Talk:UNITREF.DAT

If we have Height [49] AND "Float" [51], then their "real" height (in terms of profile you can hit) might be Height-Float. Combined with this potential Width ([48] and/or [52]) defines a cross-sectional target area a.k.a. percent of the space of a tile that they occupy, seen from the 'side'. If it does work this way, it's Area1 here:

               [49] [51] "Tall" [48]        | [49] [48]    
  Class         Ht.  Flt   H-F  Wide Area1  |  Ht. Wide Area2    Area1=(Ht.-Flt.)*Wide
  ---------     ---  ---  ----  ---- -----  |  --- ---- -----
  Sectoid       16    0    16    2    32    |   16   2    32     Area2=Ht.*Wide
  Celatid       12    6     6    3    18    |   12   3    36
  Hovertank*    12    6     6    4    24    |   12   4    48     Sorted on Area2
  Silacoid      10    0    10    5    50    |   10   5    50 
  Snakeman      18    0    18    3    54    |   18   3    54 
  Cyberdisc     15               4          |   15   4    60
  Ethereal      20    0    20    3    60    |   20   3    60    All Hover/Tanks in their
  Chryssalid    21    0    21    3    63    |   21   3    63    class assumed to be same
  Civilian      21    2    19    3    57    |   21   3    63    altho not tested yet
  Floater       21    2    19    3    57    |   21   3    63
  Muton         21    0    21    3    63    |   21   3    63
  Tank*         16    0    16    4    64    |   16   4    64
  Soldier       22    0    22    3    66    |   22   3    66
  Reaper        23               4          |   23   4    72
  Sectopod      23               4          |   23   4    72
  Zombie        18                          |   18       

With Area1, Hovertanks would be thin as a pancake!! So instead I tried simply Height x Width, shown in Area2. Much better. IOW, if a full tile is 24 "high", as Danial estimated, the Celatid starts at 6 and goes to 18 (6+12). Also note that with Area2, Civvies and Floaters add up to 23, the highest value seen for others. (And Dan - if counting starts at 0, you were right on the money with the 24!! Then again maybe they avoid 24 just to steer clear of ceiling collisions. Anyway.)

This whole concept could be roughly tested with some shooting... but it'd need a lot of repetitious work. I recommend very good accuracy (hacked soldier, best weapon; simply reduces # tests needed), fair distance so anything that's calc'd to miss really will, and then use ammo counts vs. experience counter to get percent hits entirely accurately. Choose some of the targets with extreme values (width, "float", area). Then kill them a TON of times, lol. Unfortunately, you can only use Exp. Ctr. for un-MC'd aliens, so no testing of tanks. I wonder what Cyberdisc's numbers are? But they can't be disarmed, wah. Actually anything Large shouldn't be used because you may hit other quarters, which would mess with comparing percents. Possibly even if you just compared large to large.

Another unanswered question is, if [48] and/or [52] is "width", HOW is it width? A scaler for "percent chance of hit", max 5, so Sectoid is 40%? Or they could've used cylindrical or square (actually cubic) percent-chance lookup tables or even actual "3D" models. (With such a simple geometry, it's conceivably do-able for them.) Hey - I wonder whether there are similar values for the scenery (trees etc.)... they get hit too... wth would these values look like for them, if so?

Hold the phone... great point, NKF ... hacking some or all of these four bytes to 0 (or something real high!) may tell us tons within a few shots. Or maybe even just by looking at'em! I personally am not interested in this ATM so if BB or somebody else wants to test, go for it. (Don't forget, [52] and [48] seem to be identical, so one might be unused.)

We could really use a (hacked/XCOMUTIL?) save game with ALL the different aliens, tanks, and soldiers vs. 4 armor types. Zombies too, wth. I count 22 types total. Multiples of each, to bring out "personal" vs. "class" bytes and bits. Disarmed if possible, so they don't kill all civs and soldiers by turn 2, laugh. Can't 80 units fit into a combat? if somebody makes this, pls. avoid the potential problem BB talked about re: Hobbes game (whatever it was!). Would this take forever or can anyone do this for us? Post as another Unitref Note, if you make one. I personally WOULD be interested in pulling data from this and posting it, like I did for Hobbes... I just got Unitref to where I can suck it in automatedly and query it; that's how I pulled his stuff so fast. And/or send me other savegames you'd like me to pull stats from! I like to play with data. (But don't necessarily like the endless repetition of in-game testing to get it!)

Anybody let me know if they want me to write a little applet where e.g. you press a button and it shows you the Experience Counters (or anything else) for a savegame. It wouldn't keep it locked. Thus you could move faster with e.g. shooting tests. I *think* I can do this in VBA and package it so you don't need MS Access... haven't tried before... so you may need Access. Anyway let me know if anyone's interested and I'll look into it.

Speaking of which, can somebody sooner or later post something re: the structure of the other XCOM savegame files? Currently I'm esp. interested in being able to hack weapons etc. into an existing game. But have no idea where/how that's stored.

Ethereals can fly... MC one and you'll see, although I can't recall seeing them fly of their own will. BUT they do have a 0 in [51].

---MikeTheRed