Euclidean movement, LOS, distances
Harald Korneliussen
of support for switching Crawl to square LOS. The argument was that
because ranged attacks use euclidean distance, yet movement costs the
same orthogonally and diagonally, you get some tactical oddities: If
you should sneak up on something or close in to melee range, you
should move diagonally, if you want to keep distance for using ranged
attacks, you should meet enemies head-on.
I would have thought the fix (if needed) was to make diagonal moves
cost more. This is not at all unheard of for tactical games on a
square grid (e.g. X-COM). But it seemed almost half the developers
would rather go all square, dismissing objections as aesthetic
sentimentality, it's an abstract game anyway don't you know? :-o
Fortunately (for me, at least), the idea and the code branch appears
to have died.
What I'd like to know is how many of you have experimented with
increased cost for diagonal movement. Also, if anyone of you (outside
of the Crawl fanbase) prefer square LOS in your roguelikes.
Michal Bielinski
> I saw to my horror on the Crawl development wiki that there was a lot
> of support for switching Crawl to square LOS. The argument was that
> because ranged attacks use euclidean distance, yet movement costs the
> same orthogonally and diagonally, you get some tactical oddities: If
> you should sneak up on something or close in to melee range, you
> should move diagonally, if you want to keep distance for using ranged
> attacks, you should meet enemies head-on.
These oddities can easily be lived with. You could think it is some
Crawl flavor.
> I would have thought the fix (if needed) was to make diagonal moves
> cost more. This is not at all unheard of for tactical games on a
> square grid (e.g. X-COM).
But is unheard in roguelike games since it severely harms the spirit.
If ranged attacks are to use euclidean distance moving diagonally may
not only move you into deep elf annihilator spell range but also give
him two turns to smash you with Lehudib's. In my eyes more expensive
diagonal movement is inherently evil because it allows for too many
double attacks which create instadeath chance.
> But it seemed almost half the developers
> would rather go all square, dismissing objections as aesthetic
> sentimentality, it's an abstract game anyway don't you know? :-o
Kheh-heh. Did they consider changing distance formula instead? That
would be another solution.
> What I'd like to know is how many of you have experimented with
> increased cost for diagonal movement. Also, if anyone of you (outside
> of the Crawl fanbase) prefer square LOS in your roguelikes.
I have uttered my hatred for costly diagonal steps already. My top list
of ways to deal with this problem:
1. Go square FOV and square range. Manhattan metric all the way.
No artifacts whatsoever.
2. Go circle FOV and square range. Has its oddities but just live with
them.
3. Go circle FOV and circular range. With low weapon ranges this is
slightly irritating for player. People will tend to check every
step if monster is within their range.
100. Go circle FOV, square range and diagonal cost 1.5. Beware of
feared double-jeopardy.
101. Go circle FOV, suqare range and diagonal cost 1.4. This is worse
than 100 because player has more difficulty judging when double
moves happen.
--
Michal Bielinski
Harald Korneliussen
> These oddities can easily be lived with. You could think it is some
> Crawl flavor.
Well yeah, the system today is liveable. I have learned - intuitively
- to sneak up on creatures from the diagonal, and step back out of
sight and come back face-on when encountering a sleeping monster with
a ranged user. Square LOS, though, would IMO be a big step backwards,
both in tactical richness and "realism".
> > I would have thought the fix (if needed) was to make diagonal moves
> > cost more. This is not at all unheard of for tactical games on a
> > square grid (e.g. X-COM).
>
> But is unheard in roguelike games since it severely harms the spirit.
Is it really unheard of? Not a single 7DR experimenting with it, even?
That's what I want to know.
> diagonal movement is inherently evil because it allows for too many
> double attacks which create instadeath chance.
Crawl has a pretty fine-grained timing system already (though I don't
know how much compared to other modern roguelikes). Executioners and
Blademasters get far more than double attacks already :D It would
mainly become an issue when you try to "cut corners" (quite
literally), and even then I'd think it wouldn't be all that far out of
line from the punishment Crawl throws at you.
Martin Read
>> diagonal movement is inherently evil because it allows for too many
>> double attacks which create instadeath chance.
>
>Crawl has a pretty fine-grained timing system already (though I don't
>know how much compared to other modern roguelikes).
Angband, ADOM, and DoomRL all have fairly fine-grained timing systems.
Nethack has not so much a timing system as an enormous bag of hacks
masquerading as one.
>Executioners and Blademasters get far more than double attacks already
Yes, but they're supposed to (they have multiple attacks and high
speed). Most monsters aren't.
>:D It would mainly become an issue when you try to "cut corners" (quite
>literally), and even then I'd think it wouldn't be all that far out of
>line from the punishment Crawl throws at you.
.XA##
.B@C.
#####
Monsters at A, B, C. (Sure, you've probably screwed up already, but
never mind.) Assuming they all move at normal speed, in the Euclidean-
costs world you get attacked anywhere up to five times when you try to
step through to X. In the equal-costs world, you get attacked twice.
I think there _is_ a way to make Euclidean costs acceptable: allow each
unit to do more in a single turn than simply "do a thing" or "move a
step".
