User panel stuff on forum
  6 posts on 1 page  1
Server Talk
2018-11-11, 14:55
Administrator
244 posts

Registered:
Sep 2015
Has removing randomness from this ever been discussed?

Currently the spread on SSG shots is displayed completely incorrectly in comparison to the server calculations. To fix this we could add an extra byte to the 'puff' particle effect notification (so they are more spread out) but if server & client are using their own random spread then it's still going to look like 'incorrect' hit/miss notifications.

I see molgrum has removed the randomness of the spread (but left it with entries for -1/1) previously in yawnmode, which people don't seem to use. Is there a good reason why the randomness was left in? It seems strange that this creates errors between client & server but hasn't been addressed, compared to (for example) the randomness of damage in a RL direct hit.
2018-11-11, 15:03
Member
131 posts

Registered:
Dec 2008
I think removing of small randomness from rl any other weapon is stupid.
Those who like to do strict calculations can play with calculator instead of Quake.
2018-11-11, 18:10
Administrator
2039 posts

Registered:
Jan 2006
I'm not sure i follow, is the SG/SSG patterns random client-side, i.e. the pellets you see, but not server-side which are used for calculation?
www.facebook.com/QuakeWorld
2018-11-11, 22:59
Administrator
244 posts

Registered:
Sep 2015
Server-side: randomised direction from the gun for each pellet
Client-side: randomised spread of particles around a point (for each pellet, grouped for hit/miss and each entity hit)

Server:
- SG is 6 pellets spread of 4%x, 4%y
- SSG is 14 pellets spread of 14% x, 8% y

Client
- gets sent 'x gunshot particle effects @ position y'
- spawns them randomly around y, sg = [-8, 8], ssg = [-12, 12]

Was looking at fixing the spread which is deceptively small in QW, was thinking of changing the message so the client could calculate the randomness from original point, but would need to share the random values somehow. Instead of making it fixed, we could store a fixed set of random numbers, pick a start point at random send that index in the message.
2018-11-11, 23:02
Administrator
244 posts

Registered:
Sep 2015
To be clear I'm not working on this as a priority, I was just trying to remember everything complained about in the past and "the ssg effect in the client bears no relation to the calculations on the server" was one of them.
2018-11-15, 20:40
Member
27 posts

Registered:
Jul 2017
That bullet spread bothered me much back in the days, and i'm very unsitisfied still)) so thank you for your efforts
I didn't noticed anything wrong playing my early NQ deathmatches, but when i first started qwcl and fired boomstic from
shaft to rocket on ztndm3 I was more than impressed that all bullets fired from long range can fit into player model (which was wrong for sure)
even discussed with my friend possibilities of firing quad boomstics from that position;-))
Than i saw that NQ drawings was much better to my opinion.. later some quake-coding-guru
(i can't clearly remember it was tonik @ quakeworld.ru forum, or not) talled that difference come from network traffic economy, and is not that important now..
and suddenly by my complain people made changes to ktx code and make it a toggle! pure magic.. but i can feel that its still not true..
i liked to fire at dm6 shaft door as soon as possible and even sooner, at the middle of the jump from rl corridor etc..so i feeled it VERY WELL..
So thank you very much for setting this question..
I like randomness of boomstics very much, but correct drawings of pellets is terrible.
IMO Boomstics must be randomized for sure - its as it works in real shooting)
(lets leave esoteric questions of molecules moving, gas spreading and Navier-Stokes equation out of disscussion ;-)
Theoretically i preferer more elliptic look of spreading pellets, BUT I for sure don't want best game ever to be modified so much..
By the way i once visualised spreading of q3 machingun - and it's pure hell - if carmack set a prime number in the code it would
have a uniform distribution, But he didn't - and it fits in diagonal lines, and dont cover all the rectangle!
//sorry for possibly bad english and wrong terms
  6 posts on 1 page  1