Aspen Grove (8) - Mike's Process ...
Posted by Dave Bull at 8:54 AM, September 23, 2006
Apologies to those who have been following this thread - trying to follow this thread! - in frustration. We've got the train back on track now ...
After the previous set of blocks turned out to be unprintable, we decided to put them aside and cut another set, this time using cherry-faced plywood from Woodlike Matsumura here in Tokyo. This is the wood that I use for most of my own printmaking work; well-dried 5mm sheets of cherry, bonded under pressure to a heavy plywood core, and then surfaced with planing and then under-water sanding to a very high polish.
Mike was willing to use these, but requested that the final sanding step be omitted, as he wanted to use his own carving machine to cut a guaranteed level surface on the wood before cutting the designs.
So I ordered the wood ... It arrived in late August, and I sent them over to Mike (I don't even want to think about how much the final bill for air freight is going to be for all these shipments back and forth!)
So let's pick up the thread with emails ...
From: Mike | Date: September 21, 2006
Aspen Grove block 1 is carved and side two (block 2) is carving now – at this rate, I’ll be done by the end of this week, or maybe early next… Currently calculating toolpaths for 7th block – when it’s done, only one to go! LONG day today, and I’m VERY dirty and dusty! Going home now to shower, eat, and sleep – back at it in the morning to finish #2 and move on to begin carving #3…
(Dave replied during a Skype conversation ... )
From: Mike | Date: September 22, 2006
I’ve completed carving of blocks 1 through 3 and am currently surfacing block 4 (2nd side of 2nd block). I’ve been planing off about 15 thousandths of an inch from the surface in order to make the blocks flat and parallel (they seem to be out by about 1/100th inch as they come from Matsumura – that means the faces are non-parallel by that amount)… Then I sand smooth to remove machining marks with 220 grit and finish with 400 grit to a dull shine.
I’m using the same bitmap contours as with block set #3 which are the ones I sent you last spring – those bitmaps look very good to me still. But I’ve vectorized them with more precision than before and I’ve redone all the toolpathing using a much closer tolerance (now about 1/1000th inch tolerance) in order to try to carve more of the fine detail than before – this means it took about twice as long (24 hours) to calculate the toolpaths, and is taking about three times longer to machine, but from the way the first three blocks look, it’s worth it – these appear to me to be a substantial improvement over the previous set (at least according to my memory)!
The 45 degree vbit is (so far) providing much greater precision than the 60 or 90 degree vbits I’ve used in other carving. This is like… DUH! Each axis of my machine ‘moves’ in increments of about .002 inch. I previously carved the blocks about .07 inches deep (35 ‘clicks’ vertically) using a 90 degree vbit on the first (plywood) blocks, so the narrowest mark I could theoretically make should be about .004 inches and at .07 inches deep the diameter would be .14. With a 60 degree vbit, the narrowest mark I can make would be about .0023 inches and at .7 inches deep, the diameter would be about .08 – this is what I used for on the maple and cherry sets – you have the cherry set carved this way.
This week I’m carving the Woodlike Matsumura blocks .1 inches deep (50 ‘clicks’ vertical instead of 35 at .07) and I’m using a 45 degree vbit. So the narrowest mark I can make is theoretically about .0017 inch and at the full .1 inch depth the cut is only .083 inches in diameter, so it’s clear why these are turning out better all the way around – about 5 times greater precision, assuming, of course, that the steeper tiny pyramids survive machining and that I accurately zero the vertical axis exactly at the surface of the block…
4.5 million lines of code to carve these blocks and about 3500 feet of actual carving, so I still have a LOT of hours of machining ahead of me at this point – might be finished closer to the END of next week it now appears.
From: Dave | Date: September 23, 2006
... but from the way the first three blocks look, it’s worth it – these appear to me to be a substantial improvement over the previous set (at least according to my memory)!
OK ... hope you keep the bit sharp! That steeper angle might mean more mountain tops getting chipped off ...
You said the other day:
I’m VERY dirty and dusty!
How much actual 'hands on' do you have to do with all this? And how much input does the toolpathing process need from you (decisions, etc.)? People have been curious where the 'line' between you and the machine is falling ... how much is 'automatic', and how much is 'artistic' ...
From: Mike | Date: September 24, 2006
The 'line' between me and the machine falls mainly at the tips of my fingers, I suppose. I've described this before, I think, so you already know all this -- but again, briefly:
I photographed the Aspen Grove about 50 yards from my Mom and Dad's home in Aspen, Colorado. I manipulated the image using PhotoShop, both rotating and distorting it in order to make a 10x15 inch (oban) image which I liked. Then I scaled it up to 600 dpi and began adjusting the tonal curve until the image contained roughly equal areas of tone while maintaining an overall tonality which appealed to me. Then I began to enormously simplify the image. First I converted it from color to grayscale, then eliminated areas too small to carve while still maintaining sharpness and detail (something of a trick actually). After many hours of back and forth adjustment and simplification, I had only nine values in the image (white, and eight evenly stepped grays going to black), each covering roughly the same area (square inches) of the image and the smallest piece of each tone was only a bit smaller than I might actually be able to carve. I added kento registration and masked borders for carving so the bitmapped image became a bit larger than the blocks I would carve. Then I built the contours to be carved for each block (I chose this particular image in part because I could avoid having to include paper support areas which could complicate instructions to the printer and make it easier for mistakes to occur in printing).
