targetting and LOS

[PowerDiver]

The suggested changes would make the following situation *not* LOS.

######################M
@


I don't know whether that is a good thing or a bad thing. I can't find it in this thread, but someone posted somewhere about going to the effort to code to make that targetable, so it's certainly not unanimously considered good.

Do we want to change to make something currently visible change to non-visible?

[jv123]

There's clearly no way of implementing symmetric LOS that gives realistic cone-shaped shadows behind pillars, and which also makes monsters visible in the above situation. Cone-shaped shadows behind pillars are essential, because the alternatives are line-shaped shadows, which are unrealistic and which work poorly when you're not in a horizontal or vertical line from the pillar, or no shadows, which would be ridiculous.

So, I would argue that the monster should indeed not be visible in the situation you've just described.

[PaulBlay]

[QUOTE=PowerDiver;20792]The suggested changes would make the following situation *not* LOS.

######################M
@

Frankly I think targetable / LOS at that distance is just greedy. ;-)

But actually I think it would be LOS under my suggestion (or at least it would be if I had worded my suggestion clearly).

As you can see 1/4 of any wall cell is clearly visible at any distance.

[Marble Dice]

As long as it's symmetric, I think we'd be fine.

Thinking out loud here, it makes a certain amount of sense that if you are inside of a rectangular room, you can see anyone standing at any door into that room. Assume LOS is symmetrical, so that would mean anyone standing at a door to a room could also see anyone standing inside the room. This second point makes a little bit less sense, especially for very large rooms - if I'm standing at the thresh hold I probably couldn't see someone standing in one of the near corners, unless I was actually inside the room myself.

So I think I'd be fine with no LOS in Eddie's example. If the person inside the room is close to the middle, or the person at the door is actually one square inside the room, then you'd have LOS and that makes sense to me. If you've got no LOS then the parties involved are basically hiding in corners and doorways.

Also for very long corridors you wouldn't be able to see what's coming around the corner until it had stepped around the corner, or you were pretty close. I'm fine with that too as long as it's symmetrical.

[PaulBlay]

This is what my suggestion would look like when entering a large room (green is visible).

I'm pretty sure it's symmetrical. (Obviously we can do the actual maths on this instead of just moving lines on Excel)

[jv123]

Paul --- I love the picture!

But actually I think it would be LOS under my suggestion (or at least it would be if I had worded my suggestion clearly).

Unless I'm mistaken, the problem doing it the way you suggest is that it would lead to weird shadows behind pillars, like this, where the gaps are the only invisible grid squares:

...........
.@.........
...#.......
..... .....
....... ...
......... .
...........

It wouldn't be crazy to have this behaviour, but I think cone-shaped shadows would be more realistic.

MarbleDice said:

if I'm standing at the thresh hold I probably couldn't see someone standing in one of the near corners ... I think I'd be fine with no LOS in Eddie's example. ... Also for very long corridors you wouldn't be able to see what's coming around the corner until it had stepped around the corner, or you were pretty close. I'm fine with that too as long as it's symmetrical.

I agree. Using the centre-to-centre rule, and the rule that walls only conceal the middle quarter of their grid square, this would be the limits of visibility around a corner:

    #.#    
#####@#
..M...#
#######

These two would NOT be able to see each other, because the centre-to-centre line passes exactly through the corner of the wall, and so would be classed as blocked.

[jv123]

Also, Paul, I don't see how your system can be symmetrical --- compare your pictures in posts #33 and #35; they clearly don't describe symmetrical behaviour.

Could you explain exactly what your suggestion is? You wrote

1. Pillars take up half the square.
2. Pillars are square.
3. Any square you can see at least half of you can "see" less than that and you can't see them.

but I don't get exactly what you mean. In particular, I'm confused by your point 1 --- we had been discussing the suggestion that pillars take up a quarter of the area of a square, but placed at the centre, i.e. the central 2x2 squares of a 4x4 subdivision.

[Atanvarno]

This system allows the player to see very much, even more than the digital field of view (http://roguebasin.roguelikedevelopme..._field_of_view):

  %%%%#%%
  %  .   
 %% ###%%
 % ...#
##....#
@.....#%%%%
##........#%
 % ...###...
 %  ..# %%%%
 %%%%##

Here % is a hidden wall.

