Jump to content


Photo

How Does Explosion Damage Work -


  • Please log in to reply
123 replies to this topic

#51 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 23 October 2005 - 08:53 PM

Ok, I just added a big section to the Explosions page, where I pulled out a lot of tile characteristics (armor etc.). Z, I also tried to correlate it against your blast patterns, but I think it needs some help there. Take a look and you'll see what I mean. You've got MapView, right?

NKF, there is something weird about mountain terrain. But what the heck it means, I have no idea. See the paragraph right above Note 4 in the new section.

Also, Danial - didn't you mention something once, about the Avenger ramp having wonky TUs? I saw that in MapView!

Okeydoke, getting late here. Hasta! - Mike

#52 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 23 October 2005 - 09:08 PM

Hmmm..... Well, I'm back with some good news. I spent the better part of the afternoon getting a mission in the desert and editing my soldiers to properly prepare them for the inevitable explosion. Just my luck that there was not a clear spot of ground big enough to contain the blast anywhere on the map. Didn't bother me though - I just lined up a bunch of my troops and flattened out a dune down to the bare sand with Laser Rifles. What fun! :)

With a nice clear area prepared, I determined where the blast would go and set my soldiers down around the outer perimeter of the explosion. Then I primed the normal grenade and threw it in the center. See the next two pics:

B4_Grenade.png After_Grenade.png


Preconditions for the test: All soldiers had 150 health and 0 armor.

Now, let the testing begin! Because there are 12 spots for the outer perimeter, I put a soldier in each tile. That really cut down on the number of reloads. I got a little carried away (as usual) and ran 134 trials (1608 different values). That should be sufficient to get a good idea of what is happening. So, let's jump into the numbers:

Minimum:       5
Maximum:      15
Range:        11
Median:       10
Mode:         12
Ant. Ave:     10
Act. Ave:   10.01
Count:      1608
So the range of damage values around the perimeter is between 5 and 15. Bling! I understand what is happening now:

Dist from GZ     Min     Ave     Max
      0           25      50      75
      1           20      40      60
      2           15      30      45
      3           10      20      30
      4            5      10      15
See? The averages go down by 10 per tile. The mins and the maxes don't follow the same rule because they are calculated differently: Min = Ave / 2, while Max = Ave *3 / 2. The ground doesn't appear to play a role in this case, but other tilesets may be different. I'll hold off retracting my statement that ground plays a role, until we have more data.

The reason why the damage doesn't extend beyond GZ+4 is because the Average would go down to 0. Once the Average reaches 0, then so does the Minimum and the Maximum. Pretty simple really. But it doesn't quite explain why the range of the Blaster Bomb isn't greater though. More testing is necessary to figure that out.

I can't explain why smoke would be produced beyond this point though. I had completely untouched virgin sand with no obstructions or vegetation. Still, some smoke was observed outside the blast radius. Go figure. :D

That's about it for today. I think we learned quite a bit. Tomorrow I'll gather the rest of the numbers to verify if my hypothesis is correct concerning the normal grenade. It should go much faster because I can place 24 soldiers around the next perimeter. B)

- Zombie
The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#53 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 24 October 2005 - 08:16 AM

See? The averages go down by 10 per tile. The mins and the maxes don't follow the same rule because they are calculated differently: Min = Ave / 2, while Max = Ave *3 / 2.
I see, I see! :) It's entirely consistent with my limited test. That includes, with silacoid susceptibility thrown in. Bravo for you! I was having a sinking feeling above when some of my data (the maxxes) seemed to say one thing, but the minima were saying something entirely else. Problem solved... leave it to Z's "by the numbers" approach! =b

Right, there's that smoke weirdness with grenades, plain as day in your screencap. Not seen with other HEs AFAIK. Oh well... I guess we'll assume it doesn't signify anything useful unless/until found otherwise.

The ground doesn't appear to play a role in this case, but other tilesets may be different. I'll hold off retracting my statement that ground plays a role, until we have more data.

It just occurred to me that terrain effects must be using some fixed value from the blast front damage. Not a range; otherwise we would've been seeing at least some mottling at the edges of blasts.

Whenever you get a chance, do check out the new tile characteristics section so we can see just which tiles you did your tests on. Then if I have the time, I'll try to correlate the data. It'll probably be straightforward to relate tiles to HEs and their blast radius (fingers crossed). That old data of yours will undoubtedly then serve as another confirmation of your explosion modelling. Then again maybe we should wait til you feel you have something entirely extendible to all HE types (their averages, mins, and maxxes). So maybe it's best to wait on your additional findings? Anyway...

FWIW once the model is clear for blasts vs. tiles directly lateral from GZ, it should be cake to figure the X+Y equation they use that ultimately determines the "round" pattern for each blast type. (Did that make sense?) In a sense, it's yet another sensitive indicator of blast mechanics... to know/predict just where the edge is, even in two dimensions.

So little time, so much data :)

I am taking a look at whether I can help Danial with pulling MCD info at the moment.

#54 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 24 October 2005 - 06:28 PM

Z - I've been trying to model explosions based on your findings. Good news and bad news.

In post #26, you talked about scorched earth vs. cratered earth. I compared this to tiles armor. For desert, initial armor is 5, and death tile armor is 25. Total 30.

It appears that tile armor is stripped away as a function of half the average blast strength. A.k.a. minimum blast damage, how we've been talking about min = ave/2 and max = ave*3/2, while the average decreases by 10 per tile.

   Type       BD   DD    HE   HE/2    -25     /5     DR
   AC - HE      7    0    44    22      -      -       0
   Grenade      9    0    50    25      -      -       0
   HC - HE      7    0    52    26      1     0.2      0
   Proxy       13    3    70    35     10     2.0      2
   Small R      9    3    75    37.5   12.5   2.5      2
   Alien       13    7    90    45     20     4.0      4
   Large R     13    9   100    50     25     5.0      5
   Hi - X      13   11   110    55     30     6.0      6
   Blaster B   23   23   200   100     75    15.0     12

 BD = Blast Diameter for desert, from your table (fyi)
 DD = Death Tile Diameter, from your post #26
 HE = HE strength of explosive
 HE/2 = Half of HE
 -25 = HE/2 minus 25
 /5 = (HE/2-25)/5
 DR = Death Tile radius, counting GZ
This table has reworked your death-tile info to be side by side with HE strength. Then:
1) Take half HE (i.e., minimum damage)
2) Substract 25. Actually this should be 30 (initial plus death tile) but then it misses the case where blast is exactly equal to 30... just something to make the spreadsheet act like reality...
3) If average is decreasing by 10, then half of average is decreasing by 5. So dividing the above number by 5 predicts the number of tiles of death damage.
4) DR shows your actual results, where we count death tiles as a radius which includes GZ. (Just "DD" stated another way.)

Anything look familiar? :)

Except - there's that damn Blaster being obnoxious again.

Got scorched-vs-cratered data for anywhere else?

I will try to look at other terrains. See if I can find a match for blast patterns to their armor. Also wondering whether any armor might absorb some of the blast. None of the Blaster calculations work for me... if it decreases by 10, it should go out to ~20 radius, ~40 diameter, but it doesn't. In fact, almost nothing goes as far as HE/10. Could it be because most are breaking through tiles and getting damage absorbed up? But the Blaster is still hella strong right to its edge. Wah!

We keep going back and forth on this so apologies if I missed something you already said :P

#55 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 24 October 2005 - 08:19 PM

Anything look familiar? :)

Except - there's that damn Blaster being obnoxious again.

Got scorched-vs-cratered data for anywhere else?

<{POST_SNAPBACK}>

Apologies, Mike. I did have a look at the table in the Wiki yesterday, and tried to figure out what all the data meant. It'll take me a while to understand it completely. Hobbes seems to know what is going on in that area better than I do. See... well, you did see the post in the Damage Modifier thread in the StrategyCore forums. Let's wait till he replies. :)

Anyway, I have a couple saved games trying to determine the cratered vs. scorched earth ratio. Though, I am having some trouble since some landscapes do not get the "cratered" look when hit with HE. I think the only two (well, three actually) landscapes which get cratered is Desert, Snow or Cydonian. Other landscapes have different "damaged" looks. I therefore prefer to call it: injured vs. destroyed. (Or Ini vs. Ded like Daishivas MapView prog shows). Same manure, different pile. :wink1: Gimme a chance to gather something up for ya.

I have been insanely busy today, what with explosion damage from the normal grenade and all. Nothing is as easy as it first seems. (And we thought explosives were easy, huh). But I do have some new info to present which could be considered partially earth-shattering. I just have to crunch the numbers and take a couple screenies to explain it. :)

- Zombie
The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#56 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 24 October 2005 - 09:01 PM

Sounds great, Z. It's a little hectic tieing it all together - the moment of maximum entropy? - but it does seem to be coming together. Can't wait to see what you've got!

I just looked at my data from post 21 again...

Although the sample sizes are small, it's entirely consistent with the same explosion formula you've been saying all along. Once muton armor is taken to account, we see the min/ave/max, decreasing by 10 per tile, with HE values as expected. Even for the Blaster Bomb.

The only "problem" is that most explosions simply stop short of their anticipated radius:
                    Act.  Act.   HE@    Ant.  Ant.   BR
     Type      HE     BD    BR    Edge    BD    BR   Disc.
   AC - HE     44      7     3     14      7     3     -
   Grenade     50      9     4     10      9     4     -
   HC - HE     52      7     3     22      9     4     1
   Proxy       70     13     6     10     13     6     -
   Small R     75      9     4     35     13     6     2
   Alien       90     13     6     30     17     8     2
   Large R    100     13     6     40     19     9     3
   Hi - X     110     13     6     50     21    10     4
   Blaster B  200     23    11     90     39    19     8

 HE = Explosive rating
 Act. = Actual (in game)
 Ant. = Anticipated (from formula)
 BD = Blast Diameter (includes Ground Zero)
 BR = Blast Radius (does not include GZ)
 HE @ Edge = Actual explosive damage at edge of actual blast (HE - 10 * Actual BR)
 BR Disc. = Discrepancy in actual vs. anticipated blast radius
The actual damage at the edge of blasts is precisely what you calculated it should be for that tile, as far as I can tell. But then they stop before they "should" - before the blast has decreased by 10 until it's nothing. ("HE at edge" is the average, which is decreasing by 10.)

