"Blink" - Prototyping a Blinking Eyeball Sculpture

As an artist working with various forms of technology, I often find myself imagining creations just out of my reach technologically. My sculpture Blink is a perfect example. I first thought of this piece several years ago but various problems – like how to make working eyelids – forced this project into mental storage. My recent residency at Autodesk’s Pier 9 and their facility packed with a dizzying collection of tools, equipment, and experienced shop staff presented a unique opportunity to make this piece a reality.

This instructable describes my experience designing and building three prototype sculptures for a planned installation of 20 blinking eyeball sculptures.

Step 1: Initial Design

I typically begin my designs using old fashioned pencil and paper. It’s faster and it forces me to think more generally about the sculpture and how someone will interact with it in its final form. I’d like the eyes to remain human scale – about an inch in diameter. I’d also like the option to mount the piece directly into a wall. This demands the mechanism not protrude more than 3 ½ inches into the wall cavity. Squeezing the mechanism for the eyeball and eyelids into such a small space is a challenge for the next phase.

Step 2: CAD

Computer Aided Design (CAD) is where everything gets painfully specific. I keep a bill of materials listing every part, both fabricated and purchased. This way, I have a realistic idea of how much it’s all going to cost. It also helps me decide where to begin work. Some things, making a circuit board for example, need to be sent out for fabrication. It’s better to tackle these things early so there’s plenty of time for everything to come together.

What actuators will drive the eye and the eyelids? What’s the desired speed of eye movement (sccade), how fast should the machine blink, what’s the range of motion up/down and left/right for the eye? All critical questions that will dramatically impact the final piece. A slow and methodical design saves me time in the long run. Remaking stuff sucks.

Step 3: Building the Frame

All these virtual parts need to become real world objects at some point. Long term, I’d like to create an installation with twenty of these eyeballs and I’d like the machined parts to be anodized aluminum. I expect I’ll eventually run these aluminum parts on a CNC mill but I’d like to make sure I’m completely satisfied with my design before I start spewing metal chips. I’ve found I can rapidly prototype a sculpture like this in laser cut acrylic. I export profiles for each piece and cut them on a laser cutter out of acrylic of appropriate thickness. The laser can only cut features on in the plane of the material, so any features on the edges – typically holes – get machined later on a vertical mill.

Step 4: Electrical

Like I said, it takes a while to have a circuit board made. I like to tackle this early so there’s plenty of time to get parts made. I do my initial electrical design layout in Eagle CAD but, before I send out the board for fabrication, I like to build it on a bread board so I can verify everything is working as expected. It’s easy to move a pin on a breadboard but having a new board fabricated is expensive, time consuming, and generally annoying.

Once the boards arrive I get busy soldering all those little surface mount components. I see a lot of people using home made reflow ovens online. Seems interesting but, honestly, I've found soldering those tiny parts with a regular soldering iron fairly easy.

Step 5: Gears

How to move the eyeball around? The plan is to track faces using Open CV but that means I need to accurately position the eye. Animatronic puppets usually use push rods. That setup works well enough if you’re controlling the eye manually with a joystick, but I’d like to drive the eye position with a program. Setting up twenty sculptures so all the linkages are the exact same length sounds like a pain. Gears to the rescue! I’ve made a couple other sculptures using gears and servos but I’ve always been limited to things I could purchase from a vendor like Servo City or McMaster. I scheduled some quality time with a laser cutter and developed a way to make my own gears – or anything else you might want to mount to the spline of a servo. I thought it was interesting enough to create a separate instructable: Laser Cutting Gears for a Hitec Servo

Step 6: Pivot Pins

Those eyelids are going to need to pivot on something. In this iteration, the hole for the pivot pin is very long and there’s not much of the shaft to grab onto when you need to pull it out. There’s not a good reason for this so it’ll change it for my next round of machines. It works well enough for these prototypes so I’ll leave it for now. Machining a little knurled cap on the end of the shaft makes it easy to twist out the pin.

