Object and Monster generation question

[Zikke]

How does the game determine whether or not to spawn an out of depth monster (excluding vaults)? How many levels OOD is possible?

Regarding items, I heard at a certain dlvl, any GOOD or better items are increased by 10 levels when it comes to rolling for drops. When does this start?


I am trying to make a spreadsheet for myself to document when certain key resists, monsters, or items appear at which dlvl so that I can know when to "not dive further until" or "dive to this level to find ___"


Thanks!

edit: Also, why do all armor items in the ego_items.txt edit file have a zero depth value?


Example: Boots of Speed

N:59:of Speed
X:25:0
W:0:24:0:100000
C:0:0:0:d10
T:30:0:99
F:SPEED | HIDE_TYPE

[Sirridan]

How does the game determine whether or not to spawn an out of depth monster (excluding vaults)? How many levels OOD is possible?

Regarding items, I heard at a certain dlvl, any GOOD or better items are increased by 10 levels when it comes to rolling for drops. When does this start?


I am trying to make a spreadsheet for myself to document when certain key resists, monsters, or items appear at which dlvl so that I can know when to "not dive further until" or "dive to this level to find ___"


Thanks!

edit: Also, why do all armor items in the ego_items.txt edit file have a zero depth value?


Example: Boots of Speed

N:59:of Speed
X:25:0
W:0:24:0:100000
C:0:0:0:d10
T:30:0:99
F:SPEED | HIDE_TYPE

1. happens in monster2.c

406	s16b get_mon_num(int level)
407	{
408	        int i, j, p;
409	
410	        int r_idx;
411	
412	        long value, total;
413	
414	        monster_race *r_ptr;
415	
416	        alloc_entry *table = alloc_race_table;
417	
418	        /* Occasionally produce a nastier monster in the dungeon */
419	        if (level > 0 && one_in_(NASTY_MON))
420	                level += MIN(level / 4 + 2, MON_OOD_MAX);

and in defines.h...

259	#define NASTY_MON    25        /* 1/chance of inflated monster level */
260	#define MON_OOD_MAX  10        /* maximum out-of-depth amount */

Which means there is a 1/25 chance of getting an OOD monster, and the increased level of that OOD is (level/4 + 2) or 10, whichever is smaller.

Example, at level 15 the highest out of depth possible is 6.

2. Not sure, I think at 30, when only magic stuff drops, but I'm too lazy to look through the code some more.

3. Elvenkind have a depth of 60 I think, but the rest can drop any time but may be quite rare. I'm sure it's just so people can have nice things early on too that aren't artifacts, when later on in the game artifacts often surpass most egos. (most, not all, mind you)

[Zikke]

So in theory, a pair of boots on dlvl 1 has the same chance of being "of Speed" as a pair of boots on dlvl 100?


And thank you for the code on generating OOD monsters.

[Sirridan]

So in theory, a pair of boots on dlvl 1 has the same chance of being "of Speed" as a pair of boots on dlvl 100?


And thank you for the code on generating OOD monsters.

No prob, but no, deeper you get the more likely you'll get ego items, and therefore speed boots.

But things in town that drop items (Farmer Maggot, Mean Looking Mercs, Veterans) can drop boots of speed. There was a screenshot or a post on that not too long ago IIRC.

[buzzkill]

Which means there is a 1/25 chance of getting an OOD monster, and the increased level of that OOD is (level/4 + 2) or 10, whichever is smaller.

Example, at level 15 the highest out of depth possible is 6.

This seems weak. A 1 in 25 chance of getting a monster that's sightly OoD (at dl32 or higher only 10 levels OoD). Does anyone else think that there is not enough variation here? Good old predictable Vanilla, I guess.

[Derakon]

I'd be all for more variation in monster and item placement. Back in the way-old days generation depths were a lot more variable, and you could find nicer gear much more often (for example, PDSM was more of a "once every few winners" deal rather than a "three times in one player's life" deal). Polymorph could generate any non-unique monster (so polymorphing townspeople was a fun way to check out powerful enemies in a somewhat-safe manner, by standing on the staircase). And so on.

[d_m]

This seems weak. A 1 in 25 chance of getting a monster that's sightly OoD (at dl32 or higher only 10 levels OoD). Does anyone else think that there is not enough variation here? Good old predictable Vanilla, I guess.

