|
|||||||
|
|
|
|||||
|
|
|||||||
From time to time I do some Ray-Tracing, recently I have gone through
another revival of this art.
Here is my latest one, please feel free the nick it and use the
text box for other purposes, on the grounds you retain the artwork
on the side of the mug.
I am experimenting with illustrating by comparison just how crummy the graphs produced by Excel are.
As I was producing this, I needed something to put on the mug, quickly knocked up an advert for my Civil Liberties site, then realised how helpful a vehicle images like this could be on a medium where people nick and pass around cool pictures. (Thanks largely more these days to Google's® Image Library and their pet spider.)
After looking at this online, I saw the scaling-down used in
post-production has smudged some of the text; I have re-rendered this below,
with most of the text inside the (larger) picture.
This "modern art"-looking piece is infact the successful result of an experiment into statue positioning by a sensible definition of posing, one which also allows easy animation.
I have advanced this further, inserting sensible angles and proportions for the legs, and adding definitions of hips, spine, and shoulders.
Vector maths were used to "fill in" the trunk, which is a
more complex task than it may appear: Cylinders and spheres find their own
loci, but explicit definitions are required for floating polygons. 8
Triangles are used; two for the abdomen, two for the thorax, and four repeat
these round the back. I haven't defined it here, but model will fill
correctly for a twisted posture, where the shoulders are set at a different
angle to the hips. In the same manner as you can't wobble a three-legged
table, you can't twist the triangular polygons used here.
An emergent problem is that the legs seem too far apart. This is mostly
due to an optical illusion related to the width of the legs being far too
thin (so I can see what's going on more easily), rather than the natural
trouser-width that they will end up with, and partly due to the legs being
defined from the feet up, rather than the hips down.
It's all based on me and masking tape on the carpet, and measured angles
of my legs, and proportions. It's not easy measuring from my hips floating in
non-virtual 3D space, so I didn't do it that way. Now I'll need
3D-trigonometry to set sensible limits, with the left-hand leg following the
right along a fixed pelvic-width. Tricky. (I may have to bump along with the
rest of the scene then come back to that later.)
I see Nike™ have used a "Lineman" in their Baseball Ads.
These images predate all that, and have not nicked it from them. I just
hope they haven't nicked the idea from me, but don't worry kids, this is
going in a different direction, and won't end up looking like stick figures
(I hope).
By a related method, here is a tree draw by a generator that I wrote. Looking at 2D Previews and 3D lo-res tests is clearly not enough, and some parameters are going to have to be tweaked, but it still look's passingly treelike.
Believe it or not, the algorithm used for the "tree" is
simpler than the one for the "legs".
Speaking of posing statues, how do you show the viewer which figures
represent who? By giving them stereotyped artifacts, that's how. For
instance, giving a neutral manequin a bowler hat and umbrella, and suddenly
you're looking at a City gent. This particular silly hat was created for
a similar vein.
Imagine you're an infinitely small bug, living on a mote of dust orbiting a water droplet in an American police control room (or "Dispatch", as they call them). No? Well, this is what you'd see.
It's the result of an experiment into wallpapering a sky-sphere with
a real-life image, as opposed to stars, sunny sky, &c.
The spherical mirror thingy in the middle is a test object, reflecting
the sky-sphere back onto something sensible. I had to stretch the picture
around so it made sense, and added text like "Behind
You!" at appropriate places for future guidance. The
mind-twisting optical effect is produced by lazily using an infinite plane
instead of a finite body where the ground should be. Going infinite project's
the Dispatch Operator into the celestial sphere and put's one infinite object
inside another. Maths has no problem with this, but looking at it's another
matter.
This is intended to be used with an outside background view,
and an everyday object on a car-bonnet, only part of which is visible,
non-infinite, and nowhere near any perfect mirrors, spherical or otherwise.
Focal blur will be added, with the focus on the everyday object.
This will (hopefully) create the illusion of photo-realism, nicking
an idea I saw in a Computer Art magazine where an in-focus textured
gravestone was infront of a blurred church and very-blurred somethingscape
background. This looked to the untrained eye very much like a photograph,
and no detail was required.
Now, what could this one be about? (The main image is as used in my CV.
I have managed to add comments inside the GIF file, incase anyone nicks it
and an employer finds it.) This WAS, like many of the above, a
work in progress,
here highlighting the poor quality of many PC fonts, with the prior image:
Look how those lines wobble. I can't afford the original version, so I
used a derivative. 'Looks the same in MsWord et al, but here you can see
those lines look hand-drawn, as opposed to plotted in proportion. I'll
switch to Timrom, the public domain version of Times Roman. (And no, I don't
know how it can look identical and still be within the copyright laws. [But
it is.])
Font oddities beside, I have developed a black-marble-with-white-veins texture for the lettering(, and had to adjust the lighting angles and backdrop ambient lighting to get the effect I had in mind).
I will re-use a version of this with animated Magna-style highlights in a
substitutible Applet on the main site main page, near the rotating bullet;
where the static picture of engraved "CV" lettering is currently.
Here is a GIF version of the prerelease:
Here's a blown-up debugging trace of the animation, with the
full paths (or the little "balls") shown. Not prizes for spotting
the problem; I've missed aligning equations for the planning and plotting of
the blue highlights. You can see the "miss" in the above animation,
where the blue highlights pass through a part of the 'V', instead of around
it.
The Ray-Tracer used in this is POV (Persistence Of Vision), the script files are edited in !Zap. POV only interprets the script files, which are written by me. These pictures are my copyright, though under UK Law I may choose not to exercise this right and release some of them in the Public Domain.
Later on, tests of adult (NO! Not that sort of adult!) cartoons will be published here, thanks due to Transparency International for the underlying idea.
