TA vet system.

[Xon]

Can anyone here tell me exactly how much the TA vet system improves units by? And to what stats.

[Boogie]

http://www.tauniverse.com/forum/show...hlight=veteran

That thread there is the best hint to TA vet bonuses I've seen... but it obviously leaves a lot out :/

[Xon]

Little digging into the TA exe shows the follow increases per number of kills you have;

• Unit capture time (no limit)
• Unit health (limited at 5 vet levels, )
• Weapon damage (limited at 5 vet levels)
• Weapon reload/accuracy (no limit)

I'm still working on reverse engineering the forumla used out of the exe. There appears to be roughly two versions used. One which applys every 5 kills, another for every 10 kills.

[zwzsg]

From this thread:

Originally Posted by DannyU

PeeWee veteran status versus time taken to kill a Doomsday Machine :

    Veterancy	Kills	Time

	0	0	54
	1	5	49
	2	10	45
	3	15	36	( no it's not a typo )
	4	20	36
	5	25	29
		30	29
		35	
		40	29
		45	
		50	29
		55	
		60	
		65	
		70	
		75	29
		80	
		85	
		90	
		95	
		100	29

[Firmament]

Is there any chance to see the same kind of quote for health, accuracy and capture bonus this time ?



( Let me dream, ok ? )

[Xon]

Extracting the formula is quite hard, not only is it in x86 asm and uses all types of wierd tricks, I need to convert it from it's actual formula into what it would really be (aka remove x86 warts of integer processing)

[DannyU]

Hello Xon, good to see you back.

Originally Posted by François

Is there any chance to see the same kind of quote for health, accuracy and capture bonus this time ?

Hmm, I could probably do one for health, capture time - meaning how long a unit takes to be captured based on its veterancy (not how long a Killer-Necro takes to capture) - yes I could do that too. Weapon damage would be tricky, accuracy - hard to say but maybe (it's randomized I presume), reload yes, these three parameters would have all affected the data for the original table (which is a bit odd because if there's no veterancy limit for reload some improvement might have been expected lower down the table - although if the improvement is small, bearing in mind a +/- 1 second accuracy should be allowed, maybe not so odd).

Now if Xon can convert the formulas he could produce a much more accurate set of tables... and since he's doing that anyway .

Would such tables be much use to you Xon?

[Xon]

Originally Posted by DannyU

Hello Xon, good to see you back.

(which is a bit odd because if there's no veterancy limit for reload some improvement might have been expected lower down the table - although if the improvement is small, bearing in mind a +/- 1 second accuracy should be allowed, maybe not so odd).

Trying to match which stat the veterancy is applied to an how is an utter horrible thing. I probably mistaked a few things.

One thing I can say for sure; When firing a weapon, the veterancy level is applied to at least 2 stats related to weapons. One is capped, the other isnt. I cant claim to have nailed all of the referances down, I think I'll need to breakout TSearch to hund down accesses to it.

Now if Xon can convert the formulas he could produce a much more accurate set of tables... and since he's doing that anyway .

If I can figure out the general shape, I'll be able to figure out WTF the code is actually doing and produce an exact formula.

Would such tables be much use to you Xon?

Heck yes

Make sure you do these tests on a map without any windspeed so it cant affect the results(plasma weapons aka inaccurate ones). Also using a modified unit with nice numbers for stats might be a good idea.

[DannyU]

Veterancy vs ReloadTime	101 shots = 100 reloads*, reload = 1s
			*first shot is virtually instantaneous

    Veterancy  Kills   Time(s)

	0	0	107
	1	5	100
	2	10	94
	3	15	87
	4	20	80
	5	25	77
	6	30	77
	7	35	77
	8	40	77
	9	45	77
	10	50	77
	11	55	-
	12	60	-
	13	65	-
	14	70	-
	15	75	77
	16	80	-
	17	85	-
	18	90	-
	19	95	-
	20	100	77

Commentary: The standard reload value appears to be valid at veteran level 1 (not 0). Between level 0 and 4 these values appear to be modified linearly, the modifier at level 5 deviates noticeably so I'll probably do a graph based on kills (rather than the assumption that 5 kills = 1 veteran level) and see if it can illuminate this portion of the table.

[Boogie]

I wonder if it bumps up again at 125 kills (5^1 being a significant boost, 5^2 being a significant boost, maybe 5^3 is a boost as well)

[DannyU]

I filled in a few blanks but they don't reveal anything new.

