george koller on mon 22 nov 99
I have been asked to resend part#1, and I know how
this works, so although this breaks all my own rules
for the proper length for a posting, I've combined it
into one relatively seamless whole.
george
=====================================
INTRO
Rather consistently we have been asked over time to
explain what our process does in "non-technical"
terms. This has dogged me for a long time and although
I thought I was getting "incrementaly" better I have
learned the hard way that some artists seem to easily
"turn-off" at the mention of an unfamiliar technical
term. Leaving me wondering how to explain a rather
involved process in "non-technical" terms. Well,
it may be a bit long, soooorrry, but I have, I hope
here managed to do just that.
===========
YOUR ANCHOR POINTS
===========
Our software is dedicated to working with clay
surfaces, and only clay. Already there are more than
a few allowances for our choosen material. We started
with Sign Software, what we do was similiar in our
first edition, we are well ahead of that now.
What goes IN to my software is an IMAGE and what
comes OUT are INSTRUCTIONS for a machine, please
keep that as "anchor" points as I fill in some of the
details between. '
The INSTRUCTIONS are like a paper tape loop for an
old player piano. Only we control a machine that positions tools.
And the TOOLS can be simple scribe points, or as high
techy as a Laser. ALL our work to this time has been
with a rotating diamond burr - like a Dremel TOOL.
It merely ASSISTS us, not unlike if you found a masking system
that works (a little magically). For all previous work our
one trick has been to REMOVE MATERIAL.
We stand now stand ready to add two new process tricks.
======
CHAPTER ONE, THE SAGA BEGINS...
========
Modern software supports a building module called an
"Object". This allows computers to do more work
for us programmers. These software building blocks are
rather like the blocks of clay used in a sculpture of an artist
and I have been sculpturing for several years on this - it
is not completely trivial - as to say each Object has many
details.
One of the things these OBJECTS allow us programers
to do is to avoid a single unbelievably complex "monolithic"
chunk of clay and divide the particular "problem domain"
down into individual parts that will usually model something
in the real world. My particular scepter has about 50
base OBJECTS, some of these objects are duplicated for
for me by the computer tens of thousands of times, and some
are only used once.
In the work I have set out to do I must get my objects to
"learn" and "model" all the details that are in an image.
These details must be in some useable format. Therein
lies the whole trick. Boy, Gail I thought this was going
to be rather easy when I started, and now I am humbled,
honestly humbled, by the power of what I think this is about.
Before I go into more details, let me give you an example
of what this thing can do. I can take a digital picture of
something like a barn with a stone fence, and a flock of
multi colored chickens running across our scene. Lots of
details! About two hours later I can be looking at the details
of a particular body chickens leg in the upper right hand
corner of tile #17 and decide I want to move it, resize it,
or delete it. When we have what we like we can tell the
computer to go to work again and it will then generate
(likely) hundreds of thousands of details for our tools to
follow to REPRODUCE every detail of YOUR IMAGE
step by detailed step i.e. the movements to remove glaze
from a single "polygon" of color might be defined by over
1000 points for a tool to move to, then to, then to and so on.
(This is a tedious approach, but suited to the tools we
are using, and we humans are excused while the silicon
chips control electric motors which turns screws.. and so
on. )
Below are the eight "Phases" of turning some sort of Image into
a tile Mural. It all starts with a real or imaginary image somehow
put into a BIT MAP format:
A. Real Scene (Ancient Mosaic, a stone fence....)
B Photograph or or Abstract Image
C Digital Scan of any of B, (Or skip A&B and use
that Digital Camera your going to get for Christmas)
D. Vectorizing Program (I.E. StreamLine by Adobe)
E. My Software
F. g_Code
G. A Machine that runs g_Code (somewhat like an engraver)
H. Tiles
Let's start by saying that the relationship A, B, & C are all
rather common sensical everyday stuff up to the point that you must
understand that your image from a scan of anything, or a digital
camera is in a "bitmap" or "raster" format. This is how the monitor
in front of you is working. One Pixel has One color
============
GAIL GOES TO LAMBAU FIELD
============
Here in Door County we are 40 miles north of Green Bay, home
to the Packers and Lambau field. For our exercise, and your home
work we are going to go down to Lambau field, your friends are going
to view your work from a dirigible floating above, and I'm going
to teach you about all about the two types of Image Formats.
You wanted to know what my software does... well here is a photograph
of an image that a client would like a mural of. The mural will
be 100YDs by 34 YDS, or 3,600 Inches by 1,224 Inches. There are
4,406,400 one inch squares of field down there. Here are 4,406,400
one inch squares of paper. And there as many colored paints as
could ever want stacked up along the sidelines (there are exactly
16,777, 216 different colors to pick from here)
Well your first job is to go down there and color each of those pieces
of paper, and place it exactly on the field to best represent the
corresponding location on your photograph. When you get back, you
will know a lot about bit maps. (Now please think how you would
go about this..... and complete it in your mind.)