Digital fov is symmetric and has the following nice property:
if A can see B, then A can see everything between A and B.

This new system has the following slighty weaker property:
if A can target B (with the old method), then A can see everything between A and B.

In the traditional system this is not true:

###X.B
A.....

Here A can't see X, but a shot fired at B hits X.

Roguebasin contains articles on different fov systems, and a comparison of them: http://roguebasin.roguelikedevelopme...ield_of_Vision.

[RogerN]

If I remember Angband's view code correctly, the existing view algorithm could be adapted to use the proposed center-to-center method without too much difficulty. I think Paul's suggestion would be considerably more processor-intensive to implement. I don't see any way to get around having to track the visibility state of 16 subgrids per grid.

Not that implementation concerns should always dictate decisions which affect gameplay, but I think performance is pretty important in this case. However Angband calculates the player view, the algorithm needs to be pretty fast.

[PaulBlay]

Also, Paul, I don't see how your system can be symmetrical --- compare your pictures in posts #33 and #35; they clearly don't describe symmetrical behaviour.

Hmm, my gut instinct was that it would be symmetrical but it doesn't appear to be the case. :-( I think I may have to drop my suggestion if so.

but I don't get exactly what you mean. In particular, I'm confused by your point 1 --- we had been discussing the suggestion that pillars take up a quarter of the area of a square, but placed at the centre, i.e. the central 2x2 squares of a 4x4 subdivision.

First point - I changed from needing to see all the square to needing to see half the square. This allows the 'ghost in wall' situation to be OK.

Second point that's exactly what I'm doing - pillars and walls have a 1/4 (25%) gap around them where they meet the open space (as seen in the pictures).

[EDIT] Changed text in first half.

[PaulBlay]

I think Paul's suggestion would be considerably more processor-intensive to implement. I don't see any way to get around having to track the visibility state of 16 subgrids per grid.

It's maths. But that doesn't mean that you can't fake it with tables and stuff.

[will_asher]

Take me with a grain of salt because I'm pretty sure I'm in the minority..

Why is it so important that line of sight be symmetrical?

In the following example (copied from another post), I would consider the @ to be both hiding behind the wall and peeking around it. So the @ should be able to see most of the hallway, but it should be pretty unlikely that the M would notice the @. This would involve having stealth as a factor in line of sight but I think it's much more realistic that way.

    #.#    
#####@#
..M...#
#######

The problem with assuming that the player and monsters always take the part of the grid in the middle is that any player or monster with sense is going to try to keep as much of their body as possible out of the line of fire (which should be the same as LOS).

[PowerDiver]

Hmm, my gut instinct was that it would be symmetrical but it doesn't appear to be the case. :-(

If the definition is asymmetrical, a point in one square and an area in the other, the result is unlikely to be symmetrical. If you want something along these lines, I think you'd have to define it as some sort of measure of the proportion of points in the two squares that can see each other. The obvious starting point for such a def would be bounds on some sort of integral, but maybe its properties would be such that there would be a simpler equivalent def once you worked out exactly what you wanted.

[Marble Dice]

Why is it so important that line of sight be symmetrical?

One good reason is that in a symmetrical visibility system, it is easy and intuitive to determine which enemies have LOS on you. If there are many asymmetrical lines of sight, then it becomes difficult to avoid situations where enemies have asymmetrical LOS on you, and it also encourages the exploitation of AI which doesn't strive to avoid being the victim of asymmetrical LOS (the classic hockey stick).

Additionally I would argue that certain asymmetrical lines of sight would make the game significantly more tactically frustrating in a bad way - consider breaking into a room with breathers in the corners with a system where the room inhabitants can see you at the door, but you can't see them. Effectively, they'd get two chances to breath on you before you could even see them, once at the door and once as you stepped into the room. The first aggressor in Angband is already at a disadvantage due to the turn-based nature of the game, and asymmetrical lines of sight only exacerbate the problem.

[Magnate]

However, it is exceptionally hard to get these things right and still playable, and the D&D combination of evasion and damage reduction into a single value called AC makes it impossible. The only way to fix the rule is to throw it away. That is a lesson to be considered when thinking about things like "cover" bonuses in an angband framework.

If there was one radical change I could persuade Takkaria to make to V, it would be the separation of damage absorption (armour) and damage avoidance (evasion). So much else Just Works once you sort that out.