I figure the programmers realized that blasts were too big, and put something in to curtail them. (They couldn't model everything, working in integers back in DOS days!) Whether it's hard coded per explosive type or based numerically on HE, remains to be seen.

Stop me if you already told me this :P

Injured vs Destroyed sounds good to me. The Ini vs Ded (Initial vs Dead) was more me than DaiShiva... MapView simply has first the one tile's armor value, then you can see the death tile's armor value if you look at it. Whatever works. So "injured" means "initial tile gone" and "destroyed" means "death tile gone too" (if there was one) eh.

Right, I'm waiting to see what Hobbes says, too! How does he know all this stuff anyway? hehe. Also waiting on your stuff, for sure!!

Edit: Before I forget - the table I just posted is also consistent with the previous one, about how minimum explosive damage is probably what is damaging tiles... due to the fact that tile damage mostly occurs in the center of the blast, only Blaster was large enough to get "clipped" as per this point I just made. Also re: my previous post: Notice how your second grenade is then capable of blowing through the death tile at ground zero; half of 50=25, the death tile's armor value.

Edited by MikeTheRed, 24 October 2005 - 09:06 PM.


#57 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 24 October 2005 - 09:26 PM

Hobbes knows so much about this stuff is because he designs his own battlescape scenarios and maps (not including his work in UFO2000). Impressive guy to say the least. B)

Well, I was trying to get some numbers for you concerning Injured vs. Destroyed tiles. Luky me, I landed on a farm. So I started to test explosives on grass (you know, the short run-of-the-mill variety. Seems that everything I threw at it never got it to the destroyed look. Heck, I even nailed it multiple times with a BB with no luck. Then I looked at the table in the Wiki. 5 for initial and 255 for destroyed. That explains it. Grass tiles can't go to the "next stage" of existance. They can only be injured. Well, seeing as the table is making more sense to me now, I'll concentrate on tiles which can be damaged and destroyed. :)

Oh, right. Yes, my data is coming together quite nicely. It is actually a bit shocking now that I look at it more. Completely goes against the OSG in some respects. Lemme put this out though: imagine you are Julian Gollop trying to model explosions in X-COM. How would you show a circular blast pattern with only square tiles? That's right, you have to cut some corners in order to make it work properly. Seems as though damage from a grenade follows this same idea. More on that later though. =b

[Edit: I just had a crazy idea concerning the Blaster Launcher. It's listed damage is 200 (average). The min should be 100 and the max 300. But how could this be if the game is limiting numbers to 255? Perhaps the max of the Blaster Bomb is 255 instead of 300. A decrease of 45 points could do wonders to explain why the blast radius is smaller than expected. Just a theory though. :wink1:]

- Zombie

Edited by Zombie, 24 October 2005 - 11:09 PM.

The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#58 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 25 October 2005 - 02:52 PM

I'm back with a major new development on a normal grenade's blast radius. If you remember my pics from a few posts back, it showed the absolute outer perimeter of the grenade (GZ+4) where damage was between 5 and 15 damage points. For my next series of tests, I placed 24 soldiers directly inside the outer perimeter on the next "shell" where damage should be between 10 and 30: OSG__s_GZ_3.png

That makes sense because the OSG predicts a 10 point damage drop-off, even on the diagonals. So I fired up my logging program and started to gather some numbers. Imagine my surprise when I looked at the resulting data (1608 values): a range between 5 and 30! Something is obviously wrong. After mulling this over for a while, I decided that it was possible the 4 soldiers standing on the corners were actually getting the lower damages while the others were getting the anticipated damage. However, even after removing these soldiers and gathering another 2000 values, I was still seeing a range of 5 to 30.

Before blindly taking more soldiers out of the blast, I decided to do a distribution of the values. When I compared the two trials head-to-head, the distributions of values lower than 10 was going down. Luckily, I preserved the soldiers’ names in the data. This allowed me to figure out which soldiers were getting the lower damage. After removing them from the mix, this is what I was left with:

Temp_GZ_3.png Overhead_GZ_3.png

So those screenies shows the actual GZ+3. It looks exactly the same as the pic in post #52, except now the troops are one tile closer to ground zero. Fortunately, by removing those soldiers getting less damage, I was able to compile an actual outer diameter of the blast (GZ+4). This is what it looks like from all the angles I could think of:

Actual_GZ_4.png GZ_4_After.png Overhead_of_GZ_4.png


As you can see, the tiles which take 5 to 15 points of damage increased significantly since my first test. (Actually, the number of tiles doubled). It seems as though the programmers decided that the diagonals take less damage than those within a 3 tile range surrounding GZ, or the tiles themselves are sopping up damage. (Here we go reviving this theory again). :wink1:

Currently, I am working on what the perimeter (or shell) of GZ+3 looks like.

MikeTheRed: Here is a question. You showed the blast produced by a Stun Bomb with a nifty little diagram. What program did you use to make it with, and can I get it from somewhere? See, I think after all those shells are figured out, the screenies I took could be compiled into one neat diagram with different colors representing the various damages. Catch my drift? :)

- Zombie
The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#59 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 25 October 2005 - 05:26 PM

More great stuff, Z...

Sure, here's my "blast diagram". All it actually is, is Excel (2000) with colored borders, fills, and centered numbers. Then a screencap. Done. =b heck if you wanted to get fancy you could make conditional cell colors (/Format /Conditional Formatting) based on distance from center, if you figure an "equation". But it'd probly be a lot faster to just color by hand, heh.

Actually, without the knowledge of the OSG, I felt sure that blast strengths were computed, and you'd see differences on diagonals. Look at it this way: We know that, at the very least, HE Block does cause "dynamic" propagation effects. And if they programmed that, isn't that 90% of the code needed to just let explosions make their own "circles"? So I always assumed diagonals were different.

So, you ONLY ever saw 5-15 on your test with "actual outer diameter of the blast (GZ+4)"? This means that they are doing pretty unsophisticated X+Y computations, to me. A.k.a. you're not a integer+decimal distance from the center - it's XCOM's usual integer rounding. (If it wasn't, you might've seen e.g. a decrement of 7 instead of 10 on diagonals, or anything not due lateral.) I'll work on this concept a little. I might be able to make charts for all bomb blasts before you can :) Either way though we'd need the raw data to confirm it works that way. I can't think of a better man for it than you. ;)

Would you like a Harp or is that humble brew too boring for you? :cheers:

I'm refining the Tile Characteristics table and pulling out some more numbers. Also double-checking what I did before, now that I understand it better. That table was done in a bit of a rush.

Sure hope Hobbes gets back to us.

You might consider putting numbers into the beginning of soldiers' names. Much easier to remember them either as a circle of increasing numbers, or even hack their coordinates relative to GZ into the beginning of the names. Then you can readily pull out their location, less likely to make mistakes, etc. I've found it real helpful sometimes, anyway. Let me know if you want me to edit them into a game for you. You probably don't need it but here's a little primer I wrote on hacking.

When you do your tests, do you use an automation tool like Autohotkey? I've only used AHK for very simple keyboard or mouse redirects, but something like that can conceivably entirely automate testing, although you would not be able to change conditions during any one automation run. Still, you could gather data while you went out to eat. :)

Attached Files



#60 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 25 October 2005 - 08:43 PM

Thanks for your blast diagram, Mike. Much appreciated! :)

Well, I wasn't expecting the results from my tests simply because the OSG clearly shows a 10-point drop-off per tile in figure 8-21 on pg. 225. All the "shells" shown are complete squares, and all have the same damage rating. I think that in this case, people like yourself who do not have access to the OSG are in a better position to understand the mechanics behind blasts because this information is not permanently ingrained into your skull (like it is in mine). :wink1:

My tests with the outer perimeter of the blast GZ+4 (see final three pics in my previous post), all showed a range of between 5 and 15 damage points. From the limited tests I completed on GZ+3, all had a range of between 10 and 30. Therefore, I'd say my initial calculations agree completely with the experimental results obtained. (That's always satisfying, wouldn't you say)? I'm sure I could whip up those blast diagrams in a jiffy too, but I think additional data is necessary with more powerful explosives and different tilesets. That way we aren't just spinning our wheels to keep busy if the results don't match theory. Hopefully by just obtaining data from one tileset, we can expand that equation to cover other tiles with different properties. At least, that's the goal I'm shooting for.

BTW, Hobbes did reply back to your query. See this post.

As for adding numbers to or changing a soldiers name, that can easily be done without hacking or editing the game files. I just go into the soldier stat screen and click on the "NAME" button or the name itself. Both allow you to edit a soldier's name to whatever you want - including numbers. But working with only numbers for a long time has taught me that a simple name is less ambiguous than a number - especially if you temporarily memorize the order of your squad. Numbers would lead to more confusion if you were trying to relate order to the position on a screen. (Let's see, number 1 to number 4... :unsure: )

I actually prefer to be partially in charge of the data gathering aspect. It never bothered me. Letting a computer completely take over that task would require less time, but if something went wrong while I was away, I'd never know. The problem could go unnoticed and lead to bad results. "Messy" data is much easier to remove if a person is watching as the tests progress. If something unanticipated happens, then that data can be quickly removed before it gets "lost" in the sea of numbers. :)

- Zombie

Edited by Zombie, 25 October 2005 - 08:55 PM.

