Re: 8 bit values in mode 13h?

• Previous message: Randall Hyde: "Re: Solaris and HLA"
• In reply to: JGCasey: "Re: 8 bit values in mode 13h?"
• Next in thread: sinewave: "Re: 8 bit values in mode 13h?"
• Reply: sinewave: "Re: 8 bit values in mode 13h?"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Date: Wed, 22 Oct 2003 15:02:37 +0100

JGCasey wrote:
> Whew! What was my question again? :-)

You were enquirying about switching the DAC to 8 bits in mode 13h in
order to attempt to better accurately display 24-bit images in a 256
colour mode...

Sorry, there is no simple answer, as there are a myriad of factors
involved in such a task...I would not cheat you by pretending there is
a simple answer, just because that would look so much more clever to
have a perfect "soundbite" answer, than to ramble on about all these
various factors involved and the balances that need to be maintained
between them...

Now, if your question had been "Is your favourite colour purple?" then
I'd have said "yes" and maybe a minor embellishment of "purple's
really cool" and a smiley face or something, just so as not to seem
impolitely abrupt...but the thing people haven't twigged is that I
_listen_ - truly listen - to the questions put to me and do my level
best to give them a full and useful response...case in point answering
this question too...

Just because a question is short, unfortunately doesn't mean the
answer is equally short...or vica versa with long
questions...actually, more often than not, a long detailed question
only has a simple "yes" or "no" answer, while short questions - "What
is the meaning of life?", "What do you think about politics?", "What
is Love?", "How does that gadget actually manage to work?", etc. - are
entirely non-trivial answers that even the most concise person who's
ever lived could not answer in equal or less words than it took to ask
the question...

> In fact there is a clear and crisp line at the conceptual level.

Yes, indeed...as I said, there are no crisp and clear lines..."at the
conceptual level" means "how we choose to view it in our
minds"...well, that has no actual relation to reality, it must be
remembered...existing only "at the conceptual level" actually means
that it doesn't exist, save for a fevered imagination trying to ram
the square peg of reality into the round hole of their prejudices of
that reality...

Do we have to go into the issue of photons and how they act as
particles and waves simultaneously and then the way that the back of
the eye is excited by the arrival of photons, which effectively
summates all the photons that have arrived at the receptors, that
conspires to also mean that we don't see things in "frames" of time
either ("motion blur", for instance) as well as that the lowest-level
of matter that project these electromagnetic disturbances are
separated by vast gaps and, thus, do not touch at all?

Or can you already see that "at the conceptual level", actually means
"this is how my prejudices prefer to view it in a more abstract,
simplified manner for comprehension purposes" and that we should
_always_ be careful not to confuse our notation with what we are
actually notating...

Visually, there are NO crisp and clear lines...our vision is, in fact,
out of focus looking at a blurry world that moves around causing
motion blur everywhere...that's why you can still always tell the
difference between computer graphics and the real deal...and why all
the modern trends in computer rendering are actually to bring rough
surfaces, blurry images and all the "imperfections" of reality into
the picture because the greatest problem of realistic graphics is
actually that they are far, far too perfect, crisp, smooth, etc. and
that's utterly unnatural in this "rough around the edges" universe we
inhabit...

At the conceptual level, yes, you can choose to abstract out these
awkward ugly details...but note that abstraction does not physically
make them disappear...it's just the choice to decide that we will
ignore them and delibrately operate with less than the full facts
because that usually helps our poor little human brains comprehend
things...

> When anti-aliasing "blurs out" a line it really removes the sharp
> contrast of the edges of the pixel blocks so our visual system
> can blend them together and form the "real edge". In other words
> it prevents our high resolution edge detectors reacting to the
> edges of the blocky pixels.

And this is the eternal problem with dealing with things "at the
conceptual level"...you abstract "irrelevent details", the small
details add up to something now no longer "irrelevent", you try to
"plug the gaps" with yet more theorising and then everything is
confused...like, there are no sharp lines so we abstract that there
are "conceptually" but this poses a problem in explaining non-sharp
lines so we abstract that "blurry lines" are blurred blurred lines,
which makes a sharp line...

>From the "conceptual" viewpoint you're talking about, this isn't a bad
assessment but you're starting from a weird perspective so the
conclusion is even more weird...

For example, we have "high resolution edge detectors", do we? Well,
ignoring that biologists would simply argue that as total bollocks
from the off, as of what we know about human vision...there's still
the logical issue you've got to resolve in what you've said above that
blurring a line allows the visual system to make it "sharp"...is this
sort of like a taste that's so bitter it suddenly makes a "leap" into
becoming perfectly sweet? Or that if you put things out of focus more
and more, it suddenly becomes crystal clear and "sharp"?

You've actually twigged the answer but haven't seen it yet...how can
"blurring out" a sharp line make it sharp? Even on a "conceptual
level"? The reason that our visual systems can "blend them together"
is because they are constantly dealing with "blurred out"
lines...that's, in fact, all they do deal with...better yet, it's not
just "blurred out" with respect to space - as in this line example -
but with respect to _time_ also...a phenomenon known as "motion
blur"...it's simple enough to see this...stick you hand out in front
of you...right, now wave it up and down really fast like a mad
person...when it's in constant motion like this, you'll actually see
what looks like multiple hands sort of all blurred together, including
that some of the "phantom" hands are blurred with the background
behing them as if you had transparent hands..._certainly_ not a clear
or sharp perspective at all...

After all, if your mind is deceiving you with this "conceptual level"
business then your own eyes watching your own hand, where I could not
possibly be "rigging" the results you see might begin to convince
you...this "time blurring out" in "motion blur" is the most
obvious...it's harder to clearly demonstrate to an eye not trained to
see it that _everything_ you look at is "blurred out"...really, every
single thing...there is no such thing as a sharp, crisp line in a
visual sense but inside your mind...draw a straight line on paper and
it's not actually straight...nope, stick it under a microscope and
you'll see it's distorted by the grain of the paper...magnify this a
bit more and it'll seem even worse, like a city skyline full of
skyscrapers, the notion of the "line" long since forgotten and quite
silly in this "sharp" concept...throw away a light-based microscope
(we're going down even further where light ceases to be a useful way
to look at these things) and pull out your electron microscope...oh
dear, oh dear...better yet, check your atomic theory...there's
absolutely massive "gaps" between each of your atoms that easily dwarf
the size of the atoms themselves...ummm, aren't the points of a line
supposed to be all continuous and connected?