Veterancy vs ReloadTime	101 shots = 100 reloads*, reload = 1s
			*first shot is virtually instantaneous

	Kills	Time(s)

	0	107
	1	107
	2	107
	3	107
	4	107
	5	100
	6	-
	7	-
	8	-
	9	-
	10	94
	11	-
	12	-
	13	-
	14	-
	15	87
	16	-
	17	-
	18	-
	19	-
	20	80
	21	80
	22	80
	23	80
	24	80
	25	77

I don't follow your reasoning Boogie but I checked it, nothing new again.

Veterancy vs ReloadTime	101 shots = 100 reloads*, reload = 1s
			*first shot is virtually instantaneous

	Kills	Time(s)

	125	77
	126	-
	127	77
	128	77
	129	-
	130	77

[DannyU]

I did this as an off the cuff spot check because I see comments from time to time wherein it is suggested that range is affected. This is not an accurate test but there's no evidence for that effect here (but such an effect might not apply to all weapon types so some more tests may be done).

Veterancy vs Range	rendertype 0, beamweapon, range=300
			visual comparison (range was not measured)

	Kills	Range

	0	300
	130	300

[macraig]

Originally Posted by Xon

Make sure you do these tests on a map without any windspeed so it cant affect the results(plasma weapons aka inaccurate ones). Also using a modified unit with nice numbers for stats might be a good idea.

Here's a TA:M mutator that will force the wind speed to zero on ALL your maps, so you can otherwise pick and choose for your testing.

[DannyU]

Veterancy vs Damage	default damage=100, target maxdamage=10000

    Veterancy  Kills   Shots To Kill

	0	0	100
	1	5	95
	2	10	90
	3	15	85
	4	20	81
	5	25	77
	6	30	77
	7	35	77
	8	40	77
	9	45	77
	10	50	77
	11	55	-
	12	60	-
	13	65	-
	14	70	-
	15	75	77
	16	80	-
	17	85	-
	18	90	-
	19	95	-
	20	100	77
	21	105	-
	22	110	-
	23	115	-
	24	120	-
	25	125	77
	26	130	77

Commentary: Notice any eerie similarities between this and Table1? Once again the values appear to be modified linearly but with a kink at level 5 (possibly at level 4, the fact that default values are at level 1 in Table1 but at level 0 here may be relevant).

Thanks mac but that mutator doesn't work for me... er, you know , because I don't use TA:M (besides I have only one map to modify, these tests are carried out in campaign mode).

[Xon]

The equation for damage buffs appear to be; 6*<vet level>+<Initial Damage>
And reloading; -6*<vet level>+<Initial reload time>

To test damage buff equation, use a 100 damage weapon with a 2500hp target.
It should take 25->20 hits to kill it (decreasing by 1 hit per vet level).

[DannyU]

Veterancy vs Health		default damage=100, target maxdamage=10000

    Veterancy  Shots To Kill

	0	100
	1	105
	2	109
	3	114
	4	120
	5	125
	6	-
	7	-
	8	-
	9	-
	10	125
	11	-
	12	-
	13	-
	14	-
	15	125
	16	-
	17	-
	18	-
	19	-
	20	125
	21	-
	22	-
	23	-
	24	-
	25	125

Commentary: OK, so while the "kinks" in the previous tables are illusory (being a product of inverted values, i.e. 1/x) this table does have a kink which also kinks back to the linearly expected values (a true kink, it's not clear to me why such an effect would occur). Anyhoo, capture time will be next (but requires a bit of preliminary work - unless somebody knows which factors determine capture time), I still haven't decided on how I might reliably test accuracy so that's last to do .

[DannyU]

Originally Posted by Xon

The equation for damage buffs appear to be; 6*<vet level>+<Initial Damage>
And reloading; -6*<vet level>+<Initial reload time>

To test damage buff equation, use a 100 damage weapon with a 2500hp target.
It should take 25->20 hits to kill it (decreasing by 1 hit per vet level).

I don't understand your nomenclature so bear with me - this is what I get:

Weapon Damage Modifier

Damage(M) = Damage*(100+mL)/100		where m=the modifier=6
					and   L=the veterancy level (0,5)

Weapon Reload Modifier

Reload(M) = Reload*100/(100+m(L-1))	where m=the modifier=7.5
					and   L=the veterancy level (0,5)

Or

Reload(M) = Reload*80/(80+m(L-1))	where m=the modifier=6
					and   L=the veterancy level (0,5)

But Not

Reload(M) = Reload*100/(100+m(L-1))	where m=the modifier=6
					and   L=the veterancy level (0,5)

This does not fit the values.


