Building a Voxel-based z-first Slicer

[LudovicUrbain 0]

Hello, I've been using Cura for a few months now, and I believe there might be a different approach to slicing that might yield better results.

 

I've looked a bit at the source code for Cura, and it isn't clear yet where I could work on a fork of the github repo to try my idea out.

 

Here's how I understand the current slicer:

 

Currently, slicing occurs at a per layer level, where only infill can be printed every n layers.

This causes very important detail, deserving of the 0.02mm layer height, to be printed at the right layer height, and large areas that will end up inside the model to also be printed at that layer height, including infill.

We end up with insanely long print times because 97% of what's being printed at 0.02mm is just support for the next layer, and not actually visible 0.02mm detail.

This also occurs with adaptive layer height, which incidentally has big problems with thinner layers, pillowing, top layer count vs thickness etc. - problems which might be related with an initial per layer slicing.

 

My idea:

 

• start with nozzle size, e.g. 0.4mm and a min layer height, e.g. 0.02mm
• derive a max layer height from the nozzle size, e.g. 0.28mm
• create a 3D array of 0.02mm voxels that represents the STL mesh
• for each fat layer (max layer height), for each x,y, find the z-height of each column of voxels
• put each voxel in the print layer that matches its height
• from lowest z-height to highest (0.28 or top of print), run a pathing algorithm to optimize each print layer
• print each layer

 

It doesn't really handle overhangs at this point, but other than that, I expect it could print at full quality in the same time as minimum quality for non-overhang prints.

 

That said, overhangs could still be handled differently at this point, they could perhaps use the same logic reversed (i.e. columns hanging from the ceiling) although I have no idea if that would even work physically, having not tried slicing at this point.

 

A few questions then:

1. Why wouldn't it work?
2. Will you try it?
3. If I want to try it, how should I proceed?

 

Thanks :)

 

 

 

[LudovicUrbain 0]

Something like this: