View previous topic :: View next topic |
Author |
Message |
borgrel Advanced

Joined: 20 Aug 2008 Posts: 879
|
Posted: Fri Feb 22, 2013 2:28 pm Post subject: Open Source: to-hit correction |
|
|
In the beginning there was to-hit. There was a magical tree: Guaranteed, Almost Guaranteed, Easy, Above Average, Average, Below Average, Difficult, Near Impossible ...... 8 to-hit levels.
And with this tree there came to-hit modifiers. Uncontrolled speed movement -2, Out of optimal rng energy -2, Elite gunner +3, WM -3, etc.
And things were good.
Most modifiers between 1 and 4 except for thumper. It took 3-5 modifiers to 'cross' the to-hit tree.
And then there came the reckoning, the bonuses were expanded to allow for gradiated modifiers. And the idea was good.
Out of optimal rng energy -8, Missiles +4, Uncontrolled movement -8, etc.
And things were NOT good.
Most modifiers capped close to 8. It takes 1 modifier to cross the entire to-hit tree. A SINGLE MODIFIER BREAKS THE ENTIRE TREE.
The idea of rolling modifiers is exceptionally good, unfortunately the implementation was slightly lacking.
I suggest the to-hit tree be expanded from 8 to-hit levels to a number that again requires 3-5 modifiers to cross the tree rather than 1. Maybe 25 or 30 to-hit levels. _________________ Birth is pain;
Death is pain;
Change is painful;
Embrace pain,
Embrace life!! |
|
Back to top |
|
 |
maXDooom Advanced

Joined: 07 Dec 2011 Posts: 757 Location: in perverted places
|
Posted: Fri Feb 22, 2013 3:21 pm Post subject: |
|
|
from 1% to 100% would be better, modifiers would be like 30%+ 20%-
or so _________________
Walterohdim wrote: |
I would be so mad I would Tomo Rage, and probably lose my mod ship and council seat.
Walterohdim |
|
|
Back to top |
|
 |
Walterohdim Site Admin

Joined: 15 Jun 2006 Posts: 1629 Location: Somewhere with lots of doors
|
Posted: Fri Feb 22, 2013 3:30 pm Post subject: |
|
|
Right now it is some sort of funky logarithm. I have long advacated going to something more what Maxdoom is talking about.
Walterohdim _________________ Wannabee historian of Tinywarz. Through the building of myth we will build community. M. Scott Peck
 |
|
Back to top |
|
 |
Vanoi Site Admin

Joined: 25 Aug 2005 Posts: 1403 Location: Avoiding You (The Void)
|
Posted: Fri Feb 22, 2013 4:09 pm Post subject: |
|
|
It's really not that hard to understand... problem really is there are too many buff/debuffs in the game that 'cause the to-hit to fly from one end to the other in a matter of a single turn.
My solution would just be to go into everything and reduce their effect.
1 (3) To-hit becomes <1 (.6 or .3) To-hit and so on...
Especially the later stuff like SW, XP and CD they have To-hit/To-be-hit values flying everywhere. _________________ The person who will go to great lengths to fight boredom...
 |
|
Back to top |
|
 |
borgrel Advanced

Joined: 20 Aug 2008 Posts: 879
|
Posted: Sat Feb 23, 2013 4:48 am Post subject: |
|
|
if u dont like the expand 2-hit tree option, another option is a double sliding value system.
create 2 values, source and target, which start at 50% each and all to hit modifiers only modify one of them.
eg.
out of optimal energy modifies source by -80% making the result 0%(source) to 50%(destination) to hit, so thats still an average 25% to hit
if the other player stacks then target to hit is +30%, then to hit is 0%(source) and 80%(target) an average of 40% to hit
if the modifiers are carefully grouped then there sould be NO way for BOTH values to be 0% and neither should ever b 100%
even if a player uses missiles at 10rng with DSM and AoA and atl (which modifies source by +140% for example) if the other player modifies target to-hit with 10 move speed, 1 lvl harder to hit and WM which modifies target to-hit by -60% thats still 0%(target) to 100%(source) thats still 50% average to-hit.
of course if the 2nd player stacks, doesnt move, uses no cmds, etc. the target to-hit would be +40% making to-hit 90%(target) to 100%(source) making average to-hit 95%
with an implementation somehow similar to this idea **BOTH** players have **LIMITED** control of to-hit making no1 able to guarantee perfect unstoppable hits and also making no1 able to guarantee perfect unhittable dodges.
It also ensures that BOTH players actions/choices/setups are never completely meaningless (like a max dodge setup vs units with +20 to-hit)
This also opens up interesting cost/benefit mods (like for example AoA increases source tohit by +80% but also penalizes target tohit by +20% when this unit is under fire) (makes ur chance to hit better at the cost of increasing ur targets ability to hit u also) _________________ Birth is pain;
Death is pain;
Change is painful;
Embrace pain,
Embrace life!! |
|
Back to top |
|
 |
Walterohdim Site Admin