=============
Now we are ready for Vectors, the OTHER way to deal with Images.
Understanding Vectors is not trivial, I'm sure a semester could be
dedicated to it in Art School. In our context you can think of Vectors
as curves that define a boundary between solid bodies of colors. We are
at step D. You have a bit map image on the field now, your next job
is to "Vectorize" it. There are many ways to "vectorize" a single
bit map image.
Here are the tools you have:
1. an unlimited supply of pipe cleaners
2. some sort of "color averaging tool" (digital of course)
and here are the general steps to follow:
1. Find a body of "similar enough" colored pieces of paper.
2. Take your pipe cleaners and shape each one to approximate
the edge to any "other" color (or the field limits).
3. Fasten the pipe cleaners to each other.
4. Now use your "color averaging tool" and paint all the pieces
and partial pieces of paper inside the pipe cleaners with
the exact paint for this color.
5 Go back to #1 until you have surrounded every one of those
pieces of paper. (!)
Congratulations, Gail, you have VECTORIZED an image. You had
to make some compromises (or you just wrapped 4,406,400 pipe cleaners
around 4,406,400 squares). But, grasshopper, you have learned
to vectorize. Now go back out there, take this pile of binders,
and make a label for each closed loop (polygon) of pipe cleaners. Then
for each pipe cleaner in this polygon write down the distance from the
South-East corner for four points to "peg" that curve. That is 8
numbers
to define each "Pipe Cleaner" (or Bezier Curve - Nikom?).
You probably had to use something like 1 million pipe cleaners to
do a detailed "vectorization". Still feeling like continuing? Well
type in those 8 million points and I'll be right back, ok?
=====
If you typed all those numbers, without missing one digit, and put
them in a certain format you wrote a POSTSCRIPT file. (ClayArter
Tony Hanson used to work with these files) This is the file that
program will read.
Computers do some things very well, for example the 8 million points
you typed into that file will be read, and "filed away" inside the
computer in about four minutes. The vectorizing that you did with all
those cans of paint, and pipe cleaners is done by such packages as
StreamLine(TM) by Adobe in 5 to 15 minutes on a state-of-the-art
Pentium type system. This program is not in the least artistic, it uses
simple rules, which you can modify to MECHANICALLY grind away
at making the decisions you made as an artist. THERE IS NO DIRECT
COMPARISON, NONE, ALL THIS PROGRAM CAN DO AT BEST
IS FAIRLY REPRESENT WHAT AN ARTIST GIVES IT. But think
about this, what if you give it a digital picture you took of a
wonderful old barn, or a tremendous stone fence, or a vine covered stone
house?
In any case like this the software can alleviate a terribly boring
task of Representing, NOT INTERPRETING, and provide a
very real service.
Also if, as an artist, you compose your image using certain "paint"
packages no vectorizing is necessary at all! Your work is
being built and represented by vectors AS YOU WORK. This is how
ALL our work has been done to this day, except two pieces which were
processed from a photographs. (They were only mildly successful, but
this is why I spent the last year spiffing up my software.)
Vectorizing is something of a necessary/pragmatic tool in some cases
that can perform "miracles" with some art, make a mediocre image
look great with one image, then frustrate the heck out of you on another
piece. Using a vectorizing program is an art in itself that many graphic
artists learn for one reason or another.
One more note on this: as the world goes more to computers, many
things that are designed, such as houses, ships, and every type of
product is available DIRECTLY in file formats directly convertible
to Postscript. Just today I got an order for multiple copies of a tug
boat build here in Sturgeon Bay, from the engineers that designed it.
The image will come directly from their file, which is a composite
of all the parts they designed. I can be converted directly to
PostScript.
My software starts by reading a postscript file of an image we
wish to produce a mural of. This puts us at point D, and touching
on E above. I have spent hours writing and rewriting this, if
you or anybody on ClayArt wants to know the exciting details of
what goes on between steps D and H, the last step just let me know.
==============PART I -> PART2 ========
When Gail finished "vectorizing" Lambau, she was at D. This
is what my program will read. In the time it took Gail to run
to the sideline, our silicon friend could "upload" and
sort out those 8 million points she typed in for us. But, we also
task the computer to do all kinds of things, during this upload, like
spreading image precisely over so many tiles from left to right,
and from bottom to top. So there are a modest number of things
which come from the operator interface - dialog box with pull
down item lists and etc. Nothing that would stump an eight year old.
Gail can walk proudly, SHE vectorized the image, NOT the computer
and it is HER TOOL TO CONTROL all along the way. Adobe
Illustrator will give an artist COMPLETE CONTROL to zoom-in,
merge, move, recolor, etc until she is completely happy. Also
please remember that if "building up" the image rather than
translating it is what you want to do (Paul Lewing?) then this is also
100 percent possible (I.E. using a "electronic sketch pad").