Unit Health Modifier

Health(M) = Health*(100+mL)/100		where m=the modifier=5
					and   L=the veterancy level (0,5)

Ignores the kink in the values.

[Xon]

Sorry about that; the equations are:

Weapon Damage Modifier

Damage(M) = Damage(I) + mL 		where m=the modifier=6
					and   L=the veterancy level (0,5)
					and   Damage(I) is initial damage

Weapon Reload Modifier

Reload(M) = Reload(I) + mL   		where m=the modifier=-6
					and   L=the veterancy level (0,5)
					and   Reload(I) is initial Reload time 
					in milliseconds (stored in 
					the weapon's tdf as seconds)

Unit Health Modifier

Health(M) = Health(I) + mL   		where m=the modifier=5
					and   L=the veterancy level (0,5)
					and   Health(I) is initial hitpoints

The hitpoint and damage buffs are damn tiny, while the reloadtime is quite significant!

The Reload time figures you have are too big for the value stored in seconds, and I've seen the value is also way to large in memory, so I assume it gets converted from seconds to milliseconds internally when loading the weapons info.

[Xon]

Originally Posted by DannyU

Anyhoo, capture time will be next (but requires a bit of preliminary work - unless somebody knows which factors determine capture time), I still haven't decided on how I might reliably test accuracy so that's last to do .

Capture time is directly propertional to the health of a unit, worker speed doesnt factor into it either. So it should be identical to health value

[zwzsg]

Here is a unit displaying the time between the two last fired shots, in tick. Maybe I should have used a longer reload weapon, such as a BLoD, instead of a HLT, to have more meaningfull values.

Built in all four kbots labs, for free, and invulnerable to all but dgun.

Edit: First remark, an HLT goes like: 30->28->26 with each 5 kills, while a BLoD goes like 150->140->132->123->114->105->105->105->105. So the reload decrease is proportionnal to initial reload, which is against your equations.

[zwzsg]

It is peculiar that no one ever noticed until now, but:

Reload time change with health!

Without scoring any kill:
200% health -> reload is 120 tick
100% health -> reload is 150 tick
50% health -> reload is 164 tick
10% health -> reload is 177 tick
0% health -> reload is 180 tick

It is known that veteran unit have a higher max hitpoint. Therefore I theorise that the veterancy doesn't direcly increase the firerate, but instead, it increase the hitpoint which increase the firerate.

Also, if instead of a los weapon, I use a ballistic weapon, such as the guardian gun, then, as soon as health is overloaded to as low as 101%, the gun goes bersek and fire in random direction! (Not only random heading, but also random pitch). Lowering health back restore normal operation of the gun. I guess it gets negative accuracy or something. TheRegisteredOne is telling me it's well know and that I am not the sole discover of this phenomenon, but I'll still anigif it: here is what happens when a guardian gun has its health overloaded to 500%.

To the previous unit, I added another unit that:
- Fire an annihilator BLoD
- Is instabuild, but cost 1000 metal so that reclaiming and healing is low. (Yes, reclaim speed depend on metal cost, not build cost)
- Is equipped with a damage inversal device, toggleable with the on/off button.
- Display time between two last shots, "get HEALTH", and killcount if Xon's dll is here.

To adjust its health to whatever value, make it be reclaimed by fark, then turn it on for health increase, and off for health decrease.

To rack kill, well, make sure the AI has built stuff, then, with a BLoD, murder is easy.

Don't forget to let it fire twice in a row to read the reloadtime.

Also, first unit to sport three ten digit signed digital counters!

[Xon]

Ok, I have found several locations which make use of the vetern level.

49D6E0 - Appears to have something todo with determining the accuracy of a unit. (I think, there are 2 calls to Random() depending on the vet level & health)

489BB0 - Unit_ApplyDamage - When a unit is hit by a weapon, it applys a health buff.
499CD0 - Unit_CalcWeaponDamage - When a unit fires a weapon (and hits something), damage amplification based on vet level?

Unit_CalcWeaponDamage
- Finds the per-unit type damage otherwise applies the default damage.

