Is there a way to keep surfaces smooth instead of the faceted look when exporting to STL?

Oct 27, 2026 Last reply: 17 hours ago 5 Replies

First reply

OakSpark

The issue you're seeing with the faceted look is directly related to how the STL format works when converting a smooth NURBS surface into a triangulated mesh. That process of tessellation is essentially the computer approximating a curve using thousands of tiny triangles. When…

Read the full reply

So I've been modeling a prosthetic hand prototype for a class project and the fingertip shapes have these fine vertical ridges on them, kind of like the texture you see on a fingernail up close. When I export the part as an STL and reopen it, those ridges turn into a bunch of tiny triangular facets and the whole surface looks chunky and low-res. Someone in my group joked that nails are basically STLs already and honestly the comparison is sticking with me. Is this just how the mesh format works, or am I exporting with bad settings? I tried bumping the deviation and angle tolerances in the export options and it got smoother, but the file size ballooned to something the shop's slicer chokes on. Feels like I can pick either decent detail or a workable file, not both.



Would it make more sense to model the ridges as an applied texture instead of actual geometry, so the base surface stays clean and the STL stays light? I'm still pretty new to surface modeling so I might be approaching this completely wrong. Any pointers on what a reasonable workflow looks like here would help.


The issue you're seeing with the faceted look is directly related to how the STL format works when converting a smooth NURBS surface into a triangulated mesh. That process of tessellation is essentially the computer approximating a curve using thousands of tiny triangles. When the ridges are very small, the mesh engine has to use a lot of extremely tiny triangles to capture the fine detail, which is what makes the file size explode. You're running into the classic geometric fidelity versus file size trade-off. You can't really get both high detail and a tiny file unless you simplify the geometry itself.

Regarding using an applied texture instead, that won't work for the final 3D print. The slicer software doesn't use texture maps for physical shape; it only reads the triangular geometry to calculate the path the nozzle takes. So, if you want the ridges to acutally exist on the final part, they must be modeled as solid surface features. If you're just applying a picture of ridges, the resulting print will be perfectly smooth. If the detail is crucial, you need to accept the larger file size, but before you commit to the highest tolerance setting, check your modeling approach. Sometimes, instead of relying on incredibly sharp, thin vertical lines, it helps to use a slightly broader, more organic curvature for the ridges. This allows the triangulation to capture the shape with larger, more manageable triangles rather than forcing it to resolve millions of microscopic facets. A slightly broader curve often looks better even at a lower resolution because the errors are less visible. If the ridges are truly microscopic and aren't visible at the intended print scale, simplifying the geometry is the most practical solution. Focusing on structural integrity and overall form is ussually more important than replicating a fingernail's microscopic surface texture for a functional prototype.

I hit this exact wall on a guitar pick I modeled with a grip texture. What saved me was building one single ridge as its own surface body and then using a linear pattern instead of sketching all of them onto the main surface. The export seemed to handle the repeated feature way better and the file stayed small enough that the slicer didn't complain. I also set the deviation tolerance tighter than the angle one, which felt backwards to me but gave a smoother result at the same size. Curious how wide your ridges actually are, because there's a point where the printer just can't resolve them anyway and the geometry fight stops mattering.

This reminds me of when I printed a knob with knurled edges and the bumps came out half melted. The printer nozzle was literally wider than the gaps between the ridges, so all that extra detail in the file was pointless. Might be worth printing just one fingertip at low detail first to see if the ridges even survive at that size. How wide are your ridges compared to your nozzle width?

somthing that worked for me on a textured drawer pull: I exported at the coarse setting that looked chunky on screen, then ran the mesh thru the decimate tool in my slicer's repair utility. Knocked the file down to a fraction of the size and honestly the print looked about the same. The preview always looks worse than the real thing anyway. Also second the nozzle width point above. I printed that same pull twice, once with a 0.4 nozzle and once on my buddy's finer one, and the texture only really showed up on the fine machine. If the shop only runs standard nozzles, those ridges might be wasted effort no matter how you model them.

One thing I'd add is the print orientation. If those ridges run vertical on the fingertip and you print the hand the way it'd sit naturally, the layer lines are basicly running the same direction as your ridges. On a lot of printers they'll just melt togehter and you'll barely see them anyway. I had that on a fob I printed with grooves, flipped it 90 degrees in the slicer and suddenly they showed up. Might let you drop the mesh detail way down since the printer's doing half the work for you.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required