Computers, while I'm on the subject, can't "generate" anything!
They can't even generate a true random number. They are slaves.
Fast, but stupid, and they are nothing but another tool.
Now we need to do something tricky. We need to teach a saucer
sized Smart "urchin" to do certain types of operations on each of
those pipe cleaner surrounded polygons. Can you visualize this?
The average polygon might include some 150 or so pieces of paper in
a detailed image. Many bigger, and many smaller, just 150 as
an average. This gives us an area about the size of our keyboards.
(This whole project is being done in 10,000 times the actual
default scale.)
In order to get going we have to train our super-urchin to only
move along the outside edge for a key operation I call "Trace".
There are some 12 operations we have trained our urchins to
do, and more on the books, but this is a basic one. The urchin
might be taught to position its right side snug up against the
pipe cleaner. Then always go forward, and forward, this will make
for a CounterClockwise Smart-Urchin. Anyway, our CCSU
Object is ready to go as soon as he is put into a Polygon of
pipe cleaners and shown where the edge is. Put him in there
and watch him go!!
A "real" CCSU will keep track of where he has been by leaving
a trail. And depending on "overlap" and other factors will also
leave various "marks" on those pieces of paper. Of course I'm
leaving lots of stuff out, but I hope you get the picture. Still and
all we have one more step to have something useful: We attach two
tensioned strings to our CCSU fellow. We can attach these
to the a flag post on the top center of the CCSU. Now we have
something! If we stand on the sidelines and record the movements
in the string every time the CCSU center is square over a piece of
paper as our CCSU moves itself about, WE CAN NOW RECORD
the CCSUs EVERY CRITICAL MOVEMENT.
What is this long boring recording worth? Well, think about
this, will you....
1. You find an urchin of the same size, and train it to cut the
grass, remove a glaze, dispense a liquid, texturize, remove
clay, add wax resist, or doing any ONE thing useful for
you.
2. Call this a CCTU (Counter Clockwise Tool Urchin - of
course).
3. You put the CCTU on some strings, powered by some electric
motors, and driven by another computer, that also has
software
that they call "Controller" software - not expensive, not
exotic.
4. You feed the computer that "moves list" we recorded, the
computer tells the motors to turn, the strings move, and our
CCTU is pulled and pushed around while he is busy doing
is one task.
Whallla! You can go home, turn off the lights, at Lambau
and go "unwind" with the CCSU. When you come back Lambau will be
cut, dried, wax resisted, routed out, dug up, dusted off, or whatever
the urchins that you tied the string gismo to were trained and equipped
to do.
= = = PRE-CONCLUSION: A thought on this Process = = = = =
Screen processing, and some other "printing" processes are already
"computer driven" and making inroads into the "cover a tile with
anything" markets. Well, that is whatever that is, but it is not
"ceramic art", not to my mind. This is about a chance to make
decorating tiles with traditional ceramic materials and processes
somewhat more competitive with those non-ceramic processes.
We believe that anybody knowing exactly what we are doing would
be at least three years in software development before they could match
where we are now. My career has spanned 25 years of R&D departments
and leading edge software development, I also believe this to be true.
In addition to Engraving type machines (CNCs) which are economical but
limited , the software could as well decorate regular 3d items such as
sinks, and other items large and small.
As the software can also estimate, simulate (to printouts), and help
with visualization issues there would seem to be opportunities to bring
interior designers, architects, and such in closer contact with the
artistic
process. This is all, frankly, rather foreign to me, but it is very
clear that
the Web could play a role that might excite some of the venture
capitalist types
that I have been reading about (I don't think they are covering Door
County).
Perhaps a progressive School and University might be looking for an
opportunity to combine art and technical disciplines in a truly new and
unique way. Photography, programming, and the ceramic arts do not
cross in any significant way that I am familiar with.
We have been getting the results of a series for glaze and color tests
these last several weeks and are very pleased. Our first decorating
project,
depending on the delivery of our new 8' by 4' bed machine should be
within two months. Our intention is currently to decorate large
handmade tiles, for such uses as trays, furniture, and "panels".
=================
POST CONCLUSION : And just one more thing:
=================
It has occurred to us, most seriously, that this system will be ideal
for replicating some of the world's best in-situ mosaics, and other art
pieces for full-size or precisely scaled displays. A set of quality
photographs, very careful handling all along the way, shared knowledge
of historical materials, and we should be able to replicate a historical
piece for museums and galleries with unmatchable accuracy.
The software is IDEAL for this type of challenge. For work related
to this type of challenge anywhere in the world, I would be forever
most grateful.
==========
Best,
George Koller
Sturgeon Bay, WI - Door County
| |
|
