Wanda wrote:
>Maybe I missed something Mike, but how are you doing the big drawing? I
>saw the spray/ink bottle - but don't understand the process you are using.
Well, you know that I am trying to go more and more directly from my ideas
into objects and I'm making increasing use of my electronic tool kit. So
the current batch of problems I've set out for myself present a number of
interesting hurdles -- some physical (like how to get paint to flow evenly
for a LONG time in very small amounts without drying or pouring), or just
how the heck to hold a pen on the machine without breaking it or skipping)
-- but the largest hurdles are conceptual having to do with image analysis
and coding...
The current project under production is a brush painting on large paper
(66" x 22" image) in 14 shades of sumi -- a 200 x 600 bitmap of the 'plan'
is displayed to the right of the web-cam view at
http://lyonstudio.inetcam.com/ -- the bottle pictured underneath is a
plastic condiment dispenser, modified by epoxying a water-color brush into
the nozzle/lid -- the brush was cut off 1/2" up the ferrule and a #60 wire
drill used to drill the center out of the brush leaving a small ~1/60th
inch hole -- the valve gizmo is no longer used -- it is left wide open and
is attached to a 'breather tube' to relieve the vacuum as the sumi/water
mixture is depleted -- the sumi flows slowly and continuously to the brush
-- the bottle is held in another very simple home-made plywood gizmo upside
down over the bed of my ShopBot 3-axis machine with that large paper taped
down onto the bed and I've programmed the machine to "drill holes"
(actually just touching the brush to the paper laid on the bed of the
machine to paint a 'dot' and then lifting the brush about .2" off the bed
and moving to the next spot to be dotted) -- surprisingly, it seems to
sorta work, and it appears that when the painting is complete sometime next
week (it takes a long time to paint 120,000 dots!) there ought to be a very
coarse rendering of that bitmapped image on the paper in sumi... The marks
made by the brush are interesting and vary somewhat according to the
dryness/wetness of the brush and how deeply the brush is pushed into the
paper... Working on the 4th tone (started painting with the lightest) this
morning...
The previous project (also pictured on the web-cam page), was coded on
Sunday and completed last Monday -- it was an entirely different approach
-- in that one, I used a black ball point pen to actually 'draw' each gray
contour of a different image, spiraling in from the perimeter of each gray
contour .05" away from previous lines until complete (exactly like
'routing' one of my recent woodblocks), and then moving to the next area of
the same gray level and so on -- beginning with the darkest gray areas of
the image and repeating by contouring each lighter area (plus all the
darker areas) until complete... There are 14 grays in the image, and the
differences in the shapes of the contours were sufficient to produce that
spider-webby sort of moire line drawing...
In my next drawing (underway as my machine works on the current painting)
I've converted the same gray-scale image (now 660 x 220 pixels with 256
levels of gray) into an excel spreadsheet (660 rows x 220 columns) so that
each cell holds the gray value (0-255)... Then, as a first step, I
produced a second spread sheet in which each cell contains the actual
instruction to the machine in order to control the movement of my router
with a V-bit tool (a sharp point at the end and then "V"s out to make a 90
degree cone)... Each row or column represents a fraction of an inch of
movement of the tool in X or Y, and the gray value determines the depth of
the router, so a typical 'instruction' cell looks like "M3,33,10.8,0.083"
which in English tells the machine to Move in 3 axis from it's current
position to X=33 inches, Y=10.8 inches, and Z (vertical) 0.083"
deep. Then, by ordering the instructions so that the tool moves from one
cell to an adjacent cell and to the next adjacent cell, I should accurately
produce a 'white line' block for a single block print (the line routed
could as easily be horizontal or vertical rasters, or diagonals or spiral,
or spirals or whatever!) which (in theory) ought to accurately reproduce
the tones of the original grayscale bitmap.
Then, the next project (underway simultaneously, but more complex) is to
use the pen (which produces a constant line of a certain width) is to vary
the line spacing in order to reproduce the grays of the original... So
each cell of THAT spreadsheet holds the information about how far apart
this line should be from the last -- this is much more complex because the
machine instructions no longer have a one-to-one correspondence with the
bitmap -- for example, a black cell might require 10 adjacent lines to pass
through that area of the image while a white cell only requires one, so the
machine instructions must be generated 'on the fly' depending on where the
pen is and where it needs to go, reading distances from the appropriate
image cells as we move -- and the direction of the pen becomes important
since very marked peninsulas and islands may result which must be properly
drawn... VERY interesting problem, and I imagine that the drawing which
falls out of this highly analytical process will be almost as interesting
as the process itself!
Anyway, that's what I'm working on this week and probably a couple more
weeks as well...
-- Mike
Mike Lyon
Kansas City, Missouri
http://mlyon.com