CNC Kitchen measures Z anti-aliasing cutting FDM surface roughness from 80 to 25 micrometres
CNC Kitchen measured Ra dropping from around 80 µm to around 25 µm on a 5° test wedge printed with “Z anti-aliasing” (ZAA), without any special hardware (8:31-9:00). The gain shrinks as slope angle increases: both methods converge to roughly 38 µm Ra at 25°.
The method runs through BambuStudio-ZAA, a fork of Bambu Studio by Alexander Dobkin (“adob”), or as a post-processing script called GCodeZAA by TheanInova, applied on top of regular Bambu Studio or OrcaSlicer output. CNC Kitchen printed tests on a Voron 2.4 with a Stealthburner toolhead, in PLA, measuring results with a Mitutoyo Surftest SJ-220 (2 µm stylus tip radius) and verifying print dimensions with Mitutoyo calipers.
| Test | Conventional Ra | ZAA Ra |
|---|---|---|
| 5° wedge | ~80 µm | ~25 µm |
| Dome, before sanding | 50 µm | 21 µm |
| Dome, after 1st fill/sand pass | ~26 µm | ~10 µm |
| Dome, after 2nd fill/sand pass | ~19 µm | ~6 µm |
Layer height still matters on conventional prints: on shallow 1-9° angles, conventional 0.1 mm layers measure ~15-19 µm Ra, conventional 0.2 mm layers ~22-62 µm, and conventional 0.3 mm layers ~52-91 µm. ZAA closes most of that gap at coarser settings, ZAA at 0.2 mm (~15-19 µm) matches the fine 0.1 mm conventional result, and ZAA at 0.3 mm (~18-23 µm) comes close.
CNC Kitchen also references a 2017 academic paper on the underlying concept and recommends slow 20 mm/s wall and top speeds for the technique, plus specific slicer settings including a 0.05 mm minimum Z height and 35° minimum wall angle. The video’s print-time cost comparison, run on a 300% benchy, cuts off before stating a number, so the time penalty of ZAA versus conventional slicing is untested here.