How to Make [almost] Anything

Nathan Melenbrink

Something Big - CNC Painting

I've used CNC milling for machining plywood quite a bit in the past, so I opted to use this week's assignment to put the machine to an alternative use. This is also a step in the direction of my final project proposal. Rather than using the mill to cut material, I replaced the bit with a paint brush. This required a customized toolpath to tell the machine to return to a color swatch off of the canvas and dip the paint brush in fresh paint after each stroke. Due to constraints on the machine time, I was only able to make two studies, neither of which passed my aesthetics test (I'd rather have a blank canvas on my wall), but nevertheless was a useful exercise in a somewhat obscure digital/physical interface. My only real disappointment was that it failed to produce an aesthetic that was unmistakably machined...which was actually the whole point of the exercise.

  • Tools: ShopBot, paint brushes
  • Materials: Acrylic paint, paper, canvas
  • Software: Grasshopper, Rhino, ShopBot software
  • Year: 2014
Project 01a
I used Grasshopper to build a definition to simulate a vector field based on attractor points. The curves that interpolate the points effectively become the tool paths for the machine to follow.
Project 01b
These tool paths were then converted to code in the .sbp format, which is something like gcode but even more intuitive. This was a nice change from working with the Makerbot software, which seems to be designed to discourage tampering. The shopbot (sbp) format is so straightforward that it seems it's replacing gcode in a lot of applications.
Project 01b
The first study was with a 1" paint brush and two different shades of grey paint, placed in coffee lids outside the painting area.
Project 01b
The brush strokes came out better in this first study than the subsequent one, largely due to the fact that this was painting on paper on foamboard, whereas the second was painted on canvas, which bends more easily under pressure.
Project 01b
Completed first study.
Project 01b
In the first study, it was unfortunate how much time was wasted jogging back and forth between the canvas and the paint containters. In the next iteration, I dropped paint on three specified attractor points, which were measured out and marked on the canvas. Each stroke gets its closest color, with a degree of built in randomization so as not to be too obvious. The travel time was then greatly reduced.
Project 01b
Project 01b
Project 01b
As always, I had to wrap things up due to time constraints, feeling as if I'd just scratched the surface. There are a massive number of possibilities to explore in this direction. Things I'd like to try in the future include:
Varying the brush size, as well as speed and pressure on the canvas. In order to more closely simulate the art form, the brush stroke should increase with pressure over time. This would actually by very easy to accomplish.
Move beyond replicating the way in which humans use the medium to something which is unmistakably machined. This could be either in the quantity (complexity) of brush strokes, or in their quality. For example, I can imagine activating the spindle to rotate the brush while simulataneously translating in 3D...a movement that is impossible for the human body to realize.
Creating a kind of pump or drip for paint to continuously be discharged, eliminating the need to return to swatches of paint. Additionally, there were issues with the brush getting clogged over time, which made me thing a foam brush would have been better.