Joined: 15 Jun 2006 Posts: 1629 Location: Somewhere with lots of doors
|
Posted: Sat Feb 23, 2013 9:28 am Post subject: |
|
|
I had coffee before I read it, then took a hot shower to contemplate it. After all that caffiene and hot water, I think I understand it. I admit it is still pretty complicated sounding, but all the coffee and hot water in the world won't allow me to understand the current system, so I think this might be a workable reform.
Walterohdim _________________ Wannabee historian of Tinywarz. Through the building of myth we will build community. M. Scott Peck
 |
|
Back to top |
|
 |
Frizz Site Admin

Joined: 10 Jun 2005 Posts: 3814
|
Posted: Sat Feb 23, 2013 2:40 pm Post subject: |
|
|
There's some commented out code in process_turn that will show you how much a to-hit number was modified by.
I'd be interested in seeing how "weighted" to-hit modifications work (e.g. if to-hit mods become less effective as more of them are added ... this would require queueing the bonuses and penalties, then decreasing their impact as the went beyond certain amounts ... 3-ish gunnery levels?).
_f _________________
 |
|
Back to top |
|
 |
maXDooom Advanced

Joined: 07 Dec 2011 Posts: 757 Location: in perverted places
|
Posted: Sun Feb 24, 2013 9:36 am Post subject: |
|
|
Frizz wrote: | There's some commented out code in process_turn that will show you how much a to-hit number was modified by.
I'd be interested in seeing how "weighted" to-hit modifications work (e.g. if to-hit mods become less effective as more of them are added ... this would require queueing the bonuses and penalties, then decreasing their impact as the went beyond certain amounts ... 3-ish gunnery levels?).
_f |
i have no idea what frizz just said but i support it  _________________
Walterohdim wrote: |
I would be so mad I would Tomo Rage, and probably lose my mod ship and council seat.
Walterohdim |
|
|
Back to top |
|
 |
Walterohdim Site Admin

Joined: 15 Jun 2006 Posts: 1629 Location: Somewhere with lots of doors
|
Posted: Sun Feb 24, 2013 11:40 am Post subject: |
|
|
Frizz wrote: | There's some commented out code in process_turn that will show you how much a to-hit number was modified by.
I'd be interested in seeing how "weighted" to-hit modifications work (e.g. if to-hit mods become less effective as more of them are added ... this would require queueing the bonuses and penalties, then decreasing their impact as the went beyond certain amounts ... 3-ish gunnery levels?).
_f |
Weighted averages would be way better then what we have now. That way getting a unit all one way or another,(were they end up broken) would be much more difficult. Every standard Deveation away from Average it should get twice as hard to modify away from. I know this would reduce damage a lot as many times it is very easy in this game to get to guaranteed, but I think it would make targetting buffs over damage buffs a lot more interesting.
Walterohdim _________________ Wannabee historian of Tinywarz. Through the building of myth we will build community. M. Scott Peck
 |
|
Back to top |
|
 |
borgrel Advanced