The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#61 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 26 October 2005 - 12:06 AM

That sounds great on the GZ+4, Zombie. Just wanted to double check, because it would change everything if they were using a more complicated formula. For example, using the Pythagorean formula, a diagonally adjacent tile is actually 1.4 tiles away and be expected to get 1/1.4=70% of what the directly adjacent tiles get. Those guys out at GZ+4 to the side one, are 4.12 tiles away. The dudes 3 diagonals out at the "corners" in your GZ+4 pic are actually 4.2 away. One way or the other you undoubtedly would've seen some slight nudges in the min or (more likely) the max if XCOM did not keep it real simple. Looks like they kept it real simple.

I tried plugging pythagorean into a spreadsheet and computing blast radii, as we discussed. Find it attached. Even using an integer function after the square root, it couldn't get it to work. In fact that lone tile seen due lateral in many blasts patterns is very unexpected if they used anything sophisticated. It must've been something crude or subject to rounding/trunction... but, what's new. I can't figure out what it is, though. Given how they handled decreasing explosives, I wonder if they did a "manual" calculation of square roots, but only out to say a couple of manual steps - a decent approximation, but which will give some jaggedness, basically. But it's a little too much for me to wrap my brain around and put into a spreadsheet at the moment. Alternatively they may very simply have a lookup table that can be used by any explosives; a "template" for what tiles are radius 2 away, radius 3 away, etc. But it still doesn't explain some of the variations in patterns; how some weaker explosives extend farther than stronger ones. That may be due to something else, though.

Thanks for pointing out Hobbes reply! I had just joined the board and didn't remember to turn on email notification until after I posted the message to him, and forgot I'd have to go back in and turn that one on. Let's see what he has to say. Think I went overboard? :Hyper:

When I get ready to do testing, I've traditionally lugged all my equipment in... then later numbered both soldiers and aliens. Can't edit names on the battlescape, AFAIK. Anyway, whatever works, eh? Same goes for automation... just curious! It gets pretty repetitive to reload umpteen gazillion times... my mind starts wandering badly around the 25th time.

I screencapped all explosive blast patterns and will put them in a spreadsheet. These guys really are good indicators of explsive strength. It may help elucidate some things. But it's getting late here. Til tomorrow!

Mike

Attached Files



#62 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 26 October 2005 - 05:51 PM

When I get ready to do testing, I've traditionally lugged all my equipment in... then later numbered both soldiers and aliens. Can't edit names on the battlescape, AFAIK. Anyway, whatever works, eh? Same goes for automation... just curious! It gets pretty repetitive to reload umpteen gazillion times... my mind starts wandering badly around the 25th time.

<{POST_SNAPBACK}>

Sorry, I meant I would edit soldier names before entering into combat.

It's not the repetitive nature of reloading which keeps me coming back for more; it's the satisfaction of finding something interesting about the game. And if I can solve a problem along the way which helps other people play better, that puts the icing on the cake. This type of testing is definitely not for everyone, as it takes plenty of patience to sit through "umpteen gazillion" trials. :)

Well, onward and upward. I just finished testing out the normal grenade on a desert battlescape. Nothing really new to report as the rest of the damage perimeters could be correctly deduced from my previous tests on GZ+4. But it still might be worthwhile to take a look at the screencap I took of the results:

Normal_Grenade_Damage_Shells.png

0 = GZ (Ground Zero), 25-75 damage points
1 = GZ+1, 20-60 damage points
2 = GZ+2, 15-45 damage points
3 = GZ+3, 10-30 damage points
4 = GZ+4, 5-15 damage points

So all the damage ranges check out via extremely rigorous testing on my part. I think we can conclusively say that damage drops-off by an average of 10 points per tile away from the blast. However, the 45° diagonals are a bit sketchy at the moment with only 3 data points: 1,3,4. A more powerful explosive will solve that though. :wink1:

You know, I think we may be barking up the wrong tree trying to use Pythagorean's theorem and such. Like you said the game is much simpler than that. I'm thinking more along these lines: each "normal" tile in the battlescape costs a unit 4 TU to cross from face-to-face. It takes 6 TU to cross that exact tile diagonally (or the hypotenuse). As an approximation, the hyp of the triangle could be found by this equation: Hyp = 4 / Sin(45°) ≈ 5.66 ≈ 6. Perhaps a ratio of 2/3 is used somewhere (67%). Seems like a better choice, though I haven't actually crunched the numbers yet.

I tried to formulate a theory to describe how a tileset affects blast diameter. Let me lay it out: All tiles on the battlescape have an initial damage rating. If an explosion happens, the GZ tile absorbs part of the blast (not much for most tiles). This lower level is then passed on to the next tile. That tile will absorb another small part of the damage and pass the result to the next in the line. And so on and so forth until damage left :doh:

- Zombie
The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#63 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 27 October 2005 - 08:09 PM

Today I ran some more tests on grenades and explosives in general. First test was the theory that the Blaster Bomb cannot do any more than 255 points of damage, even though its range at ground zero is between 100 and 300. Turns out, this is untrue. After editing a soldier to have 255 health and 100 under armor, and running 55 trials, I saw a damage range between 103 and 298. Close enough. BB's can dish out up to 300 points.

My second test was actually two-fold:
  • Establish whether the damage dealt at GZ determines the damage dealt to subsequent tiles away from the blast, and
  • Resolve the issue that a soldier standing closer to the blast will "shield" a soldier farther away.
Since both can be determined from one simple experiment, this is what I did. I placed one soldier at GZ and then a soldier at GZ+1, +2, +3 and +4. Now I ran about 100 trials and compared the health level of all the soldiers.

Okay, damage dealt at GZ has no effect whatsoever on damage dealt to tiles farther away. Sometimes the soldier at GZ was receiving less damage than the soldier at GZ+1 or +2. Problem resolved. This inadvertently solved the second question also. If the soldier at GZ did indeed "shield" the others, then they should never get more damage than he did. That was not the case. What this means is that from now on I can test explosives on a much larger scale by using more soldiers inside the blast radius. :crutches:

As an interesting note, I kept track of each soldier's health level for each trial and compared that to his friends. Remember that for a normal grenade the damage dealt should be this (min-ave-max):

GZ: 25-50-75
GZ+1: 20-40-60
GZ+2: 15-30-45
GZ+3: 10-20-30
GZ+4: 5-10-15

Turns out each soldier within the blast is calculated separately from the damage dealt to GZ. Therefore, units standing at GZ+1, +2, or even +3, can possibly receive more damage than the soldier standing on Ground Zero! It's true, and it happened in a couple of my trials. Just look at the ranges. Obviously, if a soldier stands 4 tiles away from the epicenter, his range is so low that it cannot overlap the GZ's range.

For instance, let's say for the sake of argument that you have one soldier standing on each different damage shell possible for a normal grenade. The soldier standing at the epicenter will get a random roll between 25 and 75 points of damage. The game will then do a random roll for the soldier standing one tile away from the blast (between 20 and 60), and do random rolls for all the remaining soldiers caught within the explosion (15-45 for GZ+2, 10-30 for GZ+3, and 5-15 for GZ+4). Very interesting.

Well, I suppose it's time to move on to more powerful blasts in the desert landscape. I think I exhausted all the possible research topics from the normal grenade. Perhaps the Alien grenade will yield some new info. :hideit:

- Zombie
The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#64 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 27 October 2005 - 08:44 PM

Thanks, Z. More testing goodness. Who needs food with all this nutritious stuff?

Still need beer, though. Prost! :cheers:

Actually it had never occurred to me that it wouldn't be random for each soldier, within the range available to that one square (min to max). It is really good to pin that down, just the same!

As for me and the issue of whether soldiers can block each other, I might (or might not) want to look at that a little myself after we entirely understand terrain effects. More specifically, once they are understood, then it should be easy to find the most sensitive test for soldier blockage. Something like, just barely have enough tile armor that even the slightest change in blast strength would be noticeable. (I.e., the next tile past wouldn't get damaged - but by the slightest of hairs.) Then, pile a ton of guys in a wedge formation between that sensitive tile, and GZ. Then pop the frag.

All in all though, this particular line is looking less and less worthwhile to me. (And you've given up on it, eh?) I mean, even if there is soldier blockage - and results to date have not supported this - it is clearly not very much. Also known as, if you're bunching all your guys up and you get exploded... well, take a note, dude LOL

Did you see my other post? For anyone following this thread who is not aware, we are also discussing explosions some with BombBloke and Hobbes, over on StrategyCore.

Right now I am manually "transferring" screenshots of all explosion types (in desert), into that Excel "blast graphic" format. I want to mull over what we see, versus HE. I still cannot get over the fact that slightly stronger explosives sometimes make slightly smaller blasts, than other ones weaker than them. I am hoping that having all patterns in front of me at once might jar a thought loose. I don't agree with the idea that tile armor is soaking some of it up and causing the difference, because one of the discrepancies is between grenade and HC-HE, and they both only affect the initial tile. No difference in tile armor effects that might be soaking it up.

I really like your example of TUs as a measure of tile distance from GZ. Clearly pythagoras is the wrong approach - and, yep, XCOM probably did something real simple sort of like with walking TUs. It's even occurred to me that maybe they actually did use walking TUs as the measure - that code was already there for them to use - but I'm thinking some of the patterns are not consistent. But I want to make my diagram for them all so that I can get my brain around them, all at once. If we can work out blast propagation entirely, we have a lot of sensitive info written right into the terrain, there for the looking. I will post it up as soon as I can.

If you do test alien grenades, note that you only need to test a slice of it (an eighth) fully, and can spot test the rest for symmetry. Also fwiw, next time(s) I will try to make test combats with all the soldiers possible :P What's the max, 24?

Bon voyage! Any and all data is great at this point.

Mike

#65 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 27 October 2005 - 11:04 PM

Okay, here's my spreadsheet with blast diagrams.

The "D" after bomb names means, how much a discrepancy there is versus expected radius (post #56).

It hasn't given me any big insights. Yet. But there are some things to see... I'll use your terms "injured" (initial tile gone, death tile showing) and "destroyed" (death tile gone, too).... Also Gren=Grenade, SR & LR = Small and Large Rocket, HX=Hi Ex, AG=Alien Grenade, SB=Stun bomb, AC=AC-HE, HC=HC-HE...