To quote Minority Report, "Can you see?"...is the value of this
"conceptual" line sinking in yet? Or do I need to give you yet another
angle by which to allow you to see the _LENS_ through which you're
looking which is effecting how you're perceiving what you see?

Indeed, let's talk about that lens specifically; The retina contains
two types of photoreceptors, called "rods" and "cones"..."rods", which
are more sensitive than "cones", aren't colour sensitive and merely
pick up the brightness of light entering the eye...the "cones", on the
other hand, are adapted for discerning colours...there are about 120
million "rods" but only 6 to 7 million "cones"...the "cones" are
mostly concentrated in the yellow central spot on the retina called
the "macula" (human vision is actually highly focussed in this spot
but quite crap elsewhere...we actually have a sort of "tunnel vision"
constantly and must move our eyes around to focus on different parts
of something and then mentally put that altogether to get an idea of
what we're looking at...look directly at this text but, without moving
your eyes, try to "look" at something to the side of you...you can't
see it properly, really, can you? I mean, it's there but the detail is
crap...there's certainly no chance of catching any "sharp lines" now,
is there? Especially as you're a man and men are generally worse both
_literally and metaphorically_ at seeing the wider picture...very
"ego-centred" and, thus, focussed and concentrated on that "macula"
area ;)...

These "rods" and "cones" aren't overlapping in any sense...they are
simply very small and densely packed, with the mind's interpretation
of what it receives from them automatically "blending" the information
into useful combined information...that's _one_ facet where "blending"
is inherently part of the visual system...you don't see "gaps" in the
colours on the screen - although, if there are "rods" in between these
"cones", wouldn't they keep them apart and cause "gaps"? You can
imagine some picture of this literal vision where you'd have colour
and greyscale pixels patterned like a chess-board...that's not what
you see...this is one level where "blending" occurs...

The second place is also there, if you know where to look...an object
has light bouncing off it and it enters my eye and hits the
retina...well, things aren't divided up into neat "grid" and that they
always "snap" to this grid...to stress, let's consider a sole photon
entering the eye...it hits the retina on a particular "cone"...a tiny,
tiny fraction of movement and, instead, it impacts on a
"rod"...changing from colour to brightness? That's pretty
radical...and what mysterious force ensures that the "edge" I'm
looking at will _exactly_ match up so that it only strikes "rods" or
"cones"...these photoreceptors themselves, by the way, aren't lined up
in any neat "perfect straight lines"...so, if the straight line of
photons from this edge is hitting the retina, it can't be being
properly received by the photoreceptors in that way...

Worse, these photoreceptors are not mathematical "points"...they have
a _surface_, they have _area_...thus, a very tiny movement can make
the photon land on the same photoreceptor but on the other
side...this, though, is not actually perceived any differently...the
photoreceptors, so to speak, just "add up" how many photons are
"exciting" them in that particular area...hmmm, couldn't a photon sort
of strike half-on / half-off such an area? Or, moreover, what about
the "borders" because the "cones" and "rods" _are_ densely packed but
they _are_ also different...that is, the surface areas are very close
that it's microscopically small but the combined surface isn't
continuous and photons are very small things - they are, in fact, just
electrons flying freely around the place - so they really could land
on a "border" and not register at all...

But, most important, having a "surface area" - however tiny that
actually is - means that, conceivably, if you're looking at an edge
then one side of this "rod" or "cone" surface is getting photons
bounced off the object, the other edge getting photons that just
"miss" the object - running parallel to it - and come off the
"background" behind it...but the photoreceptors can only discern
things at the resolution of those "rods" and "cones"...they simply
effectively "add up" how much "photon excitement" has struck _any
part_ of the surface and report this to the brain as a cummulative
signal of all impacts...as noted before, this isn't just
space-orientated "blurring out" here, it's time-orientated too so
"motion blur" is actually the accumulation of many photons that have
impacted those "rods" and "cones" during the period since the brain
last "checked" the visual signals (the photoreceptors just sit there
"collecting" - getting excited by - the light that enters the
eye...they have no real "refresh rate"...they are continuously
"on"...BUT the mind is different...it's NOT continuously looking at
the visual signals...it actually checks the "information so far" at
distinct periods...this is useful, though, or we could never fool the
eye into believing TV and films were "smooth motion" by showing a lot
of still frames very quickly unless there was also this
time-orientated "blurring out"...the eye is fooled when the still
frames are shown sufficiently more rapidly than the brain "checks" the
information coming from the eye...hence, in both cases, it receives a
combined set of information for that time period which cannot discern
between "still frames shown quickly" and "actually truly smooth
motion" that exists in reality)...

So, the "rods" and "cones" themselves are also "blurring out" the
image received because they have _surface area_ and are NOT
mathematical "points" of some imagined infinite resolution...if it
helps, imagine that the "rods" and "cones" are simply pixels on your
screen for a nice simplification where they are big enough for you to
appreciate them and their effects...

Now, if we have a conceptual "sharp line" that's 0.0001 of a pixel in
width that travels diagonally through the pixels at an odd angle, then
this contribution of this line to the pixel's colours cannot be that
finely rendered...a pixel can only have one colour and this is our
"base resolution", if you will...thus, much like tiny, tiny photons -
far smaller than the resolution of our "rods" and "cones" - hitting
the retina, we can imagine that everything within a pixel's "surface
area" is simply an accumulation of all the "photons" that are landing
in it...in fact, the colour of the pixel is the mean average of all
that lands in it...