--
\_\/_/ turbulence is certainty turbulence is friction between you and me
\ / every time we try to impose order we create chaos
\/ -- Killing Joke, "Mathematics of Chaos"
Kenneth 'Bessarion' Boyd
> On Mar 22, 1:06 pm, Michal Bielinski <dungeon_kee...@tlen.pl> wrote:
>
> > These oddities can easily be lived with. You could think it is some
> > Crawl flavor.
>
> Well yeah, the system today is liveable. I have learned - intuitively
> - to sneak up on creatures from the diagonal, and step back out of
> sight and come back face-on when encountering a sleeping monster with
> a ranged user. Square LOS, though, would IMO be a big step backwards,
> both in tactical richness and "realism".
>
> > > I would have thought the fix (if needed) was to make diagonal moves
> > > cost more. This is not at all unheard of for tactical games on a
> > > square grid (e.g. X-COM).
>
> > But is unheard in roguelike games since it severely harms the spirit.
>
> Is it really unheard of? Not a single 7DR experimenting with it, even?
> That's what I want to know.
Zaiband has it. The results are as abhorrent as Michal says, and
that's with exposing the energy system to make calculating double-
moves possible.
I need to modify Zaiband to provide more assistance to the player
regarding timing issues, but am pretty sure that won't make 3.0.10
which has missed release by two weeks already.
Konstantin Stupnik
>
> Is it really unheard of? Not a single 7DR experimenting with it, even?
> That's what I want to know.
But spells and ranged attacks are targeted at cells,
not units, and everyone first spend time on action
and only after time is spent get result.
So when you move diagonally in range of monster's attack,
you still have chance to avoid it, if monster started
action while targeting your previous position.
Konstantin Stupnik
>
> Is it really unheard of? Not a single 7DR experimenting with it, even?
> That's what I want to know.
Björn Ritzl
> On Mar 22, 1:06 pm, Michal Bielinski<dungeon_kee...@tlen.pl> wrote:
>
>> These oddities can easily be lived with. You could think it is some
>> Crawl flavor.
>
> Well yeah, the system today is liveable. I have learned - intuitively
> - to sneak up on creatures from the diagonal, and step back out of
> sight and come back face-on when encountering a sleeping monster with
> a ranged user. Square LOS, though, would IMO be a big step backwards,
> both in tactical richness and "realism".
>
>>> I would have thought the fix (if needed) was to make diagonal moves
>>> cost more. This is not at all unheard of for tactical games on a
>>> square grid (e.g. X-COM).
>>
>> But is unheard in roguelike games since it severely harms the spirit.
>
> Is it really unheard of? Not a single 7DR experimenting with it, even?
> That's what I want to know.
I'm pretty sure quite a few people have experimented with it, but I'm
not sure it stuck with anyone since it adds complexity and makes it
harder for the player to judge the result of a move. There has been a
lot of talk about diagonal movement in r.g.r.dev in the past:
I used different diagonal cost for my old roguelike Dungeonweller
(http://www.roguelikedevelopment.org/dungeondweller/) but decided to
skip it for Dweller
(http://roguebasin.roguelikedevelopment.org/index.php?title=Dweller).
Dweller has a fine grained timing system but I still do not think it's
worth adding a higher cost for diagonal movement. I prefer suggestion 1
or 2 in Micheal's post elsewhere in this thread.
/Björn
eoghan griffin
4ba644...@k15g2000prk.googlegroups.com>, vinte...@gmail.com says...
Ugly as squarelos may have looked, it played great while it was on cdo.
It had lots support from most of the ##crawl irc channel. We were full
sure it was gonna be made the default, then the guy doing the dev on it
vanished.
You did find yourself doing tricks you used to do to exploit los though,
but i think thats just a sign of how broken circular los was.
The option of making diagonals cost more would really mess with play
imo.
Jeff Lait
>
> I have uttered my hatred for costly diagonal steps already. My top list
> of ways to deal with this problem:
>
> 1. Go square FOV and square range. Manhattan metric all the way.
> No artifacts whatsoever.
>
> 2. Go circle FOV and square range. Has its oddities but just live with
> them.
>
> 3. Go circle FOV and circular range. With low weapon ranges this is
> slightly irritating for player. People will tend to check every
> step if monster is within their range.
>
> 100. Go circle FOV, square range and diagonal cost 1.5. Beware of
> feared double-jeopardy.
>
> 101. Go circle FOV, suqare range and diagonal cost 1.4. This is worse
> than 100 because player has more difficulty judging when double
> moves happen.
Additional ways:
4: Don't have diagonal movement. Only allow 4-way movement. As a
bonus, arrow keys suffice for movement so you avoid all the laptop/vi-
key shennigans. Disadvantage is that if you only allow four way
attacks you end up with a melee-multiplier of 4x rather than 8x.
POWDER does this.
6: Switch to Hexes. Then you can use manhattan for both your FOV and
your range happily since you get aesthetically pleasing hexagons for
both. Down side is that controls are a bit weirder - I recommend
people clone LaserSpigot's controls as the new standard. You also end
up with a 6x melee-multiplier, which is actually a nice mid-ground.
Both of those require a big rebalance, however, so are likely
infeasible for developed games.
--
Jeff Lait
(POWDER: http://www.zincland.com/powder)
Harald Korneliussen
wrote:> http://groups.google.com/group/rec.games.roguelike.development/search...
>
Thanks for that. I did try to search for it, but apparently didn't use
the right terms.