One interesting thing is that the inner "destroyed" patterns exactly correlate with your grenade "iso HE contour" findings. Prox and SR have destroyed radius 1, AG 3, LR 4, all consistent with your Grenade contours. I bet the outer edges of Stun Bomb are contour for 5, and Prox is contour for 6. (Prox does not have any Discrepancy.)

There is only one exception to the expected rule of, damage <30 should equal an injured tile: The lateral "tips" (radius=6) of Hi-Ex are destroyed, when they should only be injured. Hmm.

AC (HE 44) has the exact same pattern as HC (52). But HC has a Discrepancy of 1.

Prox (HE 70, Discrepancy 0), AG (90 D2), LR (100 D3), and HX (110 D4) all have exactly the same outer pattern. Could some high-level bureaucracy have decreed that all blasts must conform? laugh. I really think they put some sort of hard-coded boundaries on large blasts. Even the puny HC got snipped.

In retrospect, I guess they realized that their math was getting out of hand, so they put a simple lock on it. Real-life point-source explosions follow a cube law (r^3) since they expand in three dimensions, although in reality they're not theoretical points with nothing around... a bomb on the ground mainly explodes into half the available space (into the air, not the ground). Anyway... XCOM's approach of dropping off by 10, instead of some much more sophisticated (and steep) pattern would've led to some absolutely huge blast patterns if followed to its due end, such as a blast diameter of 39 for the BB. That's almost 4 tilesets, or 962 tiles using pi*r^2 on 17.5; how big is a map, 5 tilesets? 962=38% of 50^2. Anyway... BBs still suck, huh?...

Notice how all smaller projectiles (SR, AC & HC) have a square pattern, but all thrown objects and large projectiles (LR, BB... SB?) have more circular patterns. Maybe there is some coding difference for launched weapons?

Related to all the above, neither AC nor HC follow your Grenade contours. (That's not a radius 3 contour.)

Could they have snipped both 1) every projectile (to make them a little different from ground-exploding bombs), and 2) anything with radius >6? The HC outline may have meant some guy was pressed for time late one night (=dittoed AC snip).

Ok, that's enough questions (and not enough insights) for now. It's neat to see your contours laid out again, though...

Mike

Edit: I went ahead and put on the spreadsheet, a projected iso-HE contour for radius 5 and 6, IF (and that's a huge if) Stun Bomb is contour for 5, and Prox Mine is contour for 6. But I doubt this is correct since - news flash to self - Stun Bomb is strength 90, which means it should have a radius of 8, which means it must've been snipped. And is probably not radius 5. However, the prox mine may still be a good for radius 6.

Attached Files


Edited by MikeTheRed, 27 October 2005 - 11:53 PM.


#66 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 28 October 2005 - 12:25 AM

Z, Spurred by your thoughts, I have something that works for predicting blast radii (attached). More precisely, "the perceived distance from GZ at each point within the blast pattern". At least for what you've shown for the grenade.

I based it off of TUs, per your thought. Extended it to r=11 (BB), wth.

Take the max of X and Y distance from GZ. Then the min of X and Y, and divide by 2:

Radius = Int ( Max(x,y) + Min(x,y)/2 ) )

Always using absolute values for x and y (no negatives).

See how it relates to TUs? Diagonals cost 1.5 as many TUs as adjacent tiles. Min and max is just a way of cutting to the chase of the shortest path. As it were.

It works perfectly for the grenade.

Into the future though, it does not have that little tip of blast damage seen due laterally for many explosives. That's probably due to some sort of rounding or maybe even they just forced that little spike into their code, to make things interesting.

Note that, if this equation is right, then Prox Grenade is not the pattern for r=6. Then again, they may not have put the integer function exactly as I did, etc. There's room for play in that simple equation.

Can't wait to see your results for Alien Grenade!

Attached Files



#67 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 28 October 2005 - 12:54 AM

Incredible stuff, Mike! I'm speechless, really. OMG

Its nice to see all those blast diagrams in one spot. At least we can compare and contrast at will. :)

Listen, I haven't started testing on the other explosives yet as I just got done with the grenade today. There are still a lot of questions floating around (what's new). My question to you is "what would you like me to test next"? What is bugging you the most? I mean, my temporary schedule is completely open so I can pretty much tackle anything you want answered.

There seems to be some questions about the Proxy mine as well as the Stun Bomb. Should I start on those first to clear up the contours problems, or will testing the Alien Grenade resolve these issues? Or perhaps one of the fired explosives (HC, AC or RL) would be better?

As I said, I'm open to any testing suggestions. ;)

- Zombie
The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#68 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 28 October 2005 - 01:07 AM

Z, thanks so much for the wide open reply! Prost!

My first immediate thought is, please address the issue in the third paragraph here and let me know the MCD id's for those tiles. I think it will only take you 10 minutes, and then I can do a lot more work with your older data. That's an extensive and important dataset, but I don't want to touch it until I know what armor levels it applies to.

FWIW I have tomorrow off. We got to a stopping point at work on a big project I'd been heading up. So I'm taking the day off. Which means I'm staying up, laugh.

More to come....

#69 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 28 October 2005 - 01:35 AM

The next thought, if I can't think of any others, would be to test Hobbes' new tileset. I don't know what to make of his HE Block statements, but I do like the thought that the tileset has hacked armor of 10. This means that even just blast patterns should tell us tons about whether our insights are valid. Just one shot per weapon! Should confirm that all our observations are correct. It doesn't need the whole combat done... just test one of each explosive, once. On tiles with no HE Block. Store that savegame cuz we'll get back to HEB, I'm sure. And you my intuit some great revelations if you play with HEB there already.

I don't know about you, but I want to get tile armor figured out first. THEN move on to HE Block. I think we are almost there, for understanding explosives versus ground tile armor.

Just like you are, there are many open questions. Another important one is to double check that damage to units does occur at the expected radii, regardless of tile armor. Using your Diameter table, it looks like Mountains or Alien Bases are extreme tests - extremely high tile armor... are units still damaged as usual at radius 11 by BB??

There are so many little questions with big implications.

I have a savegame where an alien base is rounded up but not yet put down, if you want it. If so, let me know if you want me to hack soldiers.

I'll keep thinking for a bit. There's so much to nail down, it's hard to prioritize. Anyway, there's two things to pick from. But please send me those MCD ids before the rest, if you'd be so kind, sir! :cheers:

#70 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 28 October 2005 - 03:23 AM

Ok Z, it looks like I may need to head on soon. May be offline for a couple of days, it's looking like. Somebody talked me into it.

I tried to think of more things, but there are just so many. In truth what I want to see is Hobbes making a wide variety of tile armor, but that's out of our hands. And sounds like a lot of work, not only in him making it, but also us combatting each one. But it would've been a real nail in the coffin concerning "regular" ground tile armor damage. As for what I would like you to do (thanks for the request!), I think that most anything you do is great. Everything you do sets more ideas to rest - or raises new questions, as it should be.

To any extent you can gather samples on the various blast patterns to ensure units take damage predictably, that's great. For example, HC and AC and others strike me as "unconventional" patterns. I bet it's due to the programmers futzing with blast edges to make projectile explosives "seem" different than thrown ones. Which is to say, I bet HC and AC follow the same grenade contours, but just cut off at the programmer-defined edge. But as we both know, this is pure conjecture. So you see, all patterns could be tested, on just one map. Only regular tiles are needed (and I do suggest regular ones, i.e., a natural desert map or Hobbes' second tileset... the armor must be >0, that's too confusing). Keep in mind that 1) you only need to test one eight of any pattern much, and 2) every single max and min rules out any other possibility, as long as you collect data entirely rigorously. Which means that, for once, you aren't in theory collecting infinite datasets in hope against being proven wrong... you need only one datapoint to prove you're right. =b

That's the good news. The bad news is that if you ever do see one contradictory value, it's back to load and reload.

As I understand it, your middle name is 'rigorous'. ;)

Prost, bro! - Mike

#71 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 29 October 2005 - 04:21 PM

Well, it took me a while but I was finally able to construct a mission to test out all the explosives in a desert landscape. (I don't know how yourself or NKF do this so fast - it took me about 45 minutes to get everything "just right"). :master:

Anyhow, I packed 26 soldiers on an Avenger and sent them in. Ooops! I totally forgot about the HWP/Launcher. There goes another 45 minutes down the tubes. Being a bit lazy, I used Hatfarm's editor to manually put a HWP/Launcher on the Avenger in addition to the 26 soldiers. Hehe. Unfortunately, two of my soldiers spawned, well, somewhere. I think they are either underground or on Level 4 (go figure). Doesn't matter though, 24 are more than enough. :)

Now it was just a matter of clearing the battlescape of enough hills to start my tests. I wasted another 30 minutes shooting Heavy Lasers at a large hill until it was gone. Finally I had my test game set-up correctly. Arrgh! :crazya:

Right now I started testing on the Proxy mine. Since I have enough men, I'll test 1/4 of the blast to save some time.

Right after I cleared the ground from the giant hill, I decided to check the HWP/Launcher's blast pattern. Hmmm... exactly the same size as the Blaster Bomb, but 3 tiles at the edge are damaged instead of destroyed. I added this to your Excel table, Mike. =b

Attached File  XCOM_Bomb_Patterns.zip   5.1K   48 downloads

- Zombie
The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#72 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 29 October 2005 - 04:45 PM

You're right... I kept forgetting about that darn HWP! And I brought it there for testing!!!

Thanks for adding that pattern... now we have a contour for radius 8!

I'm going to make a table of all possible explosives (inc. HWPs) and put it in the wiki Explosions page...

I'm not sure what I'll do next.

