PDA

View Full Version : Training, is this possible?



Hog
18-02-2012, 02:50
Hello everyone. I'd like to improve my accuracy and headshot numbers. I play on a few servers, but I'd like to try something by myself with using the bobot mod and hosting the game on my pc. I only have a couple ideas so far.

1. Only headshots count as damage.
2. No damage to me, and also no damage to the enemy bots.
3. I'd like the bots to mimic human strafing.

Is this possible? My goal is to increase my headshots, and learn to deal with screenshake better. I'd like to learn to track the enemy better without taking damage, and without the enemy bot dying. I still want the hits to count, and I still want my screen to shake, I just want the HP number to not change.

Thanks for any input.

testforecho
18-02-2012, 11:24
Hmm I have been working recently at a computer controlled enemy for etpro (based on some short code by wolfcam author) which does almost exactly what you say, but it's still beta, and since it's gpl I should give you the source as well :D, not to mention that I'd have to port it from Linux to Windows if you use win, but that wouldn't be hard. It should probably work for other mods as well, but I tested it only with etpro.

Since I wanted to play etpro again, I wanted to train with real etpro hitboxes and not bobots or similar sh*t, and my enemy strafes left-right and crouches to control spread, has a very good accuracy for even harder training, and also has an option for unlimited hp, or to give you unlimited hp. I use that together with an udp proxy to delay the packets to the server, so I can train with the same ping I have ingame, or even test how I hit with lower/higher pings or unsteady connections. (I know there is cl_packetdelay and stuff but I wanted a real ping not a simulated one)

Eventually I realized that the best thing is playing in real servers against real players. They might not be as highskilled as that computer one, and the training might not be as intensive, but real players are a better training because they are smart. And with all the tests I did I realized that etpro hitboxes are good with any ping, you just have to aim at the head, you have to blame only random bullets so there is nothing to really train, aim at the head and that's it. I tested it with slow timescales, antilag debug hitboxes, g_debugbullets and stuff. Etpro team did a very good job despite what all the stupid whiners around say, hitboxes are quite good, and antilag works, in these tests I had the same acc with pings ranging from 0 to 300+ and steady/unsteady pings.

I don't know if I will ever release this 'training opponent' because it's not much user friendly (and not very useful after all), but you could come close to something like that by starting a server with /devmap and doing /give health 10000, this way you have 10000 hp, then /give ammo (you can bind for example mouse1 "+attack;give ammo") if you do /god you will have no screenshakes, and that's not what you want

testforecho
18-02-2012, 11:47
Another thing, you say: 'Only headshots count as damage' now that is not a good idea, I remember when I played in some hs-only servers, and it's *not* the same as the real thing, because if you can only give hs you won't get any screenshake for bodyshots.

The best thing you can do is playing in some server with good players and lots of fights, then enable hitsounds only for headshots (b_hitsounds 2 on etpro, cg_hitsounds 2 on jaymod, I don't remember for other mods, but it should be possible) and really focus on hitting the head. If you kill someone, but without a hs, you should consider it as a 'fail'

If you have a friend and a server of your own (or you can host it in your pc) you can do that training I mentioned before (1on1 with /give health 10000) but again I think you should play in real servers, and not against bots.

gl with your training

Hog
18-02-2012, 13:26
Thank you testforecho for responding. I agree with your statement that playing against humans is best. I mostly play NQ, with a little silent and jaymod mixed in, a touch of etpro also. From NQ, my accuracy usually goes up 10%-12% when I play silent and jaymod. Headshot numbers increase also. I feel NQ is harder, so it is my preference to play with my clanmates and visitors. I can lead in headshot numbers, and if my accuracy is 30%, then I've had a solid game. But there are certain players that I can't beat constistently. I'll lead them numbers wise, but in battle they beat me 75% of the time. Sometimes my xhair will be directly on, or they are doing the medic-dance (strafe) through my dot, and I won't register a hit. It gets frustrating, so this is what brings me to the idea of training.

