problems.

Yeah; we shouldn't be doing collision detection at all. -_-

-Kosh- wrote:

zoid wrote: Yes, this project doesn't have any professional devs

How do you know this?

Depends on your definition of "professional". It's true that we don't have any paid devs.

zoid wrote:Increasing projectile speed is nothing more than a hack that avoids the real problems. I mean, collision comparison being done on an FPS level?? Yes, this project doesn't have any professional devs, but copping out never solved any problems.

umm... FPS games have nothing on Supreme Commander (yes, i'm pimping it out again) when it comes to collision checking.

deterministic weapon firing categorically sucks.

zoid wrote:Dodging may not be realistic

depends on the dodging.
mobile units moving to avoid direct hits & taking evasive action are perfectly normal responses.

Zarel wrote:Maybe it's fun for some people, but not me. :[

why?

Zarel wrote:Yeah; we shouldn't be doing collision detection at all. -_-

one good reason?

BulletMagnet wrote:deterministic weapon firing categorically sucks.

truth.

psychopompos wrote:

Zarel wrote:Yeah; we shouldn't be doing collision detection at all. -_-

one good reason?

1. Because our netcode makes it really buggy.
2. Because our projectile code makes it really buggy.
3. Because this is an RTS, not an FPS.
4. Because you're supposed to win by having better strategy, not only by having better clicking skills.
5. Because when a weapon says "DPS 60", I want it to have a DPS of 60.
6. Because it's the equivalent of giving a skilled player more armor. Skilled players already have the advantage of skill, they don't need to take less damage as well.
7. Because it's the kind of thing a computer can do well. Cheaters will have bots to do the dodging for them, and there's nothing we can do about it.
8. Because we invented computers to do boring repetitive things like dodging for us. Why should we do the dodging for them?
9. Because, if we do let the computer do the dodging for us, we can just save ourselves the performance penalty and just have the computer roll the dice, which will also be more reliable and not have the eight flaws mentioned above.

well that comes down to the age-old "macro vs. micro" debate in strategy games.

do you reward or punish a player who takes close care of some of his units, while his opponents focus on the bigger picture? there is no wrong answer. i believe it's perfectly reasonable for a person to micro' his troops to gain a tactical advantage over a person who chooses the strategic one.

the only time this is a problem is when one option is grossly disproportionate to the other.