It seems pretty clear that we've worked out the concept of damage to ground tiles, although if you can establish the iso-effect radii past 4, it'll be good to have that nailed down... the Alien G should work fine... anything with radius 11, really... also it'd be good to spot check a few places against e.g. Grenade findings to make sure other blasts do have follow the exact same iso-effect radii. We assume they do, but don't know for sure. Actually I guess we're pretty sure. Right, the Prox Mine or anything else works for establishing those effect levels and checking against Grenade etc. Anyway.

I might take Hobbes' new tileset with HE Block for a test drive... hmm...

If you get a chance, can you zip up your current test savegame? Maybe I'll give it a spin sometime. Yep, it can take me 30 to 60 minutes to get a savegame set up too, and I too hate it when I forget something!!

Wow that's weird about the two soldiers. Can you get to them with the "next soldier" tab? (and then try to move them??) Or can you see them by tabbing through inventory?

#73 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 29 October 2005 - 06:59 PM

I took a Rocket Tank (HE 85) for a drive in the desert.

It has radius 5 (discrepancy of 3) and inner destroyed radius of 2, consistent with calculations. The inner radius matches your r=2 contour for the Grenade.

It has a squarish pattern, so it's consistent with how projectile explosives have squarish ("imposed"?) shapes, instead of circular for thrown ones.

I also screwed around with the Blaster a little more. On battleship floors (armor 80) it does floor damage to r=4 as expected but DOES still injure folks at r=11. So there's another confirmation that blast damage to units still happens as far away as expected, even if it's not enough to damage floors.

Also I confirmed that you can destroy a lot of the control room walls where those green-on-black "computer display" wall tiles (armor=70) exist, even though the blaster otherwise only takes out the one wall tile where it hits, for that type of 3rd-floor battleship "security wall" (armor=100) wall tile. As also discussed here.

Here's an update to that spreadsheet, now with Rocket Tank added. I'll be working on the Explosions part of wiki for the moment.

Attached Files



#74 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 29 October 2005 - 07:29 PM

Hehe. I forgot about the Rocket Tank! :blush1:

Hmmm... that's strange. The HWP Rocket has a perfect 9x9 square pattern for GZ+4. But for the Small Launcher, the GZ+4 pattern has clipped corners even though it has 5 more "damage" points. I'm starting to think a Rocket Tank would be a better choice to take along. That's interesting because it follows the same pattern as the other fired-HE weapons. :)

Sorry about this Mike: My test saved game is all messed up. Well, not really, but bad enough. I can still test the normal explosives out. But I forgot to research the Blaster Launcher/BB, Small Launcher/SB before sticking them on my craft (darn editors). Lemme fix it up for ya after I get done with the Proxy tests.

Cool point about the Blaster Bomb in the Battleship. That means my theory is wrong again. Back and forth and back and forth... I'll call it the rubber-band theory: it constantly changes to just the opposite. LOL

- Zombie
The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#75 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 29 October 2005 - 08:47 PM

Don't worry about the savegame; certainly don't do a whole mission just for me. I may or may not ever get around to it. But if you do get one where you've got everything, go ahead and post up a savegame at the point it's ready for testing.

FWIW I have that rocket tank game. The 16 soldiers have every type of weapon in the game (guns and bombs) plus rocket tank and fusion tank. However, I have not gotten worked all the way through to setting up, I only just started it. Let me know if you'd like it. It's for regular desert, not Hobbes' hacked desert.

If you get a chance, it would be really good to do a little testing at the edge of blaster explosion in a UFO (on UFO floors). If we see exactly the same damage as expected at edge then I think we've pretty much nailed that coffin shut. It wouldn't need tons of testing; just enough to get a good handle on the range for a couple squares at the edge. You could get 4 or even eight soldiers near there, in a long hallway two wide. Or an alien base game... I have a savegame with blasters and 16 soldiers in a base, and Fusion Tank and alien grenades (but no other explosives), if you want it.

I saw something really weird when screwing with the blaster. If you make it go up a battleship's elevator and then come back down to hit the ground floor of the elevator, it disappears off the map. There is no explosion. Go figure. Anybody ever seen this before? (Aim at a spot with nothing there; no soldier or whatever.)

Also when doing that, one (but only 1) of my blaster weapons did the weirdest thing... after firing, it still had a blaster bomb in it, but the bomb said "ammo=0" instead of the usual 1. WTF If I tried to fire it, I would only get the "Throw" option. So I simply unloaded it. Put it in a trophy case. "The Bomb That Wasn't." ;)

Hey, check it out - that inner radius=8 you drew for the Launcher Tank EXACTLY matches the predicted iso contour for distance 8, for that "walking TUs" sheet I made. (It's easy to highlight 8 if you use /Format /Conditional Formating, then change colors for value >8. Select the table first.) I didn't really trust that sheet before because radius=4 wasn't a whole lot to work with (lots of clipping and rounding with so little distance). But now I'm feeling a lot better about it, seeing that it matched 8.

I'm fleshing out Explosions part of wiki currently.

- MTR

#76 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 29 October 2005 - 09:06 PM

If you get a chance, it would be really good to do a little testing at the edge of blaster explosion in a UFO (on UFO floors). If we see exactly the same damage as expected at edge then I think we've pretty much nailed that coffin shut. It wouldn't need tons of testing; just enough to get a good handle on the range for a couple squares at the edge. You could get 4 or even eight soldiers near there, in a long hallway two wide. Or an alien base game... I have a savegame with blasters and 16 soldiers in a base, and Fusion Tank and alien grenades (but no other explosives), if you want it.

<{POST_SNAPBACK}>

You mind if I do that after the Proxy? (What am I saying? Of course you don't). I just started and don't want to stop until its finished. Then I'll hack another game with all the explosives. :wink1:

I saw something really weird when screwing with the blaster. If you make it go up a battleship's elevator and then come back down to hit the ground floor of the elevator, it disappears off the map. There is no explosion. Go figure. Anybody ever seen this before? (Aim at a spot with nothing there; no soldier or whatever.)

Also when doing that, one (but only 1) of my blaster weapons did the weirdest thing... after firing, it still had a blaster bomb in it, but the bomb said "ammo=0" instead of the usual 1.  WTF If I tried to fire it, I would only get the "Throw" option. So I simply unloaded it. Put it in a trophy case. "The Bomb That Wasn't." ;)

<{POST_SNAPBACK}>

Yeah. Heard of both. The first scenario is called the "vertical waypoint Blaster Bomb bug". Only happens if you try to fire directly up or down.

The second scenario is usually the "1-waypoint Blaster Bomb bug". (Some claim this happens when you reaction fire the Blaster Launcher too, but I have never got it to work). After the shot, there is a "dud" shell still stuck in there. In order to use the Launcher again, you must unload this dud and then reload the weapon with another Blaster Bomb as you found out. :)

Hey, check it out - that inner radius=8 you drew for the Launcher Tank EXACTLY matches the predicted iso contour for distance 8, for that "walking TUs" sheet I made. (It's easy to highlight 8 if you use /Format /Conditional Formating, then change colors for value >8. Select the table first.) I didn't really trust that sheet before because radius=4 wasn't a whole lot to work with (lots of clipping and rounding with so little distance). But now I'm feeling a lot better about it, seeing that it matched 8.

<{POST_SNAPBACK}>

Oh, yeah. It does match. Cool! So you had a factor if 1.5 (6/4) instead of 0.67 (2/3), right? Makes sense, and it appears that it might just be the "holy grail". Good job! B)

- Zombie
The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#77 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 29 October 2005 - 09:23 PM

Oh, yeah. It does match. Cool! So you had a factor if 1.5 (6/4) instead of 0.67 (2/3), right?

Right. Strictly speaking, it's INT(MAX(X,Y)+MIN(X,Y)/2), where X and Y are positive offsets from GZ. I think this equals minimum walking TUs, although I haven't thought about it real deeply. At this point I'm thinking that even if we do see an exception, probably a tiny adjustment will fix it up.

Another thing not yet known is whether the launched explosives will have the same "distance pattern" as thrown ones. We know their blast patterns are unusual... so did the programmers hard code the unusual patterns, or is it truly calculated based on HE? I strongly suspect that it is hard coded, because of how it's the same for all those diameter 13 blasts. IOW they probably dittoed some fixed pattern.

If anyone can ever hack the HE for projectiles, it could be interesting to see if they still have the same outline.

I also wonder whether that little tip seen due lateral might be different. We know it did something weird in the blast pattern for Hi-Ex, where it was able to destroy the ground at the tip even though it should have been at HE=25 there (half strength for tile damage). I'd bet that that's only an error vs. tile damage though.

No rush on the Prox Mine. It's all good! There's plenty of untested terroritory. If we see a solid pattern with a few different explosives, though, I doubt we'll need to test them all.

#78 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 30 October 2005 - 07:37 PM

Okay, I'm fresh off a testing spree. If you remember, I decided to gather numbers for the Proximity Grenade. Dummy me. I forgot about the bug which prevents a proxy from exploding if the game is saved. This forced one of my soldiers to prime and throw the grenade while an "overseer" (255 health and armor) set the thing off for every trial. Because the anticipated damage range for the proxy at ground zero is 71 different values, I figured that doubling the number of trials would be sufficient (142+10 more for good luck = 152). It was. :)

As always, the damage range for each tile follow my explosive calculations:

Tiles from GZ      Damage Range
      0              35 - 105
      1              30 - 90
      2              25 - 75
      3              20 - 60
      4              15 - 45
      5              10 - 30
      6               5 - 15
And of course, you are all waiting for my spiffy screen capture of the results, so here it is:

Proxy_Grenade_Damage_Shells.png

Unfortunately, it does not follow the anticipated damage contours for a diameter of 13. Where it breaks down is through the center (just as I suspected) and near the diagonals. The reason why I suspected the actual center would not follow the anticipated is this: just from the grenade tests I saw a blast "corridor" going right through ground zero in a + pattern:

   444
    333
    222
432111234
432101234
432111234
    222
    333
    444
This 3-tile "corridor" continues for the proxy grenade too. Except there are 3 - fives at each end. I suspect this corridor effect is true for all the grenade types on desert sand. But, as usual "more testing is necessary to verify". :)

- Zombie
The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#79 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 30 October 2005 - 09:10 PM

Hella job! I didn't know that the proxy won't explode if you reload...