.text:00499DAE xor     ecx, ecx
.text:00499DB0 mov     eax, 66666667h
.text:00499DB5 mov     cx, [ebx+UnitStruct.Kills]
.text:00499DBC imul    ecx
.text:00499DBE sar     edx, 1
.text:00499DC0 mov     eax, edx
.text:00499DC2 shr     eax, 1Fh
.text:00499DC5 add     edx, eax
.text:00499DC7 cmp     edx, 5
.text:00499DCA jle     short loc_499DD1
.text:00499DCC mov     edx, 5
.text:00499DD1 
.text:00499DD1 loc_499DD1:   
.text:00499DD1 lea     ecx, [edx+edx*2]
.text:00499DD4 mov     eax, 51EB851Fh
.text:00499DD9 lea     ecx, [ecx+ecx+100]
.text:00499DDD imul    ecx, esi
.text:00499DE0 imul    ecx
.text:00499DE2 sar     edx, 5
.text:00499DE5 mov     eax, edx
.text:00499DE7 shr     eax, 31
.text:00499DEA add     edx, eax
.text:00499DEC mov     esi, edx

This appears to be the formula for calcing a unit's weapons's damage when it hits something. Equation looks like(Where rounding occurs is important!);
Damage(M) = Damage(I)*((5*RoundDown(Kills/5,0))+100) /100

Kills are capped at 25/Veteran level at 5

.text:00489BF3 xor     ecx, ecx
.text:00489BF5 mov     eax, 66666667h
.text:00489BFA mov     cx, [esi+UnitStruct.Kills]
.text:00489C01 imul    ecx
.text:00489C03 sar     edx, 1
.text:00489C05 mov     ecx, edx
.text:00489C07 shr     ecx, 1Fh
.text:00489C0A add     edx, ecx
.text:00489C0C cmp     edx, 5
.text:00489C0F jle     short loc_489C16
.text:00489C11 mov     edx, 5
.text:00489C16 
.text:00489C16 loc_489C16:  
.text:00489C16 mov     ecx, 25
.text:00489C1B mov     eax, 51EB851Fh
.text:00489C20 sub     ecx, edx
.text:00489C22 imul    ecx, edi
.text:00489C25 shl     ecx, 2
.text:00489C28 imul    ecx
.text:00489C2A sar     edx, 5
.text:00489C2D mov     eax, edx
.text:00489C2F shr     eax, 1Fh
.text:00489C32 add     edx, eax
.text:00489C34 jmp     short loc_489C3A

This appears to be the formula for calcing a how much actual damage a unit takes from a weapon. Equation looks like(Where rounding occurs is important!);
Damage(Reduced) = ROUNDDOWN(Damage(Initial)(25-ROUNDDOWN(Kills/5,0))/25)

Kills are capped at 25/Veteran level at 5

This explains how a veteran Krogoth can survive a D-gun hit. No damage modifier will be applied if the damage is over 30k, but the unit's veteran status still gets applied. So a Krogoth with 25 kills takes 24k damage from a dgun instead of 30k.

[DannyU]

This is good, it means that veterancy doesn't affect health but confers damage resistance instead, therefore Reload vs Current Health (as observed by zwzsg) can operate independantly.

[DannyU]

Ok, so I haven't done Veterancy vs Capture Time yet but I have done this preliminary stuff.

CaptureTime is grounded at 5s minimum and capped at 60s maximum.
It is dependant on BuildCostMetal and BuildCostEnergy.

BCE=1

	BCM	CaptureTime(s)

	1	5
	10	5
	100	6
	500	9
	1k	12
	2.5k	23
	5k	41
	7.5k	58
	10k	60
	100k	60
	1m	60


BCM=1

	BCE	CaptureTime(s)

	1	5
	10	5
	100	5
	1k	6
	5k	7
	10k	10
	25k	17
	50k	30
	75k	42
	100k	55
	1m	60
	10m	60
	100m	60

Commentary: Capture Time is not affected by MaxDamage (initial health, though it may still be affected by current health - not tested) nor by BuildTime. The ratio at which BCM and BCE have a comparable effect is ~1:15 or ~ 1:16 but it's hard to ascertain clearly. The grounding and capping of Capture Time is going to make the Veterancy vs Capture Time table less accurate than desirable - but I'll see what I can do .

[BlueWolf]

Originally Posted by zwzsg

...has its health overloaded to 500%...