Sometimes I visit the F|A jaymod server, and hold my ground and excel in the map. I can finish with 100+ headshots (can be attributed to so many players there to shoot at). But there seems to always be "that guy" there owning everyone. I've see a guy with over 300 headshots recently, very humbling to see those stats. I've tried to play opposite of these players, but I always lose.

I play at my monitor's native resolution of 1680x1050 60hz. I recently lowered my cg_fov to 90. Could these settings be affecting my accuracy and headshots? Would a lower resolution make the player models appear bigger? I've changed the settings before, but the quality goes down, so I don't really play very long with the lower resolution. Just want to eliminate hardware and software factors before totally blaming the human factor.

daredevil
18-02-2012, 13:59
Getting headshots on jaymod is easy compare to NQ and silent.

Try r_mode 4 or 6 and use fov 100 or 110... so that u can see more compare to non config users. Might want to also lower m_pitch so that u mouse remains straight on Y axis compare to X axis.

testforecho
18-02-2012, 14:33
As regards hardware and software factors, this is what I can say:

In those tests I did on etpro with different simulated network conditions, I got the very same acc with every ping, slightly more hs with a low (around 48) ping, but not so many to make it a significant advantage, it could even be due to randomness, even though I tested it on a custom etpro server with weapon spread removed.
So ping is not an issue for your accuracy, but since hitting first in this game is very important, it could be a problem if your ping is high.

The game is about angles and coordinates, not pixels, so if you change screen resolution or you lower your cg_fov nothing will change. With a lower fov you see enemies better in long range but that's all.

To play in the best possible conditions you should aim at having a low, stable ping and high, stable fps (100 fps with 100 cl_maxpackets is the smoothest, but you will be smooth for your opponents as well, maybe 125fps is better). Make sure your monitor doesn't have input lag.


When you don't hit even if you aim on target, it can be for many reasons. Assuming that you are *really* aiming on him, and you are not confused because of screenshakes, etc. you might not hit because:
1. random bullets. Bullets have a random spread, try shooting against a wall and see. If this spread becomes too high it's more than likely that you won't hit in long range, or that you'll miss a headshot. Learn to control it with crouch and aim quickly at the target.
2. prediction errors. These happen when your opponent hits you (or shoves you, when in close range). They are a problem mostly if your enemy is above you, on an even ground they don't make you miss much. Having a high ping is bad, because your prediction errors are bigger and an enemy with low ping will frag you, all else being equal (same skill, etc.)
3. crouching. When a player crouches his bodybox is lower, and if you try to aim at the head and he suddenly crouches or taps 'c' you might miss some bullets. If he taps that button his animation could appear almost as standing, but in reality his bodybox is already low

Screenshakes are just a 'visual' thing, the only real effect is bullets pushing you back, your aim direction is not affected by screenshakes, only your position is. Make sure you hit first, when you can still see where you are aiming. Take cover if you see you are obviously losing a fight, and use all the infos you have to guess if you are aiming at the right height (hitsounds, and that 'dust' thingy that appears on the enemy when you hit him). A low m_pitch could help you, as suggested, but it's a bit overrated. Personally I use default 0.022

That's all I can suggest, I am lowskill (used to be low+ maybe) so a better player could give you more info. My acc on etpro pubs was around 38-39 when I used to play it (not much gibbing or preshoots, it seems :p), on jaymod I get a higher acc. I almost never play nq, silent or etpub but the things I told you are more or less valid for every mod.

Hog
19-02-2012, 03:21
Getting headshots on jaymod is easy compare to NQ and silent.

Try r_mode 4 or 6 and use fov 100 or 110... so that u can see more compare to non config users. Might want to also lower m_pitch so that u mouse remains straight on Y axis compare to X axis.

