Post must be at least 35 characters
1 Like
Nice summary. I would look out for further sources of displacements on top of linear and constant errors. E.g., in my experience, concave walls tend to be printed slightly to small. The smaller the radius, the stronger the effect. E.g. a hole with 20mm inner diameter comes out as 19,8mm. A hole of 4mm comes out as 3,7mm. A 1,5mm hole becomes 1mm. (Numbers just made up).
Nice to see yet another take on a problem only fixed for larger scale industrial printers so far…
I doing this 3D printing stuff now for more years than I like to count.
And if you started with the early Marlin and such firmwares on 8-bit controllers than you really have seen it all several times already - NO OFFENCE !
What I am trying to say is that both firmwares and slicers have come a long way.
Where one struggles the other provides a workaround.
The problem as I see it seems to be that no one really cares about physics any more 
Print something in vase mode, with 1 ,2, 3 and 5 wall loops and measure the results - no real consistency.
Same for holes, especially small holes that are none.
A hole is more like a wall for the printer and slicer.
Just that the infill is on the other side.
We don’t like lines, we want a proper arc so prevent corners…
All this and much more is addressed on multiple levels through the path from slicer to printed model.
We are stuck now for quite a few years with no real breakthrough in the firmware department.
Fundamentally changing pathway and extrusion parameters affects all parts of a print and would require a lot of beta testing.
Bambu prefers to import those things from Orca and other sources so I really think such nice calculations and tests should be posted there as you WILL get much for feedback from those developing than here.
I have experienced that inward shrinking on my venerable Ultimaker 2. Typically I printed at 30mm/s and input shaping was far away at that time. The X1C does it in a very similar way so I’m sure that it doesn’t come from input shaping or other dynamic variables but instead is a static phenomenon of FDM printing.
Regarding prior investigations: Stefan from CNC kitchen did a very good video on calibration cubes and why they are a terrible idea to use. There, he also explains how to avoid constant errors when calibrating filament shrinking. Vector3d created the Califlower model which takes the constant errors out of the equation by providing inside and outside measurements that are to be averaged. Interestingly all the slicers provide a tool to compensate constant errors, e.g. in bambulab via XY contour offset (dont remember the exact name). But except for Orca slicer, none allow to correct for filament shrinking, which in my opinion is the most relevant source for incorrect dimensions.