I had some thoughts about nonplanar printing and found this thread. I would be interested in knowing which of these ideas are already common notions. Thanks!
Lots of people are talking about nonplanar extrusion. Taking this to extremes, make an extruder nozzle that is long and thin. Have several angled in different directions or one whose angle of attack is steerable. Then have a slicer that follows contours of the model: Attempt to make a layer match the normal to the thinnest material laid down so far. Intuitively, if you are making a U, you build up the inside strip by strip. Lots of caveats: Have to account for extruder interference/dexterity, have to build up the uprights of the U over time.
Related to above: Make the fill path of each layer more variable. Do a back-and-forth along X and then the next layer along Y. Or three orientations. Or again, don't make the layer planner in the short dimension either. Make it a hump. So a re-anglable tip runs over it on an arced path. Again, using different infill orientations on each layer.
Stop thinking of layers as planar and think of them as external. You have a current lump of material and you are adding more to the outside. You are looking for the optimal grow shape, both for efficiency and stress. So it is a hump in two dimensions that you build up.
Basic idea is to not provide flat or consistent surfaces of weakness to an type of stress. Think of what extrusion would make a structure most durable, then design an algorithm that will approximate that path. For the standing thin shapes, you have a trade-off between vertical building and standing up against gravity. I would say that you are building up hills with the path running up and down around the outside. The rate at which you gain height is based on what the material can support. The more the material will support, the longer the threads that can be laid down along the center area and so the more resistant to tension. In general, you want stress to pull along extrusion threads, not separate layers of them.
Perhaps the ideal solution would be a constraint minimization of breaking off the extrusion as little as possible and altering the direction of path as little as possible. Ie, long continuous threads.
- Minimize layers by following model contours.
- Squiggly infills vs. dashes in one direction.
- Infills that follow long axis.
- Less planar layers = more deposition per layer.
Finally, to solve the detail problem, note that a sculpture has two parts: the structure and the surface. The structure is there to put the surface in the right place. It has no personality, it just wants to resist deformation. The surface does have some structural properties but they are different. It wants to hold its detail and resist abration. Abration isn't anything like torsion. Nor is holding small scale features.
So following with the idea that you are building up the outside of a model, make the structure and then paint on the detail. This also allows for varying color, texture, softness, sheen, etc. Areas of variability are useful both for artistic pieces and machine parts.
Ideally, detail should not even be done with the same head(s). Different temperature, extrusion rate, material, arm dexterity. The detail head would be much slower but have much more articulation. Maybe even rotate the model? Or maybe you only use as much floor as you need to support the model to allow attack from below. Ie, the slicer tells you to put a stand of a certain size on the build area before you start.
There is much less detail work to be done than structure so it is reasonable to use a slower, more precise instrument.