I'll play around with those r_modes. I've gone from cg_fov 120 to cg_fov 100.3 over a year or so, and recently took it down to 90. I feel like my movement is better with a higher cg_fov, still on the fence about accuracy as some days I'm better than others. I try to make adjustments and play for several weeks before final judgement, but I've been considering raising the fov again (at least for movement sake). I believe my m_pitch is .01501, at work right now so no access to my cfg to check. Thanks for your input. :)

Hog
19-02-2012, 03:51
Testforecho,

I think I'm on the right track then, I'm always searching the net about W:ET settings and I have mine set to: cl_maxpackets 100 and com_maxfps 125. Your 3 points: 1. Aware of spread, I'm constantly crouching. I rely on it heavily.

2. I'll search more on prediction errors. After FINALLY getting cable internet, my ping to my clan's server is usually 48. When I encounter someone on higher ground, I usually know immediately who will win. This is one of the circumstances I was referring to when my xhair will be on them and hit nothing.

3. New to me also, didn't know about the animation and bodybox mismatch. I've seen the hitbox models on the net somewhere, and it shows the position of the boxes in relation to the player model. I try to keep those images in mind while playing.

I really appreciate you taking time for the write up. I've only talked to a couple of people over the last couple of years about the game. Scarhand gave me some good advice on the screenshake, he can really adjust to it and take out the opposing player with ease it seems. My next upgrade will be a new monitor at 120hz, not sure when that will be, got bills to pay :'(

testforecho
19-02-2012, 12:01
2. I'll search more on prediction errors.
I don't know if you'll find much because most etplayers are not modders and don't know much about this stuff. Many of the infos you find around about these subjects (antilag, hitboxes, hit detection in general) are partially or sometimes totally wrong.
Just keep in mind that when somebody hits you you are pushed back in the server, before you can see it yourself. When you shoot, the server will take that position into account (the position you have in the server), when it has to determine if you hit or not. And if there is a prediction error that is *not* the position where you believe you are. Now if you say your ping is 48, the difference shouldn't be very high. The angle or direction of your aim is not affected by screenshakes so you will still aim in the same direction, but if you are pushed back the position from where you are shooting might change slightly.

There is a command to show prediction errors, it's cg_showmiss. If you set it to 1, it will write many error messages in your game console. Among these you should check for 'prediction miss'. That number tells how many game units apart you were, from where the server thought you were (sorry I couldn't explain better :p) ). Headbox is 12 units wide, Bodybox is 36. So you can have an idea if that distance could make you miss a headshot or if it's so tiny that you would still hit the head. Of course if your enemy is on higher ground, moving back would result in aiming higher. But don't worry, your ping is 48, you are lucky :)


I've seen the hitbox models on the net somewhere
Most of those models you find in those websites are old (etmain-style hitboxes) so the bodybox is reliable (for stand and crouch), but the headbox is not like that anymore in all mods (etpro was first, then all copied their new headbox, I think Zinx gave the source code for etpro b_realhead) now in all mods the headbox is exactly on the head. There are commands to show the hitboxes when you play, but you must have rcon or host a server with that config, and mostly those hitboxes are a bit delayed so they give you only an idea. The most important thing you need to know (imho) is the height of crouching bodybox, the headbox is always on the head.

BossHK
20-02-2012, 18:13
Also make sure your hardware(esp. mouse) is 'calibrated' well. To make is short: http://fearless-assassins.com/forum/topic/23840-wet-aiming-tutorial-for-wet-by-bosshk/

Hog
22-02-2012, 05:43
Also make sure your hardware(esp. mouse) is 'calibrated' well. To make is short: http://fearless-assassins.com/forum/topic/23840-wet-aiming-tutorial-for-wet-by-bosshk/

Thanks Boss, I'll check that link when I get back home. At work now and they got it blocked :| I've read your tutorials before and spoke with you on youtube, you told me to use IRC, but I haven't. (my name wasn't hog either) In your latest, you mentioned playing on someone else's system and the hardware really holding you back. I feel like I've only slightly improved over the past year, and I play regularly. So I have hardware concerns too. My mouse is MX518. I play at 800dpi @ sensitivity 2. I've read about the negative acceleration over 400 or so, but I can't really tell. I still read players using 1000+ DPI.

testforecho
22-02-2012, 14:45
I've read about the negative acceleration over 400 or so, but I can't really tell. I still read players using 1000+ DPI.

