Revolved objects are not perfect circles

I’m modelling a light fitting, essentially a cylinder, and due to the profile I sketched half of the cross section, then revolved it around an axis to produce the cylindrical body.

One part has a knurled texture, which I now need to apply to the outside surface. To do this, I’m creating semi circles, then copying them rotated around the centre. The problem is that the flat part of the semi circle (actually an arc) doesn’t intersect the surface created through revolve, because that surface isn’t a circle, but appears to be a series of lines/arcs between each radian.

Attachment shows difference between semicircle on the edge of a circle sketch, and the revolved body.

My intention is to repeat the semi circles, then union the bodies to create a single body. This doesn’t work because I end up with non-intersecting geometry. If I use circles to guarantee the intersection, I’d end up with redundant geometry within the combined body.

Revolving a sketch seems to give a much better body than fiddling with lots of extrusions (and can make revisions easier to apply), but it seems odd that the result isn’t perfectly cylindrical.

Can anyone suggest a way around this?

Hi, Could you please upload the Shapr3D design and tell a few words about how the sketches were made?
My first idea for the cause is connected to the Project tool. If you have projected anything, please note that the projected sketches will be non-constrained sketches. If any other sketch element is constrained to them, it can happen that they are moved - even if they were a projection of an edge.

I’m designing on iPad so not sure how to upload the design.

This is much the same as the ‘fins’ tutorial, in that I’m applying a sketch to the perimeter of a circle, extruding it and then copy/rotating it around a centre point. I think in the case of that motor assembly, the body was created from an extruded circle, as opposed to a revolved sketch, so the fit, if examined, would be ok.

To recap:

  1. Draw a rectangle (e.g. 1mm wide, 10mm high) on vertical plane 30mm from vertical axis.
  2. Select the rectangle face and the vertical axis, then ‘revolve’, 360 deg. This should result in a cyclindrical body, similar to a ring.
  3. Select the horizontal plane for sketching
  4. Draw a circle from the centre to the outside edge of the body (30mm radius)
  5. Move to the edge of the ring, draw a circle that intersects the outside edge (50/50) radius 1mm
  6. Trim the part of the circle that is inside the body, and the main part of the circle created at 4, leaving a semi circle on the outside of the right, close with an arc.

You may already be able to see that the inner arc of the semi circle does not match the outer edge of the ring, as the ring appears to be constructed from straight lines/arcs, and not a circle. If you don’t see this, try copy/rotating the close semi circle around the centre point of the ring, say 3 degrees per copy. Zoom in to see how well the fit is, but you should see what I uploaded earlier.

The light fitting is not a simple cylinder, but the above illustrates the point I’m trying to understand. I can see how copy/rotating works when the adjacent body is also constructed from an extruded circle, but when its a revolved sketch it appears not to be a good solution, due to the mismatch between the paths.

As an alterantive, I tried using the Sweep function, to apply the shape to the circumference of the circle, but it seems that Shapr3D also divides these into radian/arcs. If you zoom in enough on a revolved/swept body, you can see that the surface is made up of lengths of lines/arcs that don’t truely describe the expected path. Clicking on the edge selects the entire circumference though, so its probably constructed as a spline. It seems therefore that revolve/sweep has some inherent ‘granularity’, and its not clear how that might translate if one was to actually export the design, say to a 3d printer. Right now its just causing me some concern as to the best way to join objects without ending up with lots of redundant geometry to clean up.

For a circle to be perfect, one might measure an infinite number of points around the circle’s circumference to know for sure. Each point would need to be precise from the particle level to the molecular level, whether the circle is stationary or in motion, which makes determining perfection a tricky feat.

As far as 3D printing accuracy?? All in the beholder? :sunglasses:

I agree, it depends on the scale as to whether it’s significant, but on my example 40mm radius circle, I’m measuring straight sections of the circumference at around 2.3mm. I can’t measure the angle between segments easily, so it’s hard to tell how (or if) they would translate when printed (if that was the intention).

Of course, it’s possible that this is an anomaly of the shapr UI, and in fact on an exported model the spline segments are accurately modelled. It just makes intersection modelling a bit confusing that two things that ‘should’ be in alignment don’t appear to be.

1 Like

Like a lot of things, translations can become jumbled/distorted at best when incoming/outgoing transmissions/languages/numbers are just slightly askew?


It’s simply a glitch at that moment? :sunglasses:

Are you saying you are perceiving poor 3D print quality of a circle, from an Shapr3D file? The prints I’ve done from Shapr3D have always been identical to what I’ve observed from SolidWorks. They are the same engine.

I’ve not attempted to go to print at this point, so I’ve no idea how this translates (and I’m specifically concerned with revolved/swept bodies because that’s where I’m observing this segmentation).

I was asking this question to see if anyone else had noticed the disconnect between arcs and revolved/swept surfaces, and further wondering if this is a shapr3d UI issue or if it translates to print or other uses of the design file/export.

I’ve further figured out today that making a revolved body (a cylinder), then a separate body from and extruded circle of same dimensions, which is then embellished with ridges as I originally described, and finally both bodies are Union combined does leave a single body without extraneous geometry. The external face appears to lack the segmentation exhibited on the revolved body, probably because that segmentation falls inside of the Union and is omitted from the combined body. So this simplifies the way I might tackle my project, but leaves the question of whether segmentation of revolved/swept bodies is noticeable on printed/exported articles.



I have been 3D printing for 7 years, using a number of 3D drawing programs. SketchUp, 123D, SolidWorks, and Shapr3D. The only difference is that Shapr3D is easy to use. The output of 3D format is standard STL as far as I can detect in all attributes. Shapr3D uses the exact engine that SolidWorks uses, which is ParaSolid. I have printed thousands of items using Shapr3D, with curves, circles, squares, holes, etc. ALL my parts have come out excellent. There are more issues with the 3D printers themselves, than you’ll ever observe from Shapr3D itself. I own a full license for SolidWorks. I haven’t touched it in 2+ years. I have 3 of Zortrax M200, a Voxel8, an Ultimaker S5, and a Form3 resin printer. Plus a starter printer.


Check the settings for resolution. It may be set low to run faster. If the circle was done in two half’s make sure the two half’s are tangent with each other or you will get flat spots on the top of the sphere. The second thing is the reason the two bodies are not unionizing is it looks like the plate and the circle are not attached. You can do one of two things set the circle into the base plate so they intersect or draw lines straight down into the base plate. Union requires the two bodies to be over lapping.

No worries, it is a circle. What you can see is only a visualization issue (since the rendered body is a mesh). The geometry is precise.


That’s really what I wanted to hear, that real world this isn’t an issue. Thanks!