Sort of off-topic (maybe), but how is this actually accomplished?

[Xon]

Originally Posted by BlueWolf

Sort of off-topic (maybe), but how is this actually accomplished?

Negative damage modifier. Enabled/Disabled by "Set Armored = true/false"

[BlueWolf]

Originally Posted by Xon

Negative damage modifier. Enabled/Disabled by "Set Armored = true/false"

Ahh. That makes (some) sense (to me).

[TheRegisteredOne]

or by negative damaging weapons

[macraig]

I'm gonna make me a "Cleric" unit with a negative-damage area weapon who can "bless" any friendlies in his vicinity. Funny that they didn't abuse that in TA:K where it actually would've fit in, eh?

Say, it's been a while for me and this particular aspect of scripting, so I ask the question: is it possible to script a ceiling for unit health, such that if it's increased beyond 100% - say by a negative-damage weapon striking it - it will be reset to 100%? That trick could actually be used to create an alternative type of shield, were it not for the fact that it would benefit nearby enemy units as much as friendlies.

[TheRegisteredOne]

it is really bad
1. los units lose their ability to fire after a while, firing/aiming script gets stuck due to high rate of fire
2. ballistic weapons lose all their accuracy after 100%, negative accuracy value
3. damage modifier rounds down, so -0.1 = -1, so you cannot negate nega damage without making the unit invincible
4. the units explode after overdose (damage must have wrapped around)
5. it kills trees

[zwzsg]

3. damage modifier rounds down, so -0.1 = -1, so you cannot negate nega damage without making the unit invincible

Tro had an idea: Make the healing weapon have a damage of -1, and have the blessed unit have a damagemodifier such as 0.5 and set ARMORED to TRUE; when get HEALTH > 120. The idea was that -1*0.5=-0.5 would be rounded to 0, and so the blessed unit could become immune to blessing, while still getting hurt from all other weapon (which all have a damage of at least 2). But because the rounding when the other way, it didn't work.

[Huh?]

Just a quick question, might have been said or asked, but what about units in the thousands of kills? Like nukes and LRPCs?

[Xon]

The damage modifier gets rounded to the nearest whole number when read from the fbi file at load time, it isnt posible to have it as a fraction

[DannyU]

Veterancy vs Capture Time (1)

    Veterancy  Capture Time(s)

	0	40
	1	44
	2	48
	3	52
	4	56
	5	60



Veterancy vs Capture Time (2)

    Veterancy  Capture Time(s)

	0	20
	1	-
	2	-
	3	-
	4	-
	5	30
	6	-
	7	-
	8	-
	9	-
	10	40
	11	-
	12	-
	13	-
	14	-
	15	50
	16	-
	17	-
	18	-
	19	-
	20	60
	21	-
	22	-
	23	-
	24	-
	25	70

Commentary: Despite my fears of inaccurate data this is the clearest set of data yet. The veterancy modifier is 10 (%).

Initial Capture Time capped at 60s, Veterancy bonus capped at 6s (tested but not shown in this table).
Veterancy Capture Time is uncapped (tested up to 210s - there could still be a higher cap).

[zwzsg]

Originally Posted by Xon

The damage modifier gets rounded to the nearest whole number when read from the fbi file at load time, it isnt posible to have it as a fraction

I disagree with that. Super units, such as the Fat Shrew, use a fractional damage modifier, such as damagemodifier=0.4; to overcome maxdamage being limited to 32767. All the units with health noticeably bigger than the Krog use that!

[BlueWolf]

Originally Posted by zwzsg

I disagree with that. Super units, such as the Fat Shrew, use a fractional damage modifier, such as damagemodifier=0.4; to overcome maxdamage being limited to 32767. All the units with health noticeably bigger than the Krog use that!

Sorry to butt-in, but TAWP possesses a ton of units with a fractional damage modifier, including the Krogoth.

[TheRegisteredOne]

*looks at BlueWolf* O.o
uh-huh... ota solars, radars, pop-up cannons, and other units that close up uses fractional damage modifiers.

[zwzsg]

Well, save that they're broken, because the Cavedoggies forgot the set ARMORED to TRUE; in half of them.

[BlueWolf]

Originally Posted by TheRegisteredOne

*looks at BlueWolf* O.o

Ah! TRO's looking at me!

[macraig]

I'm trying not to look at you. My eyes have suffered a few too many traumas here lately!