Lovely pattern... thanks for the colors! :)

Unfortunately, it does not follow the anticipated damage contours for a diameter of 13. Where it breaks down is through the center (just as I suspected) and near the diagonals.

When you say anticipated, you mean the walking-TU contours?

If so, I only see it breaking down at the ends of the big plus sign. There should be three 6s at each of its four ends, but there's only one. The diagonals match the TU contours. The only discrepancy I see is those eight 6 tiles next to the "tips". I must be missing something?

Two things strike me:

1) In "immediate hindsight" I realize that we already knew something funky was going on with the outline - more on that in a sec. That's the bad news. The good news is

2) Every tile within the outline is following the TU contours. (Unless I'm missing something.)

We know they've jimmied the outlines for many explosives. My screw-up with the Hi-Ex shows that even changing HE doesn't change the outline (or at least it didn't in that one example). So I'm thinking that they may have "forced" an outline for all - or almost all - blasts. In fact I now think there's only one where they don't seem to have (the grenade). Some are more obvious than others. To me, your findings are suggesting that 1) outlines are not predictable based on HE or radius (but are known), and 2) TU contours still seem quite viable.

Another way to put it is that 1) they are either doing something funky with the math that makes it slightly different from the TU map, or 2) they hardcoded the outlines - and incorrectly, one could say, but hey it's their game... it's still a great game even if they cut some physical-reality corners due to integer math etc. Given those two choices, we know there are many clear examples of plain old hardcoding the outlines. Or at least, basing it off of something besides simple "decrease to 0". So why not here with the Proxy too, even if it only just barely makes it different from the TU contours.

By the way, I did Hi-Ex in desert and fixed my blast pattern map. Its inner (destroyed) tile radius precisely matches your "5" contour.

I'm just as interested in one of the small blasts with the super funky outlines, like AC-HE, as I am with the Blaster.

You're right, I totally missed things like dead and unconscious folks :doh: I knew there must be more... and it was literally standing right in front of me the whole time LOL

FWIW I played tag with the cyberdiscs on Mars. Based on their explosion radius (diameter 9), I estimate their explosive strength at GZ at 120 average, 60 min. The blast pattern appears to be snipped at diameter 13/radius 6, although that's hard to tell on Mars (tile armor=40). Also I get the impression only one quadrant is exploding. The wiki says they are equal to High Explosives when killed... I wonder how accurate that is? Hmm.

Edit - oops, double post. The site is giving me error messages tonight :(

[Edit, Zombie: don't worry about it. Posts merged.]

#80 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 30 October 2005 - 09:58 PM

Another way to put it is that 1) they are either doing something funky with the math that makes it slightly different from the TU map, or 2) they hardcoded the outlines - and incorrectly, one could say, but hey it's their game... it's still a great game even if they cut some physical-reality corners due to integer math etc. Given those two choices, we know there are many clear examples of plain old hardcoding the outlines. Or at least, basing it off of something besides simple "decrease to 0". So why not here with the Proxy too, even if it only just barely makes it different from the TU contours.

<{POST_SNAPBACK}>

So what do you think? Are blasts hardcoded, or is the game's function for determining blasts a little messed up? I'm leaning to the hardcoded choice at the moment.

I'm just as interested in one of the small blasts with the super funky outlines, like AC-HE, as I am with the Blaster.

<{POST_SNAPBACK}>

Well, you are in luck. I just started tests on the Auto-Cannon HE shells. We must be on the same wavelength! Give me a day or so to get it up. :)

FWIW I played tag with the cyberdiscs on Mars. Based on their explosion radius (diameter 9), I estimate their explosive strength at GZ at 120 average, 60 min. The blast pattern appears to be snipped at diameter 13/radius 6, although that's hard to tell on Mars (tile armor=40). Also I get the impression only one quadrant is exploding. The wiki says they are equal to High Explosives when killed... I wonder how accurate that is? Hmm.

<{POST_SNAPBACK}>

I'd still like to get the Cyberdisc on a desert landscape for some more, um, "tests". :)
120 average damage is 180 max. Impressive blast, that's for sure!
Right, only one quadrant is responsible for the explosion. That's due to the fact that the turret is located only in one quadrant. (Not to mention it's collision detection circuitry). I wonder how those aliens cram all that stuff in 1 tile... :D
As for what the wiki says about the explosive diameter: it was obviously a guess by someone. I was going to earmark that part since it hasn't been researched, but left it alone since we are pretty thorough in testing out all the explosives. :wink1:

- Zombie
The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#81 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 30 October 2005 - 10:45 PM

I'm leaning to the hardcoded choice at the moment.

Sorry I wasn't clear - that's what I get for rambling all over the place, laugh.

I definitely side with the hard coded idea. I can't think of anything that contradicts it. (Yet! - fingers crossed!) Go with the simplest explanation unless something else presents itself. You could even muse that they accidentally didn't hardcode the grenade (and/or they hardcoded it, but it was the only one that happened to match what you'd expect).

Sounds cool on the AC-HE! Are you only using a "slice" of the pattern? They should be symmetrical, so this can greatly reduce testing time.

At your excellent suggestion (cough) ;) I tested unconscious and dead unit blast survival. Turns out Flying Suit guys survive a bigger blast than most stuff. WTG home team! :cheerleader: I wonder how many times players have cringed at a blast only to find their unconscious soldier was the only thing left in the tile?

FWIW that second test was in desert (the first was on Mars). They're both consistent.

Some time I'll double-check the cyberdisc's explosion and revise the wiki, if it is wrong. I only glanced at a few cyberdisc explosions.

Ah, attached is the revised blast patterns (with HE corrected). I also kept the TU contour as another tab, and also included a little blast front table I use.

That's about it for me for the night. Thanks again for all the work!

- MTR

Attached Files



#82 NKF

NKF

    Commander

  • [Global Moderators]
  • PipPipPipPipPipPip
  • 4,798 posts

Posted 30 October 2005 - 11:16 PM

Zombie: Here's a thought regarding testing the proximity mines:

As all HE weapons differ only by their blast strength (and they are triggered in different ways), you could simply modify the obdata entry for a standard grenade to deal the same damage as the prox. mine in order for you to run your tests. It would certainly be a lot easier than having to arm a mine and having your invulnerable unit set it off over and over again.

In fact, you could do the same to test the strengths of all the HE weapons in the game - and perhaps even the strength levels of TFTD's explosives. Or even run gradual tests from 10 to 250 to see the changes in the blast patterns. Anything goes.

Just as a reminder: geodata\obdata.dat has 54 columns. Weapon strength is in column 23 and ranged-weapon damage type is in column 32. If you fiddle around you'll probably recognise some of the easy to identify weapon stats. Remember to make a backup copy of obdata.dat before you start fiddling so you can restore it later on.

- NKF

Edited by NKF, 30 October 2005 - 11:24 PM.

Lord High Generalissimo Ruler Supreme of Norm's Anti Pedant Society (NAPS).

Number of members: 1

#83 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 31 October 2005 - 12:57 AM

Good point about obdata.dat, NKF. I actually thought of that during my grenade tests but dismissed it because I don't know where everything is stored in that file. Besides, if I start fooling around in that file there could be some dire consequences if not done correctly. Hmm... how come the grenade does 70 average damage all of a sudden? ;)

Do you have any notes on obdata.dat, NKF? I tried looking for it on my computer and if I did have it, its long gone now. Anyway, if you have anything useful, PM me with it. Otherwise, I can muck around in there and hope I get it right. *gulp* :puchmonitor:

- Zombie

[Late edit: thanks NKF!]

Edited by Zombie, 31 October 2005 - 08:25 AM.

The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#84 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 31 October 2005 - 11:22 AM

A quick note... I have to do some "real" work now... but Danial just tore OBDATA.DAT apart and posted a bunch of stuff to the wiki.

This includes where he found the field for object destruction which is much more refined info than my testing... more to come on that; also see end of Discussion here...

Also, interestingly, object weights which don't always seem to match what's been put in the wiki Item Weights table...

have to get back to work atm...

#85 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 31 October 2005 - 06:28 PM

I spent most of the day working on OBDATA, with a lot of inspiration from Danial. It seems to be in great shape now. Still a few questions though... see that wiki link if you've got notes please, NKF! e.g. there seem to be two closely related weight(?) columns where one is identical to Item Weights on the wiki, and the other is occasionally less... maybe throwing weight versus inventory weight??

I would be a little leery of playing with values too... at least until we think we've got unhacked stuff pinned down. After that, some spot checks against hacked stuff could be fun. Right... what do you do if you see something weird with a hacked game? Hmm. It's still a good thought NKF... and you're probably a lot more confident with hacking those values than I am ;)

Actually come to think of it... I might be leery of doing it when spending a lot of testing time, as Zombie does... but to simply see tile-damage patterns, what the heck. I might give it a spin. That doesn't take so much time. I could even tailor it to try to see contour edges. Z - what do you think the outline will look like if I set a Blaster to HE 120? ;)

Which reminds me of another thought... could they possibly have hard coded explosion outlines because they go wonky past the edge of where their strength trails off? It's only a thought but... One way to find out!

I polished off the wiki section on equipment destruction due to blasts. Very complete now.

#86 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 31 October 2005 - 08:59 PM

I have no qualms with editing weapon damage in obdata.dat. It's not that difficult to do if you remember the offsets correctly. The main thing to remember is to archive an unedited copy of obdata somewhere in case you mess something up by accident. (Hence why NKF preaches this so often). For now, I'm going to keep obdata the way it is until I finish testing normal explosives. After that, I have a couple of ideas to test out concerning varying strengths. :)

The outline of a BB doing 120 average damage, right Mike? Dunno. It may depend on if the game categorizes HE ammo according to class. The three classes should be:
  • Thrown HE (any type of grenade)
  • LOS Fired HE (HWP Rockets, RL Rockets, AC/HC HE ammo and perhaps Stun Bombs) LOS = Line Of Sight
  • Guided HE (HWP Fusion Bomb and Blaster Bomb)
Thrown and guided HE appear to have the same pattern - except the guided bombs may have a hard-coded size limit.