Joined: 20 Aug 2008 Posts: 879
|
Posted: Sun Feb 24, 2013 1:08 pm Post subject: |
|
|
nice idea walt
how about:
if u define x as all positive to hit modifiers and y as all negative to hit modifiers:
int to-hit = trunc(log(base2)(x)) - trunc(log(base2)(y));
even with x = 32, to-hit will be 5 which is 1 over guarenteed
even with x = 32 and y = 2 (ea) to-hit will be 3
and even with x = 0 and y = 32 to-hit still wont excede -5 (just below almost impossible) _________________ Birth is pain;
Death is pain;
Change is painful;
Embrace pain,
Embrace life!! |
|
Back to top |
|
 |
rafkory Intermediate

Joined: 01 Jan 2007 Posts: 370 Location: PGF
|
Posted: Mon Feb 25, 2013 1:27 am Post subject: |
|
|
I have to say that I do like the original Frizz's idea of changing the to-hit by the standard deviation, not by percent. Reduction from 99 to 89 percent is much less important then from 11 to 1 percent. However, we should still display the resulting chance in percent, as most people understands percents much better then the standard deviation.
The numbers behind it (should the thumper reduce the to-hit by 3 or 10 STD) is another question and I think the game would benefit some tweeks here.
Raf _________________ Mammals are the best. Other animals suck.
--------------------------
The feedback I got:
Neutral: Don't know him, but has proven himself to be a complete idiot on forums. Donk+ 2010-07-27 12:31:03 |
|
Back to top |
|
 |
maXDooom Advanced

Joined: 07 Dec 2011 Posts: 757 Location: in perverted places
|
Posted: Mon Feb 25, 2013 5:07 am Post subject: |
|
|
borgrel wrote: | nice idea walt
how about:
if u define x as all positive to hit modifiers and y as all negative to hit modifiers:
int to-hit = trunc(log(base2)(x)) - trunc(log(base2)(y));
even with x = 32, to-hit will be 5 which is 1 over guarenteed
even with x = 32 and y = 2 (ea) to-hit will be 3
and even with x = 0 and y = 32 to-hit still wont excede -5 (just below almost impossible) |
so you're saying lets **** up dodge units? _________________
Walterohdim wrote: |
I would be so mad I would Tomo Rage, and probably lose my mod ship and council seat.
Walterohdim |
|
|
Back to top |
|
 |
borgrel Advanced

Joined: 20 Aug 2008 Posts: 879
|
Posted: Mon Feb 25, 2013 5:14 am Post subject: |
|
|
raf, the idea of to-hit following a bell curve *IS* a good idea, unfortunately, for that to be the case, the 'normal' to-hit needs to in the middle
as the game is now the 'normal' to-hit is either guarenteed or impossible. practically NEVER in between. if u graph the to-hits in this game atm the graph will look like this |\__/| and nothing at all like this /"""\.
yes, gr8 idea, but not at all the current implementation.
here is a diagram of the results of the suggested to-hit calculation for all values of to-hit from (0,32) negative modifiers and (0,32) positive modifiers plotted 2gether. As you can see a much greater % of possible results actually DO sit in the middle.
(this is a rerepresentation of the plotted values from http://i44.photobucket.com/albums/f49/borgrel/to-hit-color.png) _________________ Birth is pain;
Death is pain;
Change is painful;
Embrace pain,
Embrace life!! |
|
Back to top |
|
 |
Thunder_lt Advanced

Joined: 15 Jun 2006 Posts: 1218
|
Posted: Mon Feb 25, 2013 5:57 am Post subject: |
|
|
So you want to make things hit all the time? and miss just in few cases? thats really bad idea ... _________________
 |
|
Back to top |
|
 |
rafkory Intermediate

Joined: 01 Jan 2007 Posts: 370 Location: PGF
|
Posted: Mon Feb 25, 2013 6:15 am Post subject: |
|
|
Well, in fact this should be just one-tailed curve, but a triangle may be some kind of approximation.
The issue is to tweek the numbers, not to change the method. In fact, multiplying is not much more expensive then adding, so I'm not sure it it is worth re-doing all the to-hit module.
Raf _________________ Mammals are the best. Other animals suck.
--------------------------
The feedback I got:
Neutral: Don't know him, but has proven himself to be a complete idiot on forums. Donk+ 2010-07-27 12:31:03 |
|
Back to top |
|
 |
|