Dragonfly 4.0

By Joshua Killen 3/9/2018


Two flickering wings

Frame head down to a long tail

Dragonfly on high


Dragonfly 4.0 was a project that increased my appreciation of the amount of work it takes to design and engineer effectively. It started as a small-scale project; the objective, to create a dragonfly with wings that could move. As the first prototypes became assembled significant issues with my design and manufacturing methods became immediately apparent. Before I get into any details, the dragonfly model pictured here is the 4th prototype and first model to be completely assembled and is now a gift to the VIP who originally suggested it’s creation.


Butterfly Scope Creep

Last year I did a small project where I made these little butterfly pins for the Rainbow Fun Walk. Link to article. These creations had posable wings that were 3D printed using a flexible filament. It was suggested I make a dragonfly with wings.

This round I wanted to go bigger and challenge myself to do better and innovate; consequently, this strategy uncovered new and unintentional consequences stemming from my more ambitious design choices. It was a stark reminder of a term I became aware of in my previous professional life called ‘scope creep’. For the uninitiated, scope creep can be explained like this. Let’s say you are building a wooden box for the next hour, and somewhere during this build your boss (or crazy imagination in my current situation) says “No. You are building a spaceship”. In my old profession, the timeline would be the same and I’d still only have materials to make a wooded box, which I would craft into the best wooden spaceship I could. I digress; these days I do art spaceships.


Making Realistic Wings and the Transition to VR Sculpting

The butterfly wings were originally designed from my hand drawn images, for this project I wanted to use images of real dragonfly wings and transition them into a 3D printable model. This I accomplished using Illustrator to trace and export out a clean SVG file. This file is imported into the 3D modelling software for extrusion. This is far from an automatic process, it takes time to find and clean some of the odd vector points and angles.

In the original project scope, I would design and export the model in Fusion 360; however, because the wings had increased their detail of the designs, I felt having geometrically simple body forms wouldn’t look right. This is marks the time of Dragonfly 1.0, which I felt wasn’t worth showing.

Around this time, I was getting into VR modeling and my disdain of the simplistic pushed me in an entirely new direction. I made a model for printing completely in VR which became Dragonfly 2.0. In this version all the parts and joints were sculpted in VR, then exported into a printable file. All the joints were ball and socket joints, and they were made from prebuilt templates embedded in the software. I decided to print the entire model in Nylon on the MarkForged as the Makerbot was busy making snowflakes at the time.

Cleaning the printed parts was challenging. All the pieces were very organic and required multiple printing supports. Nylon material is a lot tougher than PLA and so it makes the cleaning even more difficult. During this stage I was finding flaws with my design. A couple of the socket joints had a little distortion due to my sloppy modeling in VR and those required a little more cleaning with a Dremmel.  I did a preliminary assembly and the ball and socket joints were completely inadequate for this to be practical design. Back to the drawing board with new strategies to incorporate:

  • The joints should be printed vertically and separate from the parts they are going to eventually join. This adds a step in the post process; however, the joints will print clean and be stronger.
  • Use VR to create organic and complicated shapes. Then bring them into a program like Fusion 360 for final assembly. The tools offer precise controls over the placement of joints and adjoining pieces that the VR programs are currently missing.

The Composite Approach

The next attempt was my first blend of VR and 2D sculpting. This allowed me to freely sculpt the dragonfly body in VR which I prefer. I exported my original dragonfly wings into the VR sculpting software Medium the model and the model became a lot cooler. Here I explored the limits of resolution threshold these VR modeling programs can handle, especially when working with the wings which account for an incredible amount of vector line detail and memory usage. Dragonfly 3.0 was printed.

During Dragonfly 3.0’s preliminary assembly the nylon joints were doing well, plus the sockets had a fair amount of friction so the figure had the ability to be posed. That all changed after I did the finishing work on the body parts. The weight of the XTC coating and the paint added so much weight the model could barely hold together let alone stand up on its own. I tried to salvage it using metal wire to support the joints, but it failed. This wasn’t going to work so the design had to be scrapped.

Reflections and the Success from Simplicity

I was looking into incorporating a rubber band element to provide tension between the pieces like the old GI Joe figures from the 80’s for my next attempt. Then I had a moment and reflected on how far from the original scope this project had creeped. My 4th attempt would be a simplified approach yet still incorporate VR and 3D sculpting elements. I simplified the design of the body and added a base. The wing joints would be like the butterfly joints, but the design of the wing would be the detailed realistic one.

The body and base all printed as one, the wings would be the remaining challenge. The flexible filament does not do well with exquisite detail, in fact its pretty terrible at it. I needed the joint to be flexible, so I split the wing into parts that would have to be printed separately with different materials. I eventually got it to work but the gluing process felt awkward. I know there is a better way to discover.

Post processing went smoothly, and the painting of the model was fun with the airbrush. I did some hand brushed details here and there and some dry brushing. I enjoyed it so much, after I finished and shipped 4.0, I pulled out 3.0 and just glued all the parts together and painted him up too.

Next Steps

For my next set of figures, I feel more prepared. I am going to emulate the action figure base framework into the design. Just must figure out the direction I want to go, if I want to revisit old ideas or jump into something new. Maybe as bit of both, just as long as I can keep the scope creep down to a tolerable level.