This is _really_ what anti-aliasing is, in fact...there are no "sharp
lines" and the eye cannot ever discern them...and, yes, the visual
system - in co-operation with the processing of the brain - can look
at the (blurred out) "approximation" we receive and, in that amazing
little way the brain works, it is able to "guess" where edges are...

You cannot actually see an edge...and, no, this is not just what I've
mentioned above...there's another phenomenon involved I've yet to
mention...when the Sun is eclipsed by the Moon, there's a subtle
"bending" of light around the edges of the Moon and that actually
contributes to the "corona" effect you see of the Sun's light just
poking around the edges of the Moon (the Sun actually effectively
"vanishes" behind the Moon but then, magically, some of its light
reappears to create a "halo" around the Moon...very beautiful thing,
that ;)...

Now, Einstein actually owes an eternal debt to this effect...because,
yup, it was the main "evidence" that Einstein really had correctly
re-written Newton...coincidentally, shortly after Einstein had come up
with his theories, there was going to be a convenient eclipse...and
gravity and other factors "bend" the light around an object so that,
yes, you're actually able to see a very, very small part of the light
that's strictly behind the object and were light not bended by these
various effects, it would be entirely obscured by the object you're
looking at...well, they worked out the amount of this effect according
to Newton's equations and Einstein gave his estimates from his
theories...Einstein "won" and that was the main event when people
finally started to admit "okay, Newton might not be 100% infallible
and might just have missed something here"...

Anyway, even on a small scale, there's an element of this "light
bending" going on with _every_ edge you may look at...so, this
mythical "high resolution edge detectors" is a figment of your
imagination...you've effectively invented this "conceptual" sense in
order to (slightly illogically) explain away how blurring a line
actually makes it seem "sharper"...which, of course, is simply
_wrong_, anyway...anyone that looks at an anti-aliased line or other
anti-aliased shape will see an obviously "blurred out" edge...it's
simply NOT sharper...but what it is, is that it's more accurate to
what the eye naturally _does_, what the eye naturally _sees_ and the
reality of absolutely NO "sharp lines" at all...

You've been watching too many cartoons where they draw a big thick
black line around all the objects...absolutely NO SUCH THING exists in
reality..."edges" are discerned by simply finding a discontinuity of
brightness / colour and then the brain does its usual clever
processing to work out "oh, that's a banana I'm looking at and, yup,
that discontinuity is where I'd expect the edge of this banana to be
so that's the edge of the banana"...

99.99999% of your sight is your _brain_, not your eyes...99.99999% of
what you see is your _perception_, not reality...and, yes, there
really is a massive difference...there are many optical illusions
which perfectly demonstrate just how big a role the brain is in
_perception_ and that our perception is, yes, all we know of reality
but it actually _ISN'T_ reality...

Truly, "none so blind as those who can see"...before you cast
judgements on the things you look at, you must also cast judgement on
the lens by which you look at it...you're "conceptual sharp line"
isn't actually really...it's a nice simple abstraction - a
simplification - of reality to suit nice, abstract, simplified
concepts you have in your mind...for instance, stuff about drawing
straight lines on a graph in maths lessons and the "y = mx + c"
equations...this is the way we choose to _represent_ or _notate_ these
things because that is the most convenient way...but don't confuse the
notation what is being notated...there are no "points" (Planck and
quantum theory explains that one), there are no "lines" (not just what
I've covered here about _visual_ lines, there's even more to
this...for instance, if there's not really any "points" then that's
automatically not going to be very good news for "lines"...but, also,
in a 3-D universe, how on Earth can you have any line that has no
"width" to it? Mathematically, that's how we consider most lines but,
in reality, you can't pick and choose which dimensions you want...you
either have them all or you don't exist...it may be as small as small
can be but every "line" - conceptually, even, if we're dealing in this
stuff (it just tends to be ignored in typical mathematical
calculations because any "width" concept is not useful to the problem
or its solution) - must have "width" and "depth"...all dimensions are
mandatory in this 3D universe...the 2D universe often represented on
paper in mathematics is an unreality...an abstraction, a conceptual
nonsense that merely serves a useful purpose...

It cannot be stressed highly enough, don't confuse the notation for
what you are notating...there are no "sharp lines" (even, actually,
conceptually to an extent...though, "conceptual" is the realm of the
mind...so, sure, I can "conceptually" imagine gravity that repels and
works in reverse...I can "conceptually" imagine a perpetual motion
device...I can "conceptually" imagine violating the laws of physics by
getting more energy out than I put in or accelerating passed the speed
of light or cooling something colder than absolute
zero...conceptually, you can literally imagine the impossible and
there are no "rules" or "laws" but the extents of your
imagination...this is _useful_ for mathematical manipulation...I mean,
what actually in reality is a negative number or even zero? There was
an initial reason why there wasn't conceptually a zero and it had to
be "borrowed" from the Arabs to finally make a useful numerical
notation...no-one had imagined "non-existence" as a number...I mean,
it's a pretty weird concept...I'm sure if it wasn't handed to us
already in maths lessons as the number "zero", none of us here would
ever dream of such a thing in a hundred lifetimes...someone somewhere,
though, did imagine this weird thing and it got recognised as a useful
abstraction and it stuck :)...

Anti-aliasing - though you might find it hard to credit - is the true
reality...every single thing you've seen with your eyes or that is
filmed from reality and shown on your TV set is naturally
"anti-aliased"...your brain, of course, is automatically interpreting
this and telling you things like "that's a 'sharp line'...oh, there's
an 'edge' by there"...turning what it sees in that "conceptual"
perception of the world that you have (that we all have :)...again, I
can back this all up with facts, figures, observations and experiments
you may carry out at your leisure to see that it's correct without
my - or anybody else's - interference to "rig" the results...grab
yourself a photographic image from your digital camera or feed
something off the TV into your computer...and then zoom in on
it...it's "blurred out"...and, nope, no "anti-aliasing" was applied at
any point...at least, that is, it simply happens naturally...the
"pixels" inside the digital camera collect all the light across their
surface area and report this back as a single figure...now, millions -
billions - of photons from goodness knows where (yes, though light
travels in straight lines, it may be reflected and refracted...for
instance, like the "halo" effect you get looking at a street light in
the dark...you can see some of the other light that's shooting off in
other directions because, due to the high contrast between night and
the street lamp, the very subtle amount of this light that gets
knocked off its straight line path and ends up landing on your retina
can be sensed and you see like a "halo" around the light and some
"streaks" of light coming off it :) may have struck that same
pixel...zooming in on the picture shows that it's effectively
"anti-aliased" throughout...but there's no "anti-aliasing device"
inside the camera at all...the fact that the light receptors have
_surface area_ rather than being "conceptual pin-points" means it
happens quite naturally...it's a part of the reality of vision
itself...

The actual "exception" here was computer graphics...computers _are_
very conceptual and, yes, we could load one pixel with pure black and
another with pure white and create jagged "lines"...in fact, with
earlier graphics devices, we basically _HAD TO_ do this because they
weren't capable of dealing with a fast amount of colours or anything
(monochrome - 1 bit per pixel - being the obvious starting
point...note that this wasn't a technology problem...TV sets - which
monitors are simply a "high-quality" version of - have long since been
capable of rendering the necessary colours...what made these early
graphics use a low amount of colours - that prohibited proper
"anti-aliasing" - was, of course, _memory_...more colours needs more
memory and graphics takes up an awful lot of memory, anyway...it
needed for RAM to become cheap, basically, for this to be a capability
found on most computers)...

The thing that's actually "wrong" here is the simplistic _aliased_
computer graphics that are now being phased out where easily possible
(texts is now totally anti-aliased in Windows...expect, eventually,
for the whole lot to go this way in due course...there's no mad rush,
though...people are happy with what they see...but, eventually, for
even better graphical quality...at least, "more realistic" graphical
quality, the whole lot will finally be made this way and computers
will finally meet TVs and films)...

Also, another massive clue you've walked straight over...note that
it's called _anti_-aliasing...that is, the method is to prevent a
phenomenon called "aliasing"...what does "alias" mean?
Exactly..."aliased" graphics are effectively _inaccurate_
represenations...we use a jagged "stepped" line to merely
"represent" - be an "alias" for - a real line which is
"anti-aliased"...steps are taken to eradicate this "aliasing"
effect...

The problem here, of course, is that "aliased" came before
"anti-aliased" so there's this sort of expectation that the original
"aliased" is right or something...no such deal..."anti-aliasing" has
come about because, finally, memory and hardware are up to the job of
dealing with these conceptual "sub-pixel" ideas and can render
them..."anti-aliasing" is actually _right_...it should always have
been this way...this is how things naturally are...this is how we've
always seen things...in fact, anti-aliased text, for instance, is
_easier_ on the eye because it's more in-keeping with what the eye and
brain expect to see...mip-mapping and bilinear interpolation in 3D
games tries to get the "anti-aliased" effect onto the graphics because
that's actually more _right_...go back and check out the floor in mode
13h DOOM sometime...look at it with a complex pattern in the
distance...it turns into a mess...the tiniest movement of the player
character makes the pattern jump about a bit wildly...it's not
right...this isn't how things really are...but, hey, you've got to
give room for technologies to improve...texture mapping alone was
miraculous at the time...

To quote Minority Report once more for good measure, "Can you see?" ;)

"You cannot depend on your eyes when your imagination is out of focus"
[ Mark Twain ]

> A visual processing system would have the same problem with
> detecting the edges of the blocky pixels if its image contained
> such entities. A system cannot of course "see" its own pixels.

Well, it's actually interesting...you're not wrong in the basic
premise...the thing that's out of synchronisation is thinking that the
world is a nice grid of neat pixels with crisp, clear colours...that
"aliased" is the default...that, somehow, your "sharp lines at the
conceptual level" is the correct interpretation...nope, sorry, that's
just the useful _abstraction_ and _simplification_ of the true
situation...there are no "sharp" divides or lines or anything
anywhere...none whatsoever...

For instance, there are carbon atoms in your body...an atom of carbon
is not alive...it's an "inanimate object", if you like...but then, on
the human level, you're most certainly animate and alive...where
exactly is this neat, crisp perfect line that divides the two? And, by
the way, whatever you say, no experts in the field agree even what
"alive" strictly means, let alone where the neat dividing line
is...you know like how we always argue about "what assembler?" and
"HLLs versus ASM"? Well, the biology groups' recurring thread is "the
definition of life" and "where do we draw the boundaries?"...where
does red end and orange begin? When does an out-of-tune middle C
(tuned upwardly) finally be deemed an out-of-tune C#? If we choose the
"just" scale instead of the Pythagorean scale - or tune to something
other than "A = 440 concert pitch" - doesn't this effect where we
define the boundaries? Although, in audio terms, absolutely nothing
has actually changed...where does the atmosphere stop and space begin?
At what _precise_ moment when someone has a heart attack are they
"dead"?

Categories and names are merely the necessary way humans have to think
in order to comprehend stuff...it's the nature of the brain...it
creates abstract "exemplars" from all its inputs and, in creating this
approximation, it loosely compares this abstract "perfect" example
with reality until it finds a close match, constantly refining the
abstract model with every new observation it makes...which leads to
humans being able to do some pretty clever things...but, like all
things under the Sun, there's a good part and a bad part...the bad
part of this way that the mind works is that it can lead to
prejudices, stubborn old people stuck in a routine, people creating
"stereotypes", people even falling into "stereotypes" because they are
as equally exposed (the old "oh, well, if everyone's going to say I'm
such-and-such, I might as well be like that"), etc., etc....the
abstract "example" becomes more reality than the reality itself...and
that's when people literally are operating in their own dream worlds
and tend to - because they are working with "facts" that are anything
but factual - their actions tend to cause catastrophes...

This is the nature of human intelligence and, sorry, it's NOT
all-perfect...the very same mechanism that allows humankind to create
wonders can also be used to create terrible horrors...intelligence is
a _power_ that's unique to humans (at least, "our level of" and "as
far as we know" ;)...but, like all powers, there's a _responsibility_
for its good and correct use...this obligation is often ignored or
overlooked or there's not even a realisation that it even exists...and
this, in a sense, is the true "root of all evil"...but, indeed, the
mind is also the "root of all good" too...and we cannot separate the
two because it's _exactly the same mechanism_ - intelligence - that
leads to both and the actual difference between the two comes solely
from individual usage of that intelligence...

In the words of Oppenheimer as he considered the consequences of using
his genius to create the atom bomb: "I am become death: the destroyer
of worlds"...but how was he become death? He was not proud to do
so...he did not want to do so...simple, he proceeded to work with
equations and experiments and so forth...and simply forgot to relate
what he was doing to reality...

Hence, the most important medication for sanity really is "get out of
the house some more!"...go for a walk - preferrably where it's green
and Nature still rules - and remember to _look_...to _realise_ what it
is you're really looking at...consider the beauty, but also the
unbelievable technical precision and complexity...realise that all of
man's technical achievements are still dwarfed by the technical
achievement in one solitary tree or one solitary animal - say,
yourself - and, above all, remember to _CORRECTLY_ put yourself into
this jigsaw universe at the right level...

It's funny, really...when humans strut around pretending to be gods,
who exactly is it that we're trying to impress? Inanimate objects
don't even think...the plants aren't even registering things on that
particular level to even realise that there's anything even
there...the animals, _always_, are completely unimpressed by humankind
and our eccentricities (they either ignore us or run away)...fellow
humans are in exactly the same position that we know that everyone
else is just as limited as us...and in the various holy scriptures, no
deity has ever been particularly impressed with humans...ever...the
only thing I can think of that remains is that this is just basic
insecurity...not having been given any clear purpose, we just grab the
first thing handy and stick to that, come what may...like that Linus
guy on "Peanuts" with the comfort blanket thing, permanently sucking
on his thumb...

"Intertwingularity is not generally acknowledged - people keep
pretending they can make things deeply hierarchical, categorizable and
sequential when they _can't_.
Everything is deeply intertwingled."
[ Ted Nelson ]

> As for our peripheral vision I think it is more a case of less
> resolution rather than it being blurry.

Well, actually, one and the same thing...I mean, wasn't it you who was
just saying about "big blocky pixels" leading to "blurring out"? Well,
the "pixels" in the eye aren't all of the same size...the resolution
differs across the surface...yes, there's less resolution but that's
exactly _why_ it's "blurry"...

Note that vision is 99.9% "blurry"...it's the brain and perception
that "clears up" the "blurring out" (there's actually _extra
information_ inside that "blurring out" that the brain can and does
make use of...when you hold text at the limits of your sight - where
it becomes impossible to maintain a clear focus on the text anymore
(short-sighted people have this limit closer to them but _everyone_ -
even with perfect vision has this - because, again, it's the nature of
sight...there's a heck of a lot of machinery in the eye to attempt to
get a semi-"clear" picture of what is an inherently _blurry_
universe) - and then get your assistant holding up the text to take
one step backwards so that it really is a blurry mess, it's still
potentially possible for you to roughly discern the letters from the
very rough impression of the shape...you no longer can see it clearly
but you still might be able to read it by exploiting _perception_
rather than "sight"...for instance, it's a long, thin blur...well,
candidate letters are "I" - or "L" in certain fonts - and the blur
next to it seems round...like an "O" or a "Q"...and, after working out
a whole bunch of candidates, you might still be able to work out what
the text says because you use your internal "dictionary" to look up
valid words that would fit the rough candidates you think you
see...and, in fact, that's _exactly why_, of course, those optician
charts for testing eyesight merely use _random_ letters and not actual
words...otherwise, people might get the right word by "guessing"
looking at the rough shapes of the blurs they see, even though this
stuff actually really and truly is beyond the limits of their clear
vision :)...

Vision is actually a non-trivial topic...for instance, peripheral
vision is "worse" than the macula sight? Umm, yes, generally...but not
in all instances...perhaps you might have seen this weird effect in a
cinema...wear white trainers or something and put them up in front of
you in the darkened cinema, where your vision has switched to "night
vision" because it's really dark...now, if you look at your white
shoes directly, then, amazingly, they are _easier_ to see when you
look slightly away from them to pick them up in your peripheral
vision...that is, the peripheral vision is sensitive to certain
low-light conditions that the main vision isn't...it'll seem weird the
first time you notice this...your shoes will appear somehow "brighter"
when you're NOT looking at them directly...in ideal conditions with
the exactly right brightness to your shoes, it's even possible to have
it that they are, to all intents and purposes, _invisible_ to your
main vision (not bright enough to be seen) but turning your eye away
from them slightly will actually mean that you can see your shoes
(because they seem "brighter" in the peripheral vision and this
finally makes them visible)...