----------

I too have been busy completing the tests on the Auto-Cannon's HE ammo with 44 average damage. Here again, the damage ranges follow the predicted 10 point drop-off per tile on the average:

                  Damage Range
Tiles from GZ      Min-Ave-Max
      0              22-44-66
      1              17-34-51
      2              12-24-36
      3               7-14-21
Even though the blast only extended out to GZ+3, there should be a theoretical GZ+4 having a damage range of 2-4-6. I didn't see it. Perhaps average damages of less than 10 are ignored. Though, more tests are necessary to verify this.

I know, I know. Here is the screencap of the observed shells for AC-HE 44 ammo:

AC_HE_damage_shells.png

Up next, I'm going to do a test of the HC-HE 52 ammo to verify if it follows the same pattern. That shouldn't take too long to complete if I continue to follow my rough guide of collecting two times the range. In this case 52*2=104. That’s nothing! I can do tests like that in my sleep. =b

- Zombie

Edited by Zombie, 31 October 2005 - 09:03 PM.

The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#87 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 31 October 2005 - 09:43 PM

Very cool! It was one of the weirdest. If it's still following the TUs pattern, there's hope for them all :)

I'm thinking that all explosives, with the possible exception of the grenade, follow a hard-coded outline. I suppose we'll see soon enough when we start playing with HE levels...

I just took BombBloke's mountain map for a test drive. Really neat. Updated the wiki on that effect.

FWIW I have tried a number of bombs on several different maps (old savegames) at this point. They are all producing the expected outlines for injuring and destroying ground. This includes inside alien bases.

Maybe it's time for me to look at Hobbes' HE Block map...

Edited by MikeTheRed, 31 October 2005 - 09:44 PM.


#88 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 01 November 2005 - 05:37 PM

As promised, I finished tests on the Heavy Cannon's HE 52 ammo in the desert. The radius affected is identical to that of the Auto-Cannon: 3 tiles past ground zero, even though the damages are radically different.

I ran through approximately 125 trials for each soldier inside the blast radius, and all of them fell within the predicted damage levels:

                  Damage Range
Tiles from GZ      Min-Ave-Max
      0              26-52-78
      1              21-42-63
      2              16-32-48
      3              11-22-33
However, there should be two additional damage shells: one at GZ+4 with 6-18 points, and another at GZ+5 with 1-3 points. Neither were apparent after placing soldiers in the appropriate locations.

Since the damage contours for the HC match the AC, I skipped this step. So if you want to see a pretty picture, see my last post and just imagine it is the Heavy Cannon. :wink1:

I think I might check out the High Explosive next. It seems like an interesting explosive because the range affected is smaller than expected. After that, the Stun Bomb and the Blaster Bomb need a little attention. :P

- Zombie
The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#89 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 01 November 2005 - 07:17 PM

Wow, 125 trials! That's some serious stuff!

It's great to see current theory being consistent with all explosives so far. The idea of fixed outlines and TU distances seems to be working great. Still more testing to be done, of course!

The stun bomb - now that ought to be interesting. It hadn't even occurred to me to test it per se... but... why the heck not?

Note that both the stun bomb and the alien grenade are rated for 90 - but the SB is diameter 11 and the AG is D13. We can only guess that this is also due to more semi-arbitrary outlines afoot. Any way you slice it though, it'll be cool to see the results. And again you've seen something that I was entirely overlooking.

I'm sure you haven't forgetten that all XCOM armor imparts some stun resistance.

For what little its worth, I consider the hits experience counter (UNITREF[81]) the most sensitive indicator of "any hit no matter how small, even if it didn't register" due to armor. Your testing also picks up any hit, no matter how small - but because you are courageously doing so many samples, plus have hacked your guys to 0 armor. So it's irrelevant for you, but maybe not to other readers... If they don't care about an actual damage estimate per se, and simply want to know/do a quick check of whether something was within the blast radius without hacking etc., U[81] is the way to go. Exactly one explosion is all it needs. No matter how small the blast strength; even if nothing registered due to e.g. 5 armor blocking the minimum possible (5 damage) on their one test. It does have to be on un-MC'd aliens, but that's not a big deal.

FWIW this is how I originally estimated outlines. With no repetitions, except to check different spots =b

As I wrote that, I just realized - that although in theory some explosions could've been less than 5/10/15 (min/ave/max) at their edge, in fact, due to outline curtailment, this is the real-life minimum. It is not possible to be hit by less than this. (Not counting armor and resistance, of course.) A bit of trivia, I suppose, is that for a soldier in FS/PS hacked to 0 armor, it would actually be 4 (80% modifier). Anyway... due to outline curtailment, it also means that it is impossible for electroflares to survive anywhere in an unhacked blast... interesting how they chose "8" for it, eh? :P Also, desert sand and farmland grass (both 5 armor) will always be injured from any bit of (unhacked) blast.

Your damage ranges match the little AllProp tab of my recent xls update (message 81) when 52 is plugged in. Everything's looking great. BTW you don't really need to do that shaded diagram any more, at least not for my sake... they do look cool but, as long as you tell me it matches the expectations (TUs map, decrease by 10, stays in expected outline), I'm more than good :) But if you want to, don't let me stop ya! ;)

I'm going to look at the wiki, maybe work on OBDATA a little more, then keep working on Hobbes' 3rd desert map. Unfortunately I misunderstood him - all this is, is desert with varying tilesets (10x10) set to varying armor levels (10, 20, etc.). No HE Block at all :( As long as I've gotten to the point of testing, I guess I'll test, eh? We expect no surprises. (Famous last words?)

#90 Hobbes

Hobbes

    UFO2000 Staff

  • Moderators
  • PipPipPipPip
  • 719 posts

Posted 01 November 2005 - 07:59 PM

I'm going to look at the wiki, maybe work on OBDATA a little more, then keep working on Hobbes' 3rd desert map. Unfortunately I misunderstood him - all this is, is desert with varying tilesets (10x10) set to varying armor levels (10, 20, etc.). No HE Block at all :( As long as I've gotten to the point of testing, I guess I'll test, eh? We expect no surprises. (Famous last words?)

<{POST_SNAPBACK}>


There's 3 hacked terrain versions:
- Desert.zip has the ground hacked to different levels of HE block, no armor, ranging from 0 to 120
- Desert2.zip has the terrain hacked to different levels of armor/HE Block ( armor/HE block have equal values on each different ground tile), same ranges. Both Desert and Desert2.zip are available on the strategy core forum through the links i posted.
- Desert3 (the one i emailed you), has no HE block, different levels of armor, same ranges as above.

Is that enough for your testing? :)

#91 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 01 November 2005 - 08:47 PM

Don't get me wrong, Hobbes - I greatly appreciate the time spent. In hindsight I now see that I misunderstood the difference between the second and third set (until I got it set up then checked it with MapView). I probably would've gone for the second if I'd realized it. However, now that I've got the third ready to go, what the heck, it's still good! At this point I'm thinking we've got armor itself fully understood but you never know what you might find ;) ... your tiles present a lot of variation, all in one place.

A question if I could: Now that you've made the 10x10 .MAP tilesets, do I only have to directly edit the MCD if I want to change things such as armor, HE Block, and death tile? That is, it's independent of the .MAP files, right? The tilesets you've made are fine; nice big swathes for testing things... it's just those armor/HE/death values that I'd like to play with. And I think I understand the structure of MCD - fingers crossed. (But I don't understand the MAP files yet.)

#92 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 01 November 2005 - 09:11 PM

Seriously? 125 trials to me is like nothing. It's actually fun to test out explosives because up till now the ranges have been rather small. And small ranges means fewer trials need to be run to gain an understanding. :)

The stun bomb - now that ought to be interesting. It hadn't even occurred to me to test it per se... but... why the heck not?

Note that both the stun bomb and the alien grenade are rated for 90 - but the SB is diameter 11 and the AG is D13. We can only guess that this is also due to more semi-arbitrary outlines afoot. Any way you slice it though, it'll be cool to see the results. And again you've seen something that I was entirely overlooking.

I'm sure you haven't forgetten that all XCOM armor imparts some stun resistance.

<{POST_SNAPBACK}>

Yeah, I figured I'd might as well test the Stun Bomb to see exactly what is happening inside the blast radius. We pretty much have it figured out already, but it still is a good idea. I might even test it on the ground and in the air to see if ground plays a role in determining blast radius. (I'm not talking about damage to a unit, only the pattern on the ground).

All X-COM units have armor, but only units wearing manufactured armor have resistance to stun as I'm sure you know.

For what little its worth, I consider the hits experience counter (UNITREF[81]) the most sensitive indicator of "any hit no matter how small, even if it didn't register" due to armor. Your testing also picks up any hit, no matter how small - but because you are courageously doing so many samples, plus have hacked your guys to 0 armor. So it's irrelevant for you, but maybe not to other readers... If they don't care about an actual damage estimate per se, and simply want to know/do a quick check of whether something was within the blast radius without hacking etc., U[81] is the way to go. Exactly one explosion is all it needs. No matter how small the blast strength; even if nothing registered due to e.g. 5 armor blocking the minimum possible (5 damage) on their one test. It does have to be on un-MC'd aliens, but that's not a big deal.

<{POST_SNAPBACK}>

When taking armor into consideration, sure, the hit counter is more accurate to determine hits. However, without armor my tests are just as accurate. Units having a susceptibility to HE of 100%, and 0 armor are actually excellent indicators of hits. That coupled with a minimum damage > 0 and a small range are probably perfect testing conditions.

