General > Discussion

Again on reaction fire

<< < (7/14) > >>

Telok:
Using this save under v2.5(I think it's the right one, Dam map with 3 snipers and 3 plasma rifles). I've run it five times now and have some data. One note is that my snipers are using Aimed Shot at 20TU, this is a new compile but I don't know if it's an artifact that got in the code or a reversion from the 24TU it had in my last compile.

All the sniper action is across the full length of the dam, soldier spawn fence to past the fence on the opposite end. All the plasma action is from the soldier spawn fence to targets near the pipe on the near side of the dam proper.

Interesting actions
1) A taman provoked two sniper RF at 26% chance to hit when he uncrouched, dropped a weapon, and spent 16TU moving. This is RF obviously working as intended.

2) A hovernet fired three shots without moving. Two snipers had 24% shots, one fired and the other didn't. Shooter: speed 23, accuracy 35, mind 35, sniper 32. Non-shooter: speed 24, accuracy 30, mind 34, sniper 31.
This same hovernet fired 3 shots without moving for the next five rounds. The same two snipers failed to RF until that last round. Same shooter RFed, same one didn't.

3) One taman fired two plasma rifle full-auto blasts. Three snipers at 25% chance to hit didn't shoot. Two of the three soldiers armed with plasma rifles did RF a shevarr who fired a plasma blaster burst (16TU), they had 20% chances to hit.

4) A hovernet moved and fired two shots, two snipers with 27% chances to hit did not RF. The second shot hit one of the snipers. A shevarr moved at a longer range and the wounded sniper took a RF shot at 15% to hit. Next round the hovernet fired three shots again (30TU), but neither sniper took RF that round or the next round when it fired two more shots and moved. Those RFs would have been at 27% and 16% for the wounded sniper.

% chances of RFs taken, these do not include the "Interesting actions" shots taken or not.
Sniper: 20%, 26%, 25%, 23%
Plasma Rifle: 20%, 25% (twice), 23% (one soldier did and one soldier didn't)

Action TUs and %s or RFs not taken, these do not include the "Interesting actions" shots taken or not.
move 5, shoot 16, 15% chance
move 9, shoot 28, 25% chance
move 20, shoot 10, 10% chance
move 2, shoot 20, 27% chance (two soldiers)
move 32, 23% chance
shoot 30, 10% chance (blocking thin pole), 23% chance, and 24% chance
shoot 28, 15% chance
shoot 32, 23% chance (two soldiers)
shoot 24, 25% chance
shoot 28, 10% chance (wounded sniper, at plasma rifle range)

So um... I can't see any real pattern in all this. I did note that sniper rifles with a 45%+ chance to hit will take unobstructed RF shots. There seem very little chance for RF at chances less than 25%.

H-Hour:
Dam has a lot of clutter that can complicate visibility (and thus your results). Try skirmishing on the attached map which will provide unobstructed visibility. If you've never set up a new map, place the .bsp file into /base/maps and have a look at /base/ufos/maps.ufo in a text editor. You can create your own /base/ufos/maps_temp.ufo, copy the dam mapdef from maps.ufo, and edit the names to match test_reaction.map.

Telok:
Is that "clutter" the fence posts? Because that's the only thing I can see here. I did note the one instance where a post blocked one of three possible shots.

I am trying to account for those posts, but almost every single sniper shot was into the same area. On #2 both snipers had equally unobstructed LoS and full TUs available. On #3 the plasma rifle soldiers were using the snap RF which takes 8TU compared to the alien provoking by using 16TU, but they only shot once each and never hit the alien. On #4 the LoS were not blocked by anything and the hovernet could obviously get a good line to the soldiers for 30TU multiple times.

I'm using the Dam map because it allows me to force the AI to stay at range and provoke RF. I'll check your map but don't expect any changes. By the way, shooting plasma rifle bursts and sniper aimed shots across the dam did not provoke RF from the aliens, ever. But 3/4 of the time that I used medkits did provoke alien RF.

H-Hour:
The test_reaction map will force the aliens to stay a decent range away.

Our visibility traces are minimal. I think we only draw one line between the alien and the soldier. If that line is obstructed, the alien is "not visible" regardless of whether or not other parts of the alien are visibile or if the thing obstructing that single line is very tiny. That means it's easy for very small obstructions to cause anomalies in the visibility.

There may be some kind of chance-to-hit calculation going on for reaction fire. You'd have to dive into the RF code to find out.

Telok:

--- Quote from: H-Hour on December 08, 2012, 05:49:18 pm ---The test_reaction map will

--- End quote ---

Man, that map isn't going to work. I can't control enough variables to get a good idea of what's going on.

1) It's a much shorter range than the Dam. I normally lost one or two soldiers (and more wounded) to enemy fire on the first turn, making it hard to tell if they used RF and what the %hit chance was. A starting sniper with 20 sniping and 22 accuracy had 30% to 33% chances to hit on this map, on the Dam a 30s skill sniper has ~25% chance to hit. There's no range issues here and range is one of the aspects of RF that I'm trying to quantify.

2) The aliens charge. If an alien advances towards the soldiers and provokes RF in the middle of the movement I can't tell what the %hit chance was. The Dam was good for this because the AI couldn't charge and had to move laterally. Plus they're so clumped up that I can't always tell which alien is being shot at.

3) Skirmish is lousy for this. I use 3 sniper rifles and three plasma rifles on the Dam so that I can tell who is getting RF at what ranges and %s. With skirmish I'm ending up with one or two of each weapon and I can't always tell who is shooting what.

I think a Dam map with the railings edited out would be much better for this. Then I could compare my current information with the edited map information and we could get a feel for how much the railings actually contribute.

Navigation

[0] Message Index

[#] Next page

[*] Previous page