I output the contours for each block to a 1-bit bitmap (black and white) and vectorized the perimeter of each black contour -- in effect, each area to be carved away was outlined with vectors. At 600 dpi and using a 1.2 pixel 'average' for the vector conversion, and eliminating any contours of less than 6 pixels (.01 inch), the outline of each area to be carved was contained in a dxf file. Then I loaded the dxf vector information into my toolpathing software, defined the router bits I would use during machining (their geometry, including shape, diameter, inches per second, offset for clearing strategy, and similar information) in this instance, a 45 degree V-bit and a 1/8 inch upspiral end mill. Then I computed the toolpatsh -- using the v-bit to engrave the outline of each contour (calculating the Z-axis height of the v-bit in order to accomplish very narrow lines) and used the v-bit to clear interior areas too narrow for the eighth inch end-mill to reach. Larger areas were toolpathed for clearing with the eighth inch end-mill. Then I saved the toolpath instructions as text files in a format my CNC machine can interpret. Although the toolpath files contain some instructions to set the speed of tool movement and other parameters, and many instructions to lift the bit out of the block and 'jog' to a new location without cutting, more than 99 percent of the toolpath file instructions are "move in three dimensions from the current location to x,y,z where x y and z are the measurement in inches from 'zero', a point about six inches above the bed of the machine at a corner. All these millions of lines of instruction are generated automatically from the parameters I enter and using the vectors I built from the original image.
Then I decide how to mount the block onto the bed of the machine. In the case of the current block, I use vacuum. I built a vacuum plenum over my machine bed -- it's a 49x96x1 inch sheet of light weight MDF and I machined away deep grooves from the bottom side so it is 'waffled' and I screwed it down to the bed of my machine and caulked the sides all around so they are air-tight. I have a Fein shop-vac attached under the bed to provide about 1 pound per square inch of suction. I cut 10 3/4 x 16 inch (the size of the blocks you provided me) hole out of a sheet of 4x8x1/4" corrugated plastic sheet, mounted the sheet onto the vacuum plenum and used masking tape to seal the edges. Then I placed a block into the hole I'd cut in the plastic, turned on the vacuum, and taped around the edges of the block onto the plastic to effect a tight seal. So the block is held down by about 172 pounds of air -- enough to keep it from moving around during machining.
Then, since the two cherry-faced sides of the block were not completely parallel or flat when they arrived, I used a 1/4 inch end-mill to plane about 15 hundredths of an inch from the surface of the block (I do this with a 'cut rectangle' command to my machine giving parameters for how deep to cut, how long and wide, the step size (tenth of an inch for these) and that it's to begin at the outside and move in until the whole surface has been machined.
Then I sand the freshly machined surface (already quite smooth, actually), first with 220 grit and then with 400 grit. The whole setting the block, zeroing the tool, machining flat, and sanding process takes about 30 minutes per side.
Then I install a 45 degree vbit into the router, zero the vbit to the surface of the block, check the zero carefully by machining small lines at the edges until I'm satisfied with the accuracy of the setup, load the toolpath instruction file and run it. I watch for a few minutes to make sure all appears to be 'right', then I run upstairs (and write this email) and do other stuff, usually checking on progress every hour or so until the carving has completed (for these blocks probably an average of 7 or 8 hours of v-bit carving). When the v-bit movement has been completed, I lightly sand the surface with 400 grit to clear any burrs. Then I remove the v-bit and install an eighth inch end-mill, zero it to the surface of the block (doesn't require as much precision as it's only clearing and cutting kento), then load and run the toolpath for that tool and block. On these blocks that's only 30 minutes to an hour to complete.
Then I blow the dust and chips from the carved surface, turn off the vacuum, remove the tape and either flip the block to begin the other side or (if this IS the other side) put in a new block and repeat until done.
As I think I've mentioned before -- the ink and paper have NO idea and receive NO information about what sort of tool might have removed the portions of the block which don't print, so there's really NO way to determine by looking at the print that the block was carved by router, by hand, by fingernail, teeth, or whatever! The only evidence of the block is the ink transferred from the UNCARVED surfaces, right?
OK, so where's the 'line' between me and the machine? I don't really believe there IS any such line. The machine is an extension of myself in exactly the way ANY tool is an extension of oneself. Mind goes, everything else follows, right?
Minor disaster yesterday... My router shot craps mid-block-4... Picked up another (and dropped $350) this morning and am now re-carving #4 which should turn out just fine.
This thread continues in Aspen Grove (9) ...
Added by: Steve Knoblock on January 12, 2008 8:08 AM
I stumbled upon the Aspen Grove print while browsing the Mokuhankan catalog a few days ago. It had never occurred to me to use a CNC machine to mill the wood blocks for a print. It is a fascinating idea and the results look beautiful. I can envision some of my photographs being good candidates for a wood block print.
It would be a new way of printing a photograph, different from anything else, whether a gelatin silver print on paper or a photoceramic. Having trudged my way through the saga of making the print, it appears to be a daunting task, so perhaps I should curb my enthusiasm. Still, I would love to have some blocks cut from my images and try my hand at printing them. I am just too impatient to learn carving.
I can understand the problem of interpreting a pixel based image in a way the milling machine can follow. I read your explanation of the lengthy process involved in readying the image for cutting and wondered what software tools you use. Have you ever seen VectorMagic? http://vectormagic.stanford.edu/ Would this be a useful tool?
So many things to learn.
Add Your Input