The next step for me was to make that continuous walls also rounded in the corners: For this i programmed some bezier curves for the printhead to move along:
The next two models show the basic principle again: There are more curls produced at the lower layers and less at the top layers. With the right model i started already at a lower layer not to elevate the printhead in the middle of the segment any further than the overall elevation and because of that the right model looks more “closed” on top:
As i thought i had produced enough curly looking models for now, i decided to try to produce less curls and see, if this way the wall would become straight again on the top of the model.
For the next model i used way less extrusion than for the models above (about a third) and i also lowered the speed of the movement of the printhead. I got my numbers from that perfectly built up cone model i did before (see: tests 07).
Even if the filament fell down onto the model quite randomly where the printhead was elevated high enough, similar looking structures were evolving:
Due to the density on the rounded corner parts of the wall, i first had to make a grid again, so that the model would stay stuck to the bed of the printer during printing.
After about 120 layers (and more than 45 minutes), some really nice looking structures became visible:
As you can see in the photo below, even if there is random, there is nevertheless an underlaying structure becoming apparent:
Unfortunately with this model the height differences became too big and, as you can see in the photo below, the printhead would burn the model in the corner areas where the angle of the model is almost the same as the angle of the side of the printhead.
There are two possible solutions to that:
a) don’t make a model where the differences in heights are too big in a small area.
b) don’t be lazy like i was and programme the movement better, so that the printhead would not just simply change directions in the corners, but approach them already lowered down a bit more.
I decided in favor of b) and i also changed the bezier curves a little bit, so that the rounded corners would be a bit more open.
With the next model i made less curly layers on the bottom to see, if the model would “close” earlier on top than the previous one, and yes, it did: not yet on top of the big curves, but you can see it on the right side on top of the smaller curves:
Because of a similar burning problem with some earlier wall-models, i made the elevation in the middle of the segments also depending on the length of the segment: That means if the bezier curve is shorter, there is less elevation going on and it will look more “closed” at a lower layer already.
These experiments about “closing on top” were showing me, that exactly the opposite from what happend with that twisted bezier curve model (see: tests 20: some small error on the bottom of the model continued until the top layers without going away) is also possible!
Some curls (“errors”) can actually be covered up again, if there are just enough “straight” layers on top or if the way the wall is built remains slightly curly. But the exact behavior seems to be depending on the factors elevation, speed and extrusion - again.
And finding out the right combination between these three parameters seems to continue to be the really hard but of course also really interesting part! :)