Maybe. It's actually a lot more likely to generate tougher monsters than the old code which it replaced (which was a lot more complicated and a lot less likely to generate a monster 10 levels out-of-depth). There was a discussion about the change on the mailing list and on IRC if you're interested.

There has been talk about moving to a function that could produce *any* depth of monster or item (obviously with a correspondingly small probability) but tuning the function so it "felt" right would be tricky and no one's done it yet.

Patches welcome

[buzzkill]

There was a discussion about the change on the mailing list and on IRC if you're interested.

There has been talk about moving to a function that could produce *any* depth of monster or item (obviously with a correspondingly small probability) but tuning the function so it "felt" right would be tricky and no one's done it yet.

Mailing list? IRC, what's that? I didn't know of such things.

The OoD object generation seems to have a *lot* more variability and seems about right. It would be nice if the monsters were also. I guess I'll just sit around and wait.

EDIT: A quick fix, if anybody notices this, would be to change

259	#define NASTY_MON    25        /* 1/chance of inflated monster level */
to this
259	#define NASTY_MON     [COLOR="Yellow"]5 [/COLOR]       /* 1/chance of inflated monster level */

[Derakon]

It seems like a simple bell curve would work fine. You could start with a curve where monsters are generated more than the current max OOD-ness less than .1% of the time (basically similar to current behavior), and gradually increase that until it feels right.

[d_m]

Mailing list? IRC, what's that? I didn't know of such things.

The mailing list isn't very active right now but sometimes there is discussion on it. I found the email detailing the changes to OOD monsters:

http://angband.oook.cz/pipermail/angband-dev/2009-September/000881.html

The IRC channel is #angband-dev on freenode. It too is pretty inactive right now.

[PowerDiver]

It seems like a simple bell curve would work fine. You could start with a curve where monsters are generated more than the current max OOD-ness less than .1% of the time (basically similar to current behavior), and gradually increase that until it feels right.

What feels wrong now? Do you really want people dying to ethereal dragons at DL 30?

[Pete Mack]

What feels wrong now? Do you really want people dying to ethereal dragons at DL 30?

Well, I would have said no, but after seeing someone gain 33 levels in a single turn, I'm thinking death molds on dl 1 would be nice.

[Derakon]

What feels wrong now? Do you really want people dying to ethereal dragons at DL 30?

I wouldn't want this to be common, but yes, I think it should be possible. You'd have to be pretty phenominally unlucky to get an ethereal dragon (or other invisible passwall badass) at DL30, though: not only would you need to get a large OOD roll, but then you'd also need to pick that particular monster instead of one of the other much more common monsters from the new depth.

In my opinion it makes the game more interesting to have the occasional totally unfair challenge and the occasional totally "undeserved" item drop. Maybe I'm just viewing things through the lens of nostalgia, but early versions of Angband definitely did have a wider spread on item drops and monster placements, and were quite fun. And it seems like there's a certain faction of players here who want to make the game "solvable" (which is to say, all deaths should be clearly the player's fault), which to my mind makes things less exciting. Without the risk of failure, the joy of victory is lessened, and one of the ways of injecting that risk into the game is to make it possible for there to be anything around the next corner.

[PowerDiver]

I'm not personally opposed to instakills, but there has been a movement in that direction. There is even a ticket something like "avoid unavoidable deaths".

Why not add a 1/10,000 each time you take the stairs that you fall and die? That's one way to inhibit stairscumming. I bet people would complain.

[Derakon]

The difference between random low-odds "you tripped on the stairs" and random low-odds "there's a huge badass monster on this level" is that the latter probably isn't an unavoidable death. Witness the fact that some players routinely take woefully under-equipped and under-leveled characters deep into the dungeons just because they can, and often make it quite a bit deeper than you'd think possible. Most monsters aren't going to kill you before you get a chance to do something about it, even if they massively out-power you. In other words, the "It breathes poison. -more- You die." really isn't that common an instadeath.

So in other words, I support a 1-in-10,000 chance of generating really powerful monsters, because given the way the game works, oh, 98 out of 100 times that such a monster is generated, the player will still be able to escape after seeing it. That makes for a .0002% instadeath rate, which I call acceptable.