General > Discussion

2.5 sucks completely

<< < (12/37) > >>

H-Hour:
The explanation is more basic: the person doing the tests was not a developer or coder, just someone interested in figuring it out. See the later pages of this thread.

Quizer:

--- Quote from: Triaxx2 on February 15, 2013, 08:40:35 am ---Actually I was saying that since the hit chance displayed by the game itself is only a general estimate of the actual chance to make a shot, due to the calculation necessary for an accurate estimate being more complex than can be done in a timely manner[...]
--- End quote ---
I wonder if it could be coded in such a way that the game first gives you the estimate, and then, if you continue holding the mouse over that square, it would start calculating the real accuracy, but break off if you move the mouse somewhere else. Due to the way aiming works, you don't ever need to calculate more than one accuracy value at a time, so I don't see why it should be any more complicated than actually firing.

Is it that firing doesn't actually calculate accuracy in any way, and calculating hit chance via the combat engine would amount to running a Monte Carlo Simulation with a large number of shots to approximate the hit chance? If it's like this, I could see how it's troublesome...

Telok:

--- Quote from: Quizer on February 15, 2013, 12:42:57 pm ---Is it that firing doesn't actually calculate accuracy in any way, and calculating hit chance via the combat engine would amount to running a Monte Carlo Simulation with a large number of shots to approximate the hit chance? If it's like this, I could see how it's troublesome...

--- End quote ---

From my understanding this is essentially correct. What I believe happens when you fire is that the engine takes the soldier's accuracy and the weapon's spread value, mushes the two numbers together, sprays your ammo down range at an angle determined by RNG and the mushy spread number, and then checks the hit boxes like a first person shooter. So the shooting engine never develops or uses a %hit chance. The number reported on your targeting mouse-over is a fuzzy guesstimate from somewhere in the UI code (I think, don't quote me on that one because it's a logic guess).

For real fun:
The reported %hit has no effect on RF. Soldiers may or may not RF anywhere between 0% and 90%+
RF is more likely to happen if your chosen RF mode shoots more bullets.
RF will not happen outside the weapon's listed range (in squares).
Higher soldier stats and skills will mean more frequent and accurate RF. Expect +30% RF if you double the soldier's stats and triple the skills.
Some weapons (and it's definitely the weapons) will RF through a LoS blocking alien.

From the tests I've run a noob soldier with a sniper rifle at 50 squares will take about 50% of the RF (Aimed Shot, crouched, 23% reported hit chance) chances he is offered. And it's about a 50/50 chance as to whether he'll take the RF before or after the provoking action.

From the tests I've run a noob soldier with a machine gun at 65 squares will take about 60% of the RF (Full Auto, standing, 00% reported hit chance) chances he is offered.

I do actually know some coding and am currently going back to school to complete the degree, so if you'll tell me which module the RF code is in than I'll try to look at it.

krilain:
Thanks for this complete summary. However, I would like to point an element.

--- Quote from: Telok on February 16, 2013, 10:04:19 am ---/.../
the engine takes the soldier's accuracy and the weapon's spread value, mushes the two numbers together, sprays your ammo down range at an angle determined by RNG and the mushy spread number, and then checks the hit boxes like a first person shooter.
/.../

--- End quote ---
This is a thing that I've tried to figure out many times. Is there any hit box check, really? It would fall finally on such a thing as depicted below so that the effective hit rate would finally be given by:

* Hit_rate = Hit_base / Spread_base But this includes so much parameter in uphill, as the presence of obstacles, that would be a little the same to calculate the things with no real evaluation of those areas. Using shooter parameters (skills / weapon), environement parameters in the line of sight (night/day / obstacle_on/off and amount), and targeted object parameters (armour / crouch_on/off indicator) would lead to similar result, even more quickly.

I add this just as a matter of discus. Final rates remain testable as you did. Only causes are obscured.

GPS51:
Just wanted to say that I really like the changes to the tech tree made in the recent days/weeks. The researching of the alien light armor leading to laser weapons feels much more "right" and adds to the gameplay. More specifically the rewording of the tech briefings (like the electromagnetic rifle) makes more sense and feels less like a placeholder paragraph :) Keep up the good work.

Navigation

[0] Message Index

[#] Next page

[*] Previous page