# Holy. Shit. https://www.youtube.com/watch?v=w8Fl8L4yk8M These guys came over to E3D HQ the other day,

Holy. Shit.

These guys came over to E3D HQ the other day, exciting stuff all round.

omg o_O

Impressive. Having been a 5 Axis CNC programmer for many years, it brings with it a new set of interesting challenges.

god damn it

super awesome though. Good bye supports, just sad they beat me.

Building a 5-axis machine is easy, it’s building 5-axis CAM/slicer software that’s hard. There’s probably a couple dozen comp sci PhD theses worth of work in making a generalized 5-axis 3DP slicer. When you add the fourth and fifth axes, the slicer has to start making optimization decisions instead of following a simple deterministic algorithm. That makes today’s hobbyist-level slicers look primitive in comparison. It’s a good challenge – the five-axis machine control software that does exist today is primarily subtractive machining, which has different CAM constraints (like whether the tool can cross its own previous paths or not.).

Stuff that’s easy for a human to do – like recognize the surface of a sphere and say the machine should keep the nozzle normal to the surface – is really hard for a computer to do. Likewise for collision prevention, strength optimization, bridge optimization, overhang optimization, extrusion strand geometry optimization, and so on. So today all we really see in this space is awesome custom-coded demos. I’m really looking forward to the development of some good open-source general 5-axis tools.

Its such an interesting problem though. Something I’ve really wanted to dive into. But your totally correct.

This adds a whole new level of complexity to the process.

Wow!!!

@Sanjay_Mortimer1 what’s their slicer like?

Printing on the surface of a sphere is a 5 minute Python script. Is that all they’re doing, or are they further along?

I’ll probably reach out and see if I can help with Bread.

Fun times! CAD with user define surface normals? STL’s seems like a poor choice for file format, and when does G-Code run out in terms of being able to handle some decent 5-axis operations?

I guess 5 axis slicing can be generalized as a Reversed Set of Eroded Shells, constrained by Ambient Occlusion and an inner Straight Skeleton that “hardens” unreachable areas and connecting structures to be eroded with some delay (printed earlier) .

nice the final result will be less layer dependent as they crisscross. ideal also for wire print

A path on a shell might be expressed as TSP solution on some selected density of points on the surface optimized for least per axis acceleration.

@Torleif_Ceder yeah, that’s what I was thinking too, but I didn’t want to jump to the obvious solution.

No, just kidding - I am counting myself smart just to get your concept, though not familiar with the specific terms. Please don’t tell me you made it all up, because if you did, you should be a sci-fi writer!

i’m impressed … what is the status of your nonplanar-slicer and the firmware for this 5axis-printer?