Another interesting thing is, if you're a sea captain guiding a ship
in the night, having to look out for icebergs by sight...or an
astronomer going out into the field with your telescope to a "dark
sky" site unpolluted by the bright lights of the city (which spoils
the "viewing" of the night sky :)...then if you have gadgets and
instruments to look at, they almost all certainly use _red_ light...in
fact, you might even have yourself a digital alarm clock with exactly
one of those red light displays (I do ;)...the choice of colour is
delibrate...these don't excite the "cones" sufficiently to trigger the
iris to open and for your "night vision" to switch to "day
vision"...the mechanisms involved just simply don't react to red light
but, for other colours, the relative brightness of the light would
otherwise trigger the eye to become accustomed to the brighter light
and lose its "night vision"...so, you've got your eyes accustomed to
the dark but you can stare at the bright light on the instrument panel
(or the front of the alarm clock...often viewed in the dark, which is
why they tend to make these red-coloured displays usually so you can
quickly glance at it in the middle of the night, without it hurting
your eyes or anything...so you can easily fall straight back to sleep,
if it's the wrong time to be awake ;)...so, the sea captain looking
out for icebergs can look at his instruments but this doesn't require
him to spend the next half hour trying to get his eyes used to the
dark all over again (which, of course, could be absolutely
disasterous...well, okay, this stuff is a bit out of date because they
try to back up good old "night sight" with other instruments these
days too...but the essential point is still there and before modern
times, it was 100% dependent on someone sitting out, looking to the
sea...sounds very low-tech but when human sight is fully in "night
sight" mode, it's, in fact, incredibly sensitive and the eyes are
actually a very good instrument for that...and having the human brain
behind it means it's being processed by a very impressive
super-computer...the actual tricky point about using humans this way
is just the problem that they get bored, fall asleep, decide it's
"harmless" to take some reading material with them to relieve the
boredom and, oops, end up more interested in the book than looking out
for dangers...and the such like...the hardware's impressive, it's the
software that's all a bit flakey with human beings...not particularly
designed for the tasks that they are often alloted to fulfill ;)...

Anyway, yes, peripheral vision has less resolution...but, well, grab
an image and let's simulate "less resolution"...load up an image in a
photo package...now, we want to effectively reduce the resolution of
this image...well, let's simply make it a 1/10th of its size with the
"stretch / resample" menu option...but, well, this is just "less
resolution" in the eye, it's still the same actual size, though (I
mean, things don't get smaller because they are being viewed by the
peripheral vision)...so, "stretch / resize" the image back up to its
original size by a factor of ten...if your paint / photo package has
the option, you should _apply_ any sort of "anti-aliasing" or
"smoothing" when reducing it (because the lesser resolution of the
peripheral vision would stil have the same amount of photons striking
it and it would just be a case of "blurring out" because of the
"blocky pixel" stuff you mentioned)...but _don't_ apply any such thing
when zooming it back up (because we simply want each pixel to be made
"bigger" to get things back to the original size to see it
properly...and, thus, we just want simple "scanline / pixel"
duplication...no fancy "resampling" tricks or anything :)...

What you should see is that "lesser resolution" actually does mean
"more blurry"...one is the cause, the other is the effect that this
cause creates...I tend to usually be more concern with results -
"content first" - than methods or technologies to get those results
(as long as it's an accurate method, they are all equally good :) so I
was talking in terms of the "effect" rather than the "cause"...

But, of course, to see that "less resolution" means "more blurry" with
_real_ images, as opposed to the conceptual graphics that get drawn on
computers (this stuff is an "approximation" of reality and the main
issue that most computer graphics suffer from in looking realistic is
that they are too simplified, too perfect, ignoring the "blurring" out
effects, etc....DOOM III looks absolutely fantastic and incredibly
"real" because, using bump-mapping, they've put the "roughness" and
"non-uniformity" of the real world into the graphics...pick
practically any object around you and look _real close_ at it...only
metals or plastics will not actually be "textured"...run your finger
over it slowly because the sense of touch in a fingertip is
exceptionally sensitive...as Fred Nilsson - an animator on the new
DOOM III game - put it, "People probably won't understand just how
much details there _is_ in the world"...our visual system is
delibrately trained to abstract - kick out as much "irrelevent"
details as possible to get to the "essential essence" of an object -
that if you really stop to look and _really look_ at the world, the
detail _is_ stupidly massive...that's always been, really, the biggest
stumbling block with creating totally realistic computer
graphics...they always look too perfect, too polished, too flat,
etc....too much of an "abstract mathematical model" of how we think of
the world rather than what it _actually looks like_ ;)...

> Also less processing so
> it is mostly limited to large areas of texture and movement.

[ Okay, one thing...be careful of the word "texture"...it means one
thing in computer graphics and something completely different
elsewhere..."texture" elsewhere means that rough pattern most
materials have and is a _tactile_ thing rather than a visual
thing...the "texture" of a material in artistic terms, from a visual
perspective, would be how light bouncing off this rough pattern so
that, yup, you can see that something is made of fabric by its
"texture"...but, note, that with a smoothly plained piece of wood (so
that it has no real-world "texture"), the "wood grain" is NOT actually
part of its "texture" at all...that is a marking _on top_ of the
texture...although, in computer graphics, "texture" has come to mean,
in fact, almost the reverse...that "texture" is the markings...which
is actually one of those nasty accidents of history...what's called
"bump mapping" is actually far, far better to be referred to as
"texture mapping" because it really does give things "texture" in the
real-world sense of the word...but, unfortunately, the earlier methods
of approximating this on a flat polygon - that is, simply slapping a
bitmap of some wood grain or whatever - got called "texture mapping"
and the word was taken...taken too early, in fact...if we're talking
about peripheral vision rather than computer graphics then it's best
to use the jargon in the biological field to avoid confusion...and
they'd almost certainly refer to this as "large areas of colour,
brightness and movement", leaving "texture" out of this
completely... ]

> As
> for color try viewing a rainbow strip of random colors held just
> outside your foveal visual area.

Okay, ummm...I can rig something up with some differently coloured
paper I have lying around to attempt to follow along...right, I can no
longer read the text that's printed on them (they are some "spam"
leaflets printed on different coloured paper to, I suppose, entice you
into reading them, if you're wondering :) but I can still make out the
"strips" of colour...

> D
> D
> D
> D
> D
> D
> D D D D D
>
> For example you can see the D's with your foveal visual area
> and left brain processing areas and the L shape (without moving
> your eyes) with your peripheral vision and the processing areas
> in the right brain.

