The Idea That Takes Too Long 

This post is waaay overdue – I should’ve written it ages ago. But hey, better late than never! For a while now, I’ve been thinking about new lighting for our living room. I had this vision in my head – inspired by the lights I saw over 15 years ego at Platige Image post-production studio.

 I really liked the organic structure of it, but I needed something that would work as a ceiling lamp.

Sunflower Math? No Thanks!

At first, I was all about a circular shape – inspired by sunflower seeds and, of course, the golden magic of the Fibonacci sequence. But after a few sketches, I ditched the idea – too regular, too predictable… and it could easily end up looking a bit tacky.

Square, But Make It Interesting

So, I went for a simpler square shape, but to keep things interesting, I gave it a bit of a “pillow” effect for better light distribution. I decided on 100 independent, addressable light points. Their layout was based on the sunflower pattern – more spaced out in the center, denser towards the edges. From this layout, I generated a Voronoi diagram, which became the foundation for my 3D model.

Prototyping & Scaling Up

And that’s when the fun began. Lots of iterations, tweaks, rounding, and adjustments. I 3D-printed the first scaled-down model – and well, it wasn’t anything spectacular, but it gave me a good sense of how the final shape would look. 

I knew the full-size model needed to be CNC-milled. The size? 1200x1200x200 mm – quite a hefty piece!

The Material Hunt

I chose non-flammable polyurethane foam. Sounds great, right? Except… I couldn’t find it in the right format in Germany. The alternative? Gluing several sheets together… meh. Luckily, while in Poland, I popped into the first construction store I found. Of course, they didn’t have the sheets either, but they directed me to a nearby place where I found the perfect ones: 2400x1200x200 mm. Problem? Transport! Quick measurement – if I cut the sheet exactly in half, it would just about fit in my car. So, chop chop!
 

CNC Milling and Endless G-Code

Material secured, now onto the milling. And this is where things got tricky… Working with Delmia was a struggle – I tested various machining strategies, each with its pros and cons, but the biggest headache was generating the G-code. The process took over 6-12 hours and had a nasty habit of crashing midway… which meant starting over. 

Even after breaking it down into smaller tasks, the generated G-code still had around 200,000 lines! So yeah, it was quite the challenge, but after a few days, the final model was ready.

To Be Continued…