Why are Class-A surfaces such a big deal for cars, but not industrial design / architecture?

That’s mostly a limitation of your machine (controller and motion control) as well as your software. With decent machines and controllers any kind of free-form (2D or 3D) is absolutely no problem.

The moment you go freeform, the toolpaths are gonna be interpolated.

Even on a high end mold making machine like Makino, I ended up with parts that needed a lot of polishing work to look “perfect”. These were the “Class A” surfaces I designed with “perfect” G2 continuity for the reflection I intended them to have.

The only problem is the toolpath interpolation, it always show up in the final part. And the only way to get rid of them is by extensive polishing (or lowering ones’ standard for perfection.)

Now my parts designed with manufacturing constraints are coming out more “perfect” than ever. It does not mean I am avoiding freeform curves/surfaces. I just have to interpolate them directly within Rhino when the appearance is critical.

If only machines can run directly off NURBS…
[Edit: It looks like some higher end machines can run off NURBS, so I guess it’s all about cost!]

He maybe had in mind that to CNC-mill a 10 millimeter radius in a wooden or aluminum block requires just a single pass with a 20 millimeter ball tool. I also did CNC-milling some 10 years ago and G2 transitions required applying tens of milling passes every 1 millimeter in order to achieve a smooth surface. Reducing machine time and cost (which is a combination of time, CNC operator wage, CNC-machine amortization, tool amortization, electricity, inability for both the operator and the machine to do other work meantime) are the main reason why so many mechanical parts that are made via CNC-milling have a G1 inner radius for the deep areas and chamfer (usually 45 degrees) to eliminate the outer sharp edges.


Yes, toolpaths are going to be interpolated - and especially anything in 3D. As I said, it’s about the quality of the machine as well as the CAM software. If you have a good controller and motion control you can throw massive amounts of very fine G1 (linear interpolation) data at the machine and the machine will deal with calculating the accelerations and interpolations necessary to cut a smooth toolpath. This already since 10-15 years or more.

You will always need some degree of polishing - part of that is simply due to machine and tool vibration and the fact that a spherical end mill cuts differently depending on where on the arc edge you are cutting (the center of the tool cuts not at all) and which direction you are moving.

Absolutely - that’s how it works. You let the machine do the work. And again with a 20mm ball end cutter, the “dead” zone near the center of the tool is quite large, so near-horizontal surface areas may end up with a poor surface finish. I have a tendency to do finish milling with a smaller tool and even more passes.

I guess I didn’t put the full context here. 3D machining with stepover <0.1mm on hardened steel (62hrc) and grade 5 titanium. I’m just chiming in why it could happen, on topic of while it isn’t always common in industrial design, especially in actual part made.

I guess we’re nitpicking in different level of surface finish and outcome here. For most machines actually, I agree 3D milling isn’t a problem at all. We’re living in 2020 after all.

1 Like

The following video is interesting from a design and mechanical standpoint. There is one part of the video where the guy talks about the importance of Class-A surface finish. But my favourite part of the video is just before the 3rd minute, because it reminded me that Rhino has the coolest icon ever! :smiley:



There is the surface according to the design math data, and then there is the reality of the resulting physical part.

Stamping sheet metal will always result in a “smooth” transition, possible small in size, when there is a change in curvature. Even if the dies have “exact” G1 continuity the metal coming out the press will have a curvature transition, though it may be narrow.

Any time a die or mold is polished any G1 transitions will be smoothed to some extent. How much depends on the amount of polishing and the skill of the person doing the polishing.



1 Like

…and locked down, put on our knees by one of the smallest creatures in the known universe.

Ok, so this turned out pretty crap, but maybe someone can get something out of it:

These are almost the exact same zebra checked G2 blends on the same product, but:

  • Top surface is milled aluminium
  • Bottom surface is injection molded plastic

There’s a line on the milled part that is much more apparent on the entire product, but unfortunately I couldn’t leave any identifiable information in the picture. :sweat_smile:

Injection moulding matrix are hand finished for better appearance, therefore the area between a flat surface and a rounded corner is less apparent. Also, injection moulded plastic products usually have a matte finish or a certain type of added texture for several reasons:

  • Hiding the 3d model imperfections and/or the marks leaved by the faster CNC-milling.
  • Making the surface more scratch resistant (in fact, it’s less resistant to scratches this way, but it’s much easier to spot even the slightest scratches on a polished plastic surface than on a textured one).
  • Reducing the possibility for sweating (important for game controllers, remote controls, knife handles, door handles and many others)
  • Increasing the heat dissipation capabilities due to the larger surface area.
  • Reducing the chance for gaining static electricity via friction with other objects.

Well, just to let you know how bad the above pictures above really are… that injection molded part has a high-gloss mirror finish. :rofl:

Anyway, what I forgot to say with regards to the CNC machining is again: Apple cares.