[ Okay, where the images are being processed isn't completely relevent
here because we're talking about what we perceive in a visual
sense...I mean, the left brain / right brain stuff is involved, of
course...but it's the "artistic" results we're interested in here,
right? In which case, the brain can be simply considered a "black
box"...we stick things in front of our eyes and then attempt to
describe what we see...yup, the brain is doing this processing but
_how_ it does it is the realm of brain surgeons and neurologists, not
artists or graphics programmers...it's not wrong to mention this
stuff, it's just extraneous information...not particularly useful for
the discussion at the moment... ]

As for your "L of Ds" then, yes, when you look straight at it, you see
the Ds and you see the overall L shape...the main macula part of the
retina is the part that can see things in fine detail and we can read
everything just fine there...but, when you look to one side to view it
peripherally, it's true that you no longer really see the "Ds"
anymore...just the basic overall "L" shape and perhaps - at least,
this is what I'm seeing - that each of the "blobs" making up the "L"
shape do seem to have a rounded right edge to them...no, if I hadn't
seen that they were "Ds" already, then I couldn't say that they were
"Ds" merely from my peripheral vision (well, not without making a
really lucky guess that they are letters and then coincidentally
picking the right "right edge is curved somehow" letter out of the
alphabet...like "B" or "D" or "P" or whatever...in other words, it's
not really vision there, it would be perception instead ;)...although,
if you move your eyes away so that it's really only just entering your
field of vision at all, then it actually becomes difficult to even
make out that it's an "L" shape and certainly not anything about the
letters that form the shape...at this point, the "resolution" is now
so low, it's become "there's some sort of dark 'blob' thing
there...can't make out its shape at all"...if you like, it's all
become one or two massive "pixels" with awful resolution, just telling
us "something dark in this general area"...

But this is all to be expected from what I've been saying...peripheral
vision has much lesser resolution and, therefore, the "surface area"
covered by each "pixel" ("rod" / "cone" in the eye) is bigger and the
"blurring out" is more pronounced...as noted, you can't see anything
below the resolution of your pixels ("pixel" does, after all, mean
"picture element", where the word "element" should be stressed...as in
"it don't get smaller or more decomposed than this, folks!", as these
are our "elements" ;)...

> My interest is actually in visual processing and recognition.

Then, straight off, the first thing to realise is that the world is
not uniform, it's rough, it's blurred, it's not "neat", edges aren't
clearly defined by a thick black outline like in the cartoons, etc.,
etc....

I know I don't often give off much of a "professional" image or
anything but this isn't just fancy talk of some personal opinion...I'm
a general programmer but graphics is where my main forte' is and the
commercial AI work I was involved with was actually specifically about
processing visual data (well, actually, there was more than just
visual data to be processed but inputs from simple cameras that needed
AI recognition was a large part involved in it)...

And the world is actually a bit of a mess visually...most people
perhaps have no appreciation of just how messy it really is because
the brain does such an excellent job of perceiving it and "clearing it
up" conceptually and creates order out of the chaos...but, as I try to
stress, when looking at an object, you also have to consider the
_lens_ you're looking through at the same time...perspective, for
instance, comes from the distortion of looking through a curved lens
at the world and the way light is focussed from a cone towards a
conceptual point...this, though, of course, is quite a delibrate
"design decision" for the eye because this "distortion" gives you a
very useful perspective on the world for comprehending depths and
automatically means things in the distance are smaller and lesser and
don't take up as much space...because, well, things in the distance
tend not to be as important or significant as things close up to you
(a lion in the distance just means "okay, there's a lion over there,
let's walk over here to avoid it"...a lion right in front of you means
"okay, I'm in deep, deep crap here, I'm liable to be eaten any second
now...time to panic / run / fight / anything, as my very life is in
danger" ;)...plus, with two eyes, the brain can process both and check
the differences between them and gains even better "depth perception"
by exploiting the distortion of perspective (perspective means
implicitly that closer things "move" faster than distant things across
your vision...hence, by holding the eyes slightly apart so that they
get similar but slightly different views of the world, the brain can
do some "triangulation" of what it sees and, thus, things become "3D",
even though, technically, the eyes only see _2D_ images
individually...the third dimension of "depth" is, in fact, entirely a
"perception" thing, not a physical visual thing at all...we don't
directly see depth...we see differences between two distorted flat
images and then can "extrapolate" the depths, thanks to the distortion
of perspective :)...

So, the first thing to cultivate when dealing in a subject like this,
is an appreciation of the subtle yet great difference between "vision"
and "perception"...they really are amazingly different...

Did you catch an earlier post I made about a BBC documentary series
into the human senses where they prove the point in a staggeringly
amazing way? They made the point in one of the most effective ways
I've ever seen...at the start of the programme, the presenter asks you
to record the show so that you can look back at it later...then you
forget all about this and watch the show, where he demonstrates all
the different things about human vision...then they reached a section
where they wanted to demonstrate the essential difference between
perception and vision...how people can actually stare straight at
something and, yet, _still_ don't see it...they had a section of film
where there was a group of people passed around basketballs between
them...then, suddenly and quite comically, on walks a man wearing a
big monkey suit...he walks into the middle of the picture, stops and
looks straight at the camera and then walks off again...as the
presenter says after showing us the footage: "there's got to be
absolutely no way people could NOT see this man in a monkey suit,
right? Wrong"...they showed the film to some volunteers but carefully
prepared them by giving them a particular task to do...they were to
count the number of basketball passes the group are making...that is,
they delibrately directed people to _concentrate_ on the group of
people to direct where they were looking...and, yup, only two people
of about fifteen volunteers actually saw the monkey man...and, no, he
wasn't in any way attempting to hide at all...I mean, the monkey man
just stands there in the middle of the screen and looks toward the
camera...the two who did see the monkey man (weren't concentrating on
their task properly, methinks ;), were laughing their heads off that
everyone couldn't see something so obvious and quite, quite surreal
and silly...

