Giant 3D-printed LED sphere with 2,200 addressable pixels

Oct 25, 2026 Last reply: 15 hours ago 4 Replies
Jump to replies (4)

When it is running, this sphere puts on a proper show. Animated multicolour sequences roll across the surface with rainbow gradients, sweeping runs and pixel waves, then swirling vertical bands of red, purple and blue take over before the display goes dark. The clip finishes with every LED at full brightness in solid cool white, producing a strong lens flare. The sphere stands on a vertical stem over a rectangular base, and with the power off, the white curved slats around the inner frame give it a clean look.



The construction is the part that earns my respect. The sphere measures 530 mm across and is assembled from more than 240 individual 3D-printed parts. Curved vertical ribs carry outward-facing strip segments, arranged as 10 rows of 42, 10 rows of 40, 20 rows of 34 and 40 rows of 18 LEDs. That totals roughly 2,200 WS2815B pixels, 12 V, IP65, at 60 per meter, fed by a Mean Well HLG-600H-12A delivering 12 V at 40 A. A row plan like that takes real patience to map out.



Two details show this is a build in motion. Only 3/8 of the final sphere exists so far, and diffusion is currently being tested with painter's tape, due to be replaced with satin-finish acrylic later. There is also a silver-toned enclosure at the base with wiring routed exposed up the support shaft; that part still looks like bench-test territory to me.



The wider plan is what makes it ambitious. The sphere will anchor a gyroscopic frame with three spinning rings, and eight sensors should detect human interaction, shaping brightness, colours and animations in real time. A Raspberry Pi will run the whole system, including ring-drive motors and sensor inputs, with WLED handling synchronisation across sphere and rings. The maker is openly asking how to connect the Pi to WLED and which protocols and software setup would work best, and that architecture question feels like the real challenge ahead.


Which supply is powering the sphere during these tests? With around 2,200 addressable pixels on board I am curious how close to its limit the unit actually runs once everything is driven at full white.

Running 12 V strip instead of 5 V across roughly 2,200 pixels seems like a smart call to me. At that density and diameter, voltage drop along each rib would be harsh on 5 V, and the IP65 rating looks sensible for parts that people will eventually touch in an interactive installation. The trade-off is slightly coarser colour depth per pixel, but at this scale I doubt anyone would spot it.

Before committing to the satin acrylic across the whole surface, I would fit just one rib with it and run an identical sequence beside the taped sections. Comparing diffusion directly on the real curvature should show fast whether the acrylic spreads light evenly between rows at this spacing.

The maker has listed the supply as a Mean Well HLG-600H-12A, 12 V at 40 A. Nothing has been said about how much headroom that leaves at full white, or about the reasoning behind picking it.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required