[MUD-Dev] Condsiders

rayzam rayzam at home.com
Sat Feb 24 11:41:14 CET 2001


JB wrote:

> I would tend to rely on a third person view, but where the objects
> in the world are presented according to the character's perceptions.
> Where the character is looking, the player would be presented with a
> full color, detailed view of the area, with the actual viewing
> frustrum clearly demarked.  As we transition from the foveal area of
> the character's field of view into the peripheral areas, the objects
> start to lower in polygon count and the colors might wash out.  As
> we go beyond even the peripheral view, objects are only displayed
> when they make a sound and are presented as grey lumps.  Noises that
> your character makes can mask noises around it.  So if you are
> walking down a country road singing a song, it means that a bandit
> might be able to sneak up behind you and club you.


Though this mimics the human visual system, it isn't a complete visual
solution which may cause problems. You're correct that the foveal area
is the highest acuity area and has the largest concentration of cones
[for color vision]. As you move towards the periphery in the retina,
the cone count goes down, the rod proportion goes up, and the acuity
drops [on the flip side, detection goes up due to the numbers of rods
per RGC. This is why, when you look up at the night sky, you can see
faint stars in your periphery, but not if you look at them
directly. Actually, this is a simplification, there is a bypass
circuit in the retina to handle just that sort of system via an
amacrine II cell, but that is probably unimportant to this model :)].

The point being that reducing the polygon count and lowering the
colors can mimic the human visual system, except for the fact, that
the eyes aren't static and it doesn't include the
persistance/memory/object coding. In general, your eyes are moving
around everywhere with rapid ballistic eye movements. This builds up a
representation of space around you. Now you pick up the colors/forms
best by foveating, but once you determine that's a yellow sticky on
the table, it remains a yellow sticky on the table even when you're
foveating elsewhere, and it's moved into the periphery where you
cannot see it as a nice sharp-edged yellow sticky.

So you've modelled the retina, but now the player has to rapidly move
the character's fovea around [something done naturally for the player,
not the character]. And lacking the persistance/object representations
means that the objects in view will fade out when they wouldn't for
the character.  On top of that, when we move our eyes around, we
account for that motion internally, via motor efference copies [one
theory] or the speed of the ballistic movements [up-and-coming
theory]. Now the player has to rapidly move the fovea around his 3d
world, loses his natural persistance/object models, and cannot
internally compensate for the motion [this is why some people are
susceptible to motion sickness when playing FPS].

As an alternate suggestion, why not limit the field of view, but keep
it all clear and colored. However, fade things in depth [generally
done anyway with lower poly counts]. And give hidden/harder to notice
things a percentage chance of not being rendered at all. But once it
is spotted for that character, always render it until he leaves the
area and some amount of time passes. This is more like the natural
visual system. If you're sitting at a conference table, it's a nice
long black table. If you get bored and start checking it out closely,
you may notice some imperfections in its surface. You didn't notice
these imperfections before, even though you saccaded around the room
and table. Once you notice them, they'll persist in the periphery of
your vision when you shift your eyes. If you then pay attention to
whomever is speaking or presenting at the meeting, when you move your
eyes back around the table, it may take you a little bit of searching
to find the imperfection again.

    rayzam
    www.retromud.org


_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev



More information about the mud-dev-archive mailing list