Your damage ranges match the little AllProp tab of my recent xls update (message 81) when 52 is plugged in. Everything's looking great. BTW you don't really need to do that shaded diagram any more, at least not for my sake... they do look cool but, as long as you tell me it matches the expectations (TUs map, decrease by 10, stays in expected outline), I'm more than good :) But if you want to, don't let me stop ya! ;)

<{POST_SNAPBACK}>

Actually, those diagrams are more for me than anything. I need an actual diagram to compare against the anticipated table to verify that everything holds. And I figure that as long as it's handy, I'll post that too. :)

- Zombie

Edited by Zombie, 01 November 2005 - 09:12 PM.

The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#93 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 01 November 2005 - 09:28 PM

In the air? Again, I never thought of that! You must mean for projectiles eh? No way to get grenade types to explode in air, is there? Although... you could blow a grenade by the edge of a UFO roof and have guys off to the side (in midair) I suppose. Hmm... what's something with exactly 1 "roof" tile, to suspend it in midair, I wonder? hehe

Working through Hobbes' 3rd map now

#94 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 01 November 2005 - 10:30 PM

Everything on your map checks out, Hobbes... It was straightforward, which got me bored, which got me trying a hack at MCD and ... tada! =b

MCD can even be hacked between reloads; no need to exit XCOM. Now I can try a number of things... Thanks very much for making the tileset!! :Hug:

#95 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 02 November 2005 - 12:56 AM

OMFG!

The desert MCD has 13 ground tiles.

I gave each of them 5 armor.

Then I linked them sequentially as death tiles:
  Tile 0 dies -> Tile 1
   Tile 1 dies -> Tile 2
   Tile 2 dies -> ok you get the idea
Then I set off a blaster in the middle.

Blam. A painted picture of the damage level every single tile received. Tile 0 = 5 damage, 1 = 10, 2 = 15, etc.

Apparently the code works all the way through tile chains until all damage is done. In fact I'm sure I could've wound around in a circle if I wanted.

Attached, find a picture of Before and After, when setting off a blaster bomb.

For this, I actually made Tile 0 = 45 armor, and the rest of the chain 5 - otherwise the blaster has enough strength to blow all the way through the stack of 13 (65 armor total; blaster does 100 tile damage at center).

The blaster does 45 damage at the edge. So by setting tile 0 (the surface tile) to this, the blaster cut through to tile 1 at its edge, tile 2 one radius inward, etc.

The "tile key" for the picture is as follows. IR=Inward radius (counting backwards), OR=Outward radius:

 IR  OR Dam  Tile
   1  11   0  Human skull
   2  10   5  Small sage (dot)
   3   9  10  Cow skull
   4   8  15  Straightish snake
   5   7  20  Snake curved down at end
   6   6  25  Driftwood branch (bone?)
   7   5  30  Cactus
   8   4  35  Large sagebrush
   9   3  40  "S" snake
  10   2  45  Scuffed moving to upper left
  11   1  50  Furrowed scuff
  GZ  GZ  55  Central scorch
Add 45 to all damage levels - tile 0 was arranged to reveal what was underneath. So "Damage 0" actually means "damage 45" (thus showing the first death tile past tile 0).

The map I played this on does not have one nice big place with tile 0, unfortunately. This spot next to a UFO was the best I can find. I blew the blaster off right in FRONT of the soldier. (He moved away a sec!) The cursor box is at the edge due lateral (GZ+11). Notice that all the surrounding tile was blown away because it didn't have the tile 0=45 "hard cover".

It is entirely following the TU contours except that it's nipped a little at the edge. Not just on the ends of the plus sign, but a little more than that... if you compare the TU contour from my spreadsheet (attached to message 81 here) to the blaster outline pattern in it, you see where the other nips are. IOW, we knew this already if only we'd looked =b

Zombie, I think your work just got a lot easier :) At this point, I would think that all that really needs work is spotchecking things, now... but I still need to study each blast to make sure they're ok.

The stun bomb will still need work - and might turn out to be real interesting. Or might not - which is also real interesting! (that it follows the same pattern as HE blasts)

Note that one can specifically tailor the tile chain and armor values in order to detect even 1 hitpoint differences to tiles, etc. (although it'd be a little work) Keep in mind that you can edit MCD on the fly... all you have to do is reload the same savegame. (You can't change the tileset, but that's ok.) Hey it just occurred to me that I can set all other tile's armor to 0 and death tile to tile 0 (armor 255), then shoot a few bombs and basically repaint the whole map to plain terrain, LOL... man there's a lot of tricks that could be done. One could conceivably make circular chains and stuff, too. BombBloke even pointed out that you could even make "gates" that wink on and off, if one of the tiles has TUs set high (see offset 42).

Want a copy of this game? In addition to the savegame, you'll have to replace your desert terrain files before using this savegame, too. I can bundle them in there.

Attached Thumbnails

  • BlastedSequentialTiles1.jpg


#96 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 02 November 2005 - 09:39 PM

Z, BombBloke has way one-upped me and posted a tileset that has 100 tiles that SAY 0 - 99 on them... now you can directly read damage from the ground! See this thread but read it all way through... there's more after that... BTW I've already verified that everything uses the TU contours! Next up... shot weapons!!

#97 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 02 November 2005 - 10:55 PM

Very interesting. That should reduce tile testing times by a bunch! =b

Anyhow, I ran through about 50 or so HC-HE trials today with all my men in the air instead of on the sand. Preliminary results are indicating that the damage range is exactly the same. Yeah, I know. I did test this out during my damage modifier trials, but I never checked each tile - only ground zero. :wink1:

An identical damage radius on the ground or in the air suggest that the radius may be predetermined to be a specific size based on the primary tileset on that mission. This is only theory yet, but how else would you explain different sized blasts on different landscapes? It's either that or the game just hard-codes a saved blast contour for that specific map. *shrugs* Perhaps you are better equipped to answer this due to your tests on landscape tiles. :)

- Zombie
The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!


#98 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 02 November 2005 - 11:18 PM

Sounds great on the HC-HE... air and stun are the two big remaining questions... there's one down!

I am not sure I am following your argument concerning radii and tilesets. Keep in mind that damage to units was never affected by tiles, as far as we know (although it definitely could use one or two good tests to make absolutely sure). This is, of course, for regular ground tiles; no HE Block or objects in the way. Example: A blaster in a UFO only hurts floor tiles to GZ+4. But it still readily kills units at the edge (GZ+11).

I never saw any reason to not think that outlines are always the same regardless of tileset. I'm also willing to bet that even if we hack HE values, they'll STILL have the exact same outline... but will otherwise correctly follow the decrease-by-5 rule within their outline.

Conversely, of course, different tilesets do have different armor, for the different tiles. UFO floors have armor 80, so only hurt to GZ+4 by a blaster (ave/2=100). The blaster was always doing ave 200 at GZ and ave 90 at GZ+11 (100 to 45 vs. tiles), but the ground tile armor is high inside a UFO. It's much lower in e.g. the desert.

Is that clearing things up or was I missing your point?

Danial just did a bunch of tests using BB's numerical tileset. Guess what:

Damage from weapons is ave/4 to 3/4*ave. That's why Hobbes was only able to shoot through a UFO wall when he hacked it to armor 85 but not armor 90... a heavy plasma does 115; 3/4*115=86. QED.

Danial has found it to be true for plasma, AP, and laser. He did not do rigorous testing like you, but spent an hour on it and never found an exception.

Back to the tileset... trying to monkey with HE Block now...

MTR

#99 MikeTheRed

MikeTheRed

    Captain

  • Forum Members
  • PipPipPipPip
  • 179 posts

Posted 02 November 2005 - 11:35 PM

By the way, here's BombBloke's latest greatest numerical-damage tileset, plus a savegame to get down to business immediately.

Totally rocking, it even color codes tiles according to damage level (sound familiar?) ;)

I also threw in a screencap for anybody that just wants to see what it looks like, without actually messing with their XCOM. The feller standing there just popped a grenade.

It also includes a little hack util BB made that lets you change armor, HE Block and explosive value for your top (initial) tile, and then the remaining chain of 99 (secondary tiles).

Place the files within each zip, into the respective XCOM directory. BE SURE TO BACKUP. This will replace your desert terrain etc.!

Also note, this tileset will NOT work with larger UFOs. The UFO tiles plus the 100+ BB made, choke the game. Not sure just how large is too large, but the savegame with medium scout works.

Attached Files


Edited by MikeTheRed, 02 November 2005 - 11:47 PM.


#100 Zombie

Zombie

    Colonel

  • Xenocide Creative-Text Departmen
  • 2,947 posts

Posted 03 November 2005 - 10:39 PM

Is that clearing things up or was I missing your point?

<{POST_SNAPBACK}>

Nope, I understand completely. Thanks!

Danial just did a bunch of tests using BB's numerical tileset. Guess what:

Damage from weapons is ave/4 to 3/4*ave. That's why Hobbes was only able to shoot through a UFO wall when he hacked it to armor 85 but not armor 90... a heavy plasma does 115; 3/4*115=86. QED.

Danial has found it to be true for plasma, AP, and laser. He did not do rigorous testing like you, but spent an hour on it and never found an exception.

<{POST_SNAPBACK}>

Wait, this was for normal weapons (not HE), correct?

Looks like the programmers are just reusing the explosive code and dividing by two. Guess this proved my theory that weapons do a range of damage and not always their max against objects. Very interesting. :jawdrop:

So what happens with high explosives? What kind of damage is that ammo producing? Or do we not understand the mechanics behind HE block and armor yet, oh great one? :hmmm:

- Zombie
The Mr. Grognard of X-COM

Zombie: Empirical data's your only man, when formulating a research plan.
A soldier's death is never in vain if it makes the formula more plain.
A few dozen make a better case for refining that third decimal place.
They call me Zombie because I don't sleep, as I slowly struggle to climb this heap,
of corpses, data points, and trials, but from the top - I'll see for miles!