And, yup, the viewers at home were also laughing, I'm sure...I
was...but then the presenter said "don't laugh too readily...you've
just done exactly the same thing yourself"...and then they rewound the
programme to earlier sections and, yup, there he was...Mr.Monkey man
just sitting or standing in the background while the presenter was
talking in the front...again, he wasn't particularly hiding
himself...just not dancing in the foreground to draw attention but
simply sitting around with other people in the background...but,
looking again, knowing he was there, it was crystal clear and obvious
that not seeing Mr.Monkey man was the perfect way to demonstrate the
very big difference between "vision" and "perception"...visually,
everyone watching undoubtedly _saw_ the monkey man with their
eyes...but their brain did not "perceive" the monkey suited man at
all...and, of course, if you'd recorded the show then you could prove
to yourself at your leisure that, yes, he really was there all
along...the presenter, in fact, asked people to record the show
because, for sure, many people watching would simply not believe that
someone could so easily "hide in plain sight" - like a Ninja - and
they really didn't see it...I mean, the monkey suit wasn't something
you'd naturally expect to see...delibrately chosen for its extreme
surrealism and comic value, I'm sure...how could you NOT see a big man
wearing a big gorilla suit when he stands right in front of your eyes?
But it _is_ possible to completely miss even something this obvious
when your concentration is somewhere else...

Unfortunately, I don't know of any way for you to also see this show
(and the point might now actually be ruined because you'll know to
look for the monkey man...and the point is more brilliantly made when
you do actually fall for the trick :)...but, well, here's a URL to the
BBC website describing it...

http://www.bbc.co.uk/pressoffice/pressreleases/stories/2003/06_june/17
/human_senses.shtml#two

And, yes, there was also an amusing trick where the presenter asked a
person on the street a question and then some builders walk passed
them carrying a door...then they switch people completely and this
other person carries on asking the questions...almost NO-ONE was able
to spot that the people had switched and that they were, in fact, not
talking to the same person they were talking to before...the door - a
bit like how magicians do this sort of "slight of hand" all the time
for their shows - had concealed the actual switch and the two people
_behaved_ as if they were the same person...

I Love these little BBC documentary shows because they make some
really cool scientific points using a really fun and funny way of
demonstrating it...Sir Robert Winston is another one who does shows
like this and he's great because he makes his points in the stupidest
ways...he suffers for his art...like, in order to demonstrate how the
body reacts to cold, he stripped off his clothes and then stands in
one of those freezing meat lockers and uses himself as a "guinea pig"
to demonstrate all the unpleasent things that happen as your body
freezes up...a complete nutter to put himself through all these mad
"experiments" but that's what makes his shows so great...he makes the
science easy but also makes it stupidly funny...you know, you sort of
panic if he utters some words like "the human mind reacts to fear
by"...because the next scene is almost inevitably going to be him
throwing himself off a building or having a fight with sharks or
something that will "demonstrate" how fear and danger effect
humans...he's never afraid to prove his science with a practical
experiment...no matter how insane...oh, and he's got the knighthood -
the "sir" in front of his name - because as well as being a nutty TV
scientist, he actually is a bona fide world-leading expert in
fertility, involved in inventing IVF treatments and things...he really
_is_ a scientist who knows his stuff...he just happens to be an insane
extravert too, who happily makes a complete fool of himself on TV "in
the name of science"...knights have come a long way since the days of
King Arthur, eh? ;)

> My routines all assume the images are in byte arrays. Thus if
> I use assembler, DJGPP or VC++ I have to have a different version
> of the Display(image) routine. In DOS I do display color images
> in mode 13h using a standard palette or a palette generated from
> the colors in the image itself. The actual images however are
> kept as arrays of bytes.

Well, yes...of course...memory itself is nothing but an "array of
bytes" so I wouldn't expect anything different...

If what you mean is, is that you keep it as a "device-independent"
image then, yup, that's a sensible approach...but, for display, it has
to be converted into another array of bytes - in the video
framebuffer - for actually being able to see it...thus, my suggestions
weren't necessary about actually changing the image itself by reducing
the colours or displaying two images that "blend" to make a good
approximation...these images can, of course, be completely generated
in memory only for display purposes...there's no compulsion to
overwrite the original image and change it or lose information...but,
well, as mode 13h does not have the "oomph" in it to display the
24-bit images in their original glory, you'll need to generate some
sort of "approximation" images which convey the image...

This doesn't actually mean changing the original image at all, if you
don't want that...do you know about MIDI files, perhaps? Well, there
you have a sort of "description" of the music to be played...which
notes are being played, when the notes are turned off again, change of
instruments...not any actual audio data like a .WAV waveform file but
rather "commands" like "switch channel 1 to the piano sound / play
middle C / rest for a minim / release the middle C note"...and, during
playback, the MIDI player - a software synthesiser in Windows or the
sound chip inside an electronic keyboard or something - interprets the
commands and does what they say...thus, MIDI files are actually tiny -
say, 20KB for a typical pop song which would take more like 4MB in MP3
form - but, well, the quality of the sound isn't guaranteed because
the MIDI device is in charge of how the sound finally sounds...

Same thing here, basically...your original image may be a bitmap but
it's perfectly possible that it's "device-independent"...an abstract
"description" of the picture...and then the things I was suggesting
are merely ways to "render" the description on a particular "device" -
mode 13h - to try to get a good approximation...so, sure, your image
is an array of bytes...leave that alone...just read from that array of
bytes for working out how best to render the image...

Again, it's an attitudinal and conceptual thing...draw a distinction
between what you store and what you render...they need not be the same
thing whatsoever...in fact, if you want to get the best approximations
or simply the most compact storage, it tends to be _best_ - smallest,
anyway - when they are not similar and code is responsible for
rendering some basic description onto the screen...

> Thank you for your response,

My pleasure...I Hope somewhere in the maze of words, there's something
useful for you there...

Beth :)

• Previous message: Randall Hyde: "Re: Solaris and HLA"
• In reply to: JGCasey: "Re: 8 bit values in mode 13h?"
• Next in thread: sinewave: "Re: 8 bit values in mode 13h?"
• Reply: sinewave: "Re: 8 bit values in mode 13h?"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]