It depends on your mouse sens, resolution (r_mode) and fps. If you play with high sens and high resolution, and you get 125fps you won't probably have any negative accel issues. I play at 1680x1050 125fps and 2000 dpi and I don't have negative accel because my mouse sens is high.

I think 800dpi sould be safe, but you can test yourself if you have negative accel or not. Aim at some corner, with your mouse on the edge of your mousepad, move away really fast then go back slowly. If you are aiming at the same corner as before then you have no accel.

BossHK
23-02-2012, 06:38
It depends on your mouse sens, resolution (r_mode) and fps. If you play with high sens and high resolution, and you get 125fps you won't probably have any negative accel issues. I play at 1680x1050 125fps and 2000 dpi and I don't have negative accel because my mouse sens is high.

I think 800dpi sould be safe, but you can test yourself if you have negative accel or not. Aim at some corner, with your mouse on the edge of your mousepad, move away really fast then go back slowly. If you are aiming at the same corner as before then you have no accel.

Actually you are in danger to have negative acc because of your high dpi(assuming that your windows sens is at 6/11 aka the mouse input is 1:1).

Btw Hog: In my tutorial there is this http://phoon.us/mouse/ link, but nowdays the site is down permanently, here is correct link: http://www.funender.com/quake/mouse/index.html
I'm gonna edit it to tutorial soon.

Hog
23-02-2012, 09:45
Played a little on F|A silent, wasn't pretty. I got a lot of learning to do. Scenario: I crouch→strike first on enemy→he is crouched, and returns fire→my screen shakes→I continue to shoot and hear my hitsounds→I get defeated :( Anyway, I appreciate yall's input so far. I'll continue reading and learning.

testforecho
23-02-2012, 13:07
Actually you are in danger to have negative acc because of your high dpi(assuming that your windows sens is at 6/11 aka the mouse input is 1:1).

Well, you are right, but I get neg. accel only if I make extremely fast moves with my forearm, which I normally don't when I play.

With the graph in \cl_debugMove 1 you can sort of see if you are having negative accel or not. Atm I can't test it, and I often play from Linux, so I have directinput there and no negative accel.
But as far as I remember, you can clearly see when you have neg. accel. The top part of that graph would be flat, because the extra mouse input is 'cut' at some height (based on your fps, resolution and dpi/os settings).
And to get that flat peak, I need to make really fast moves. When I make normal moves, as I would do in a fight, I never reach such high peaks in that graph. Maybe if I need to turn 180 when someone is shooting from behind, but in that case negative accel won't be such a handicap.


Played a little on F|A silent, wasn't pretty. I got a lot of learning to do. Scenario: I crouch→strike first on enemy→he is crouched, and returns fire→my screen shakes→I continue to shoot and hear my hitsounds→I get defeated :( Anyway, I appreciate yall's input so far. I'll continue reading and learning.
Hmm, first of all remember to check player pings, maybe you think you strike first but your opponent has a lower ping than yours, so he actually shoots first.

In these crouch vs crouch situations I think what counts most is not only hitting first, but also the amount of hs you make.
Some micro-movements with your keyboard to keep your player model's head in motion could be useful to make your opponent miss some hs, and at the same time it's strafe-aiming, to adjust your xhair position on him.

It's all about how much damage per second you deal, and a hs counts as 2 bodyshots. So even if you start shooting first, you can still lose, if you don't make damage fast enough. If you think you are going to win, then you can keep crouching and shooting, otherwise it's wiser to quickly stand and move a bit, then crouch again in a different position. You can judge by the amount of screenshakes you receive and your hitsounds.

khaled
24-02-2012, 22:11
exactly