Step 7: Eyeballs & Eyelids

I’ve literally though about how to make eyelids for a couple of years. They need to be relatively thin but they need to hold their shape also. Thick eyelids look awkward but thin eyelids that distort will rub the eyeball. Neither works. I went as far as machining a hardened steel die so I could try stamping them out of thin brass. The thought of stamping and soldering 20 sets of eyelids soon ended that experiment.

3D printing opens a whole new world of possibilities. I can print thin delicate eyelids that hold their shape and the relative ease of printing (versus stamping and soldering, or casting) let me work through a couple of design iterations quickly. I’m honestly not sure how else I could have made these parts.

Still, a couple of challenges cropped up: First, the upper and lower eyelids need to pivot on the same axis so somewhere the upper and lower eyelids need to rub against each other. My solution was to machine small brass bushings that slide over the axis for the upper eyelid. That gives me a brass on ABS running surface that has a chance of functioning well.

Another problem: the ABS material I’m printing only comes in white but I need black eyelids. I was happy to discover the parts take black dye very well. A little graphite to lube the joints and the machine starts blinking.

Step 8: Vacuum Forming

I’ve spent some time around vacuum forming equipment but I’ve never had a reason to try this process myself – until now. In this sculpture, I’d like to conceal the mechanism as much as possible so attentions is focused on the eye rather than the mechanism making it move. (Contrast this sculpture with my previous piece, Look, where the mechanism is entirely exposed.)

There’s not much to vacuum forming in concept. You make the plastic soft with heat and suck it down over a mold. There’s a good overview of the process here.

I was excited to discover there’s an option to automatically create a mold from parts in Autodesk Inventor. I had no idea it was even an option – one of the perks of an Autodesk residency. This let me easily and quickly develop a form that would fit perfectly over my front plate with just enough clearance. Once that was done, I turned to machining.

Of course, there are demons in every detail:

What material for the mold? MDF was suggested and readily available so I generated some g-code and started up the Shopbot. It worked well enough but the surface finish was marginal and, since I’d like to eventually build twenty or more of these, I was a little concerned about the longevity of the mold. Forming a few pieces reinforced these concerns so I dug around and found some scrap acrylic. Machining acrylic is more than the Shopbot can handle. Thankfully, the Autodesk residency gives me access to a Haas Milling machine. Same g-code but a beautiful surface finish. Next time I’ll machine the mold from aluminum.

What kind of plastic (ABS, PETG, Styrene, …) and how thick? I started with 0.02" styrene but it was too thin and easy to damage (like the lid for a disposable coffee cup). Thin PETG (0.03") worked better but still felt too floppy. Thicker PETG (0.06") felt much more solid but didn’t take detail as well. Eventually I landed on 0.06 thick ABS. It was the best combination of properties – relatively easy to form, held detail well, and machined nicely. Once I had a few ABS parts formed, I laser cut jigs to trim the edges cleanly and locate all the required holes. Mission accomplished.

Step 9: Painting Eyeballs

I used the 3D printer to create basic eyeballs but the sculpture doesn’t really come together until the eyes are painted.

These 3D prints are the nicest I’ve ever worked with but the surface is still surprisingly rough. Once I get all the support media cleaned off the surface they get numerous coats of primer: Prime, sand, repeat. When the surface is smooth they get a final glossy white top coat.

The iris then gets a base coat. I want each eye unique so the color varies slightly for each eye. Details are gradually built up on the surface of the iris and the pupil get a coat of black.

More layers of paint and resin are gradually built up. This gives the pupil a subtle depth that really comes through in the final piece.

Blood vessels (AKA red thread) are added and, finally, the whole mess is coated with another thin coat of polyester resin. This last coat gives the eye a slightly amber cast and helps pull everything together.

Step 10: Temporary Wall

I’ve got my sculptures blinking but I need a place to temporarily display them so I can convey how they’ll look mounted on the wall - something easily transported. A standard sheet of 4’ X 8’ plywood tilted upright makes a nice proxy wall and is easy to build. A few gussets hold it upright and a laser cut router template makes it easy to machine cutouts for the eye mechanisms.