Modding-related theories/experiments/findings

[Dusk Voyager]

Let's post our wild theories or discoveries about why strange stuff happens and/or how the engine executes it. Of course posters can also validate or refute them if they like.

For example, this finding: http://forums.taleworlds.com/index.php/topic,173667.msg4175905.html#msg4175905

Now, I have a guess on why ranged attacks (reportedly) kill far more often.

 

[Mad Vader]

Now, I have a guess on why ranged attacks (reportedly) kill far more often.

Not likely, as there's no decision to hit like in D&D, only randomized inaccuracy modifier (most likely). Also there's some time between a shot being fired and a hit, when there's a randomized check for a kill or knock-out. It's a short time between a shot and a hit, but there's usually plenty of other action around that consumes random numbers too, so I don't believe it's the random generator at fault, just some ranged kill modifier in the engine that is heavily biased towards killing for some reason. Maybe it's even a bug that doesn't take Surgery into account.

Edit: why not try to reproduce SD's not-so-random sequences? The info might be outdated.

 

[cmpxchg8b]

As far as I can see, there is nothing about ranged that influences whether an agent gets killed or wounded.

 

[Dusk Voyager]

Not likely, as there's no decision to hit like in D&D, only randomized inaccuracy modifier (most likely). Also there's some time between a shot being fired and a hit, when there's a randomized check for a kill or knock-out. It's a short time between a shot and a hit, but there's usually plenty of other action around that consumes random numbers too, so I don't believe it's the random generator at fault, just some ranged kill modifier in the engine that is heavily biased towards killing for some reason. Maybe it's even a bug that doesn't take Surgery into account.

Fair enough. I've read someone imply  there's one before, but nobody else openly disagreed on it then (I can't find the link now). That could be wrong though. If there isn't, the shot direction is pre-determined no matter how small the reticle is ATM and as you say, I'm mistaken. If there is, it depends on more than just the accuracy stat on paper (proficiency, power draw etc), as it also has to apply to ranged weapons with 99. It would be still wrong, because each surgery point is supposed to give 1/25 (the best simplification) more chance to survive, but the results below suggest surgery itself would be pretty random.

Edit: why not try to reproduce SD's not-so-random sequences? The info might be outdated.

Apparently it's slightly better now:

Spoiler: digits from 0 to 9, what I got in 4 tries

1823648169
6146018370
2671434726
9363537130

The results above weren't that bad, so I tried a smaller range:

Spoiler: digits from 0 to 4, 4 tries

4433030411
3343311102
1004143302
0322421133

That was fairly repetitive, though.

 

[Mad Vader]

That makes sense, nice results. Only cmp can tell us why there is a kill bias with ranged hits on player's troops.
I'd bet on a Surgery bug or piecing damage kill modifier.

 

[cmpxchg8b]

That makes sense, nice results. Only cmp can tell us why there is a kill bias with ranged hits on player's troops.
I'd bet on a Surgery bug or piecing damage kill modifier.

There isn't one.
These are the condition for a troop to be wounded/killed (if one is met, the ones below will not be checked):
- if the mission template has the flag mtf_arena_fight, the agent will always be wounded
- if the dying agent's troop has the flag tf_allways_fall_dead, the agent will always be killed
- if the damage type is blunt, the agent will always be wounded
- if the dying agent's troop has the flag tf_hero, the agent will always be wounded
- if the dying agent is allied with the player, the agent has a 25% chance to be wounded (+4% per level of surgery)

 

[Mad Vader]

Thanks for that. It shall remain a myth, then.

 

[xenoargh]

if the dying agent is allied with the player, the agent has a 25% chance to be wounded (+4% per level of surgery)

Which is partially governed by the scripts handling casualties, at least for the main battles.  All of that can get over-ridden there.

 

[cmpxchg8b]

OP is talking about the "x knocked unconscious by y" message during a battle, and that can't be overridden anywhere.

 

[Dusk Voyager]

OP is talking about the "x knocked unconscious by y" message during a battle, and that can't be overridden anywhere.

Yes, it's nothing to do with attrition.

 

[xenoargh]

You can alter that, too; converting damage to be zero and converting blunt to lethal is possible, although it technically involves two "hits". 

The real outcome is determined by the script anyhow; you can convert the dead to wounded and vice-versa there, regardless of side.

 

[Dusk Voyager]

I know, but I was just wondering how it happens just when the agent gets killed or wounded with no hacks.

 

[xenoargh]

Even in Native, there's a script operating, in terms of who's converted to dead / wounded.

 

[cmpxchg8b]

No, he's still talking about being knocked unconscious during a battle. There is no way to operate on that apart from ti_on_agent_hit hacks, and Native doesn't use ti_on_agent_hit at all.

 

[Dusk Voyager]

@ xeno: So you're talking about an 'internal' non-Monty script, right? Sounds good.

 

[dunde]

OP is talking about the "x knocked unconscious by y" message during a battle, and that can't be overridden anywhere.

Yes, it's nothing to do with attrition.

Yes, it can be. I did by displaying my own message, followed by set_show_messages to 0 and mark a global variable on ti_on_agent_hit trigger. Another trigger check every frame and enable the message showing if it find the global variable marked.

 

[cmpxchg8b]

Ok, maybe that was phrased badly. We're talking about overriding the mechanic of getting wounded, not the message.

 

[dunde]

Well, I overided the message because I applied permadeath for companions and lords. It's not fun when in the middle of the battle we got a message about wounded companions but then to find out that they were death in the battle result report.
We can do it for regular troops by set a slot to the agents if they are death or wounded by our own rule. Displaying correct message, and then adjusting the battle casualities later to the parties based on the slot.
Adding party members if we find dead agents with wounded only slot, and vice versa. It's not pretty, but I think it will work.