Some New Footage..
the_hole
Join Date: 2004-01-03 Member: 25019Members, Constellation
Join Date: 2004-01-03 Member: 25019Members, Constellation
<div class="IPBDescription">With a little explanation</div> Well, I was playing the other day with some friends, smurfing.. when I had noticed that something was rather wrong with my NS, but not with everyone elses, lerks were bugged, but to nobody else but me... and those of you who know me, know that I complain about this problem alot. Here are 2 clips I made after I had noticed that something was really wrong. I changed my style of aiming drastically, but it turned to be for the best. I think it has to do with my rates, so if anyone can help me with those, that would be great. I know how to set them, and I'm not a new player, but for some reason these rates don't work for me, obviously.
<a href='http://www.lexoelectronics.com/upload/buggar.avi' target='_blank'>buggar.avi (2.2 megs)</a>
<a href='http://www.lexoelectronics.com/upload/buggar2.avi' target='_blank'>buggar2.avi (14.8 megs)</a>
Grep, if/when you see these, would you get on dn's ventrilo? You should have the information already, you've been on before to do the radio show with jink and myself =P. If you don't have it, PM jink on steam friends, we would like to talk with you about this.
Edit: Almost forgot! If you notice that when the shotty is fired, the spread goes into the ground, the FULL spread, but the lerk takes full damage about a second later, from ... well ... nothing..
<a href='http://www.lexoelectronics.com/upload/buggar.avi' target='_blank'>buggar.avi (2.2 megs)</a>
<a href='http://www.lexoelectronics.com/upload/buggar2.avi' target='_blank'>buggar2.avi (14.8 megs)</a>
Grep, if/when you see these, would you get on dn's ventrilo? You should have the information already, you've been on before to do the radio show with jink and myself =P. If you don't have it, PM jink on steam friends, we would like to talk with you about this.
Edit: Almost forgot! If you notice that when the shotty is fired, the spread goes into the ground, the FULL spread, but the lerk takes full damage about a second later, from ... well ... nothing..
Comments
1. what's your rate
2. what's your cl_cmdrate
3. what's your cl_updaterate
4. what's your ex_interp
5. what was your ping to the server
My update rate is low, due to what I said in a post in the interp thread.
STEAMEDHAMS, a veteren NS player from the exceptional HAM clan, which was one of, if not THE best NS clan to come as of now, had told me how to set it, not to that specific number, but until I have no milliseconds of lag (which can be shown on your netgraph 1). I had lowered it until the # of MS lag was gone.
I think the reason for me to notice the bug more than most others, is because I've never dealt with excess model lag, because my update rate (as far as I know), was correct for my system, thus causing no model lag. But with the bug, it drastically effects what I see, and what is actually happening, where as people who are used to the milliseconds of lag, actually see the models lagging to their corresponding hitbox.
A little cheesey, but it's all I can cope with at the moment.
Try setting your updaterate higher (i'd suggest 25-30) and see if anything changes.
I don't think update rate effects motion that drastically only changing it by 10 or 15, or even if it would effect motion like that at all.
Between that, your client basicly guesstimates how it should draw the skulk to present you with fluid motion. For all it knows, the graphical model could be entirely false. <b>To detect a hit, the server backlog is checked for the time of firing.</b> The server does not have backlog information on where your client may have predicted a target - it only has that backlog for where people actually where at that point in the servers time.
So yes, a low updaterate and default interp settings are extremely prone to severe incorrectness. People seem easier to shoot, because you're shooting a dumb prediction routine and not an actual person - but theres no magical force allowing you to register that as a hit. Personally I've been playing with 0.05 ex_interp and cl_updaterate 40, and it cuts down alot on the effects of randomly hitting backlog locations. It hasn't helped much with closerange bites sometimes not registering, but that could either me because 50ms per update is still not enough for highspeed melee (doubtful, but eh) or theres another issue with skulkbite.
As for the Shotgun, you could have read that very popular thread too - the Shotguns miss decals are completely random, whereas its firing pattern is <b>freaking static</b>. It always shoots in the exact same angle and pattern, which happens to be fairly unrelated to the standard crosshair. At best, a core 5 pellets are near the upper right of the stock crosshairs.
Edit :
Just got to read about "not affecting drastically". The amount of updates on a targets position per second is the most drastic effect you can possibly imagine - record a demo of yourself with updaterate 15 and watch it (if that still works). It might force 20 updates, I haven't tried. Do the same thing at 40 or higher, and watch it. And thats basicly assuming you didn't have the problem of clientside prediction - prediction, while making gameplay seem smooth, EXTREMELY heightens the problem of low update rates as far as inaccuracy goes. An old update, now erroneous, will still define the path of a target (means it'll keep moving in its old direction) - which can in the meantime be reversed, turned, stopped or what have you.
My update rate is low, due to what I said in a post in the interp thread.
STEAMEDHAMS, a veteren NS player from the exceptional HAM clan, which was one of, if not THE best NS clan to come as of now, had told me how to set it, not to that specific number, but until I have no milliseconds of lag (which can be shown on your netgraph 1). I had lowered it until the # of MS lag was gone.
I think the reason for me to notice the bug more than most others, is because I've never dealt with excess model lag, because my update rate (as far as I know), was correct for my system, thus causing no model lag. But with the bug, it drastically effects what I see, and what is actually happening, where as people who are used to the milliseconds of lag, actually see the models lagging to their corresponding hitbox.
A little cheesey, but it's all I can cope with at the moment. <!--QuoteEnd--></td></tr></table><div class='postcolor'><!--QuoteEEnd-->
your rate is too high, Max of Steam is 20000. The cl_updaterate is way too low, should be 40 or higher for cable modem users. I treat anything below 20 as an exploit, which it basicly is because your not giving the server enough information to tell the other players where you are.
Many people dont know how to set their netgraph right, so i encourage server operators to run plugins to control those settings, it will be a great benefit to all players on the server.
The ms on the netgraph means almost nothing.
rate 25000
cl_cmdrate 101
cl_updaterate 101
ex_interp 0 (steam will adjust it automatically)
lowering your updaterate will drastically reduce your ping in the scoreboard, but your connection will suck that way. just trust me and the others... i played cs for years and know the netcode and its issues quite well.
(and just dont flame because of the high settings... when he doesn't get loss and choke that way, everything is fine)
Those settings may be fine for CS (i should know i play cs a lot and use those) however in NS the servers take a LOT more load, so they are running at around 40fps in many cases, meaning the updaterate needs to be lowered otherwise the interp value you have will be incorrect and cause VERY jerky model movements.
First post there.
Theres no such thing as perfection though, unless you receive a packet for every frame you render, and no server is going to allow that (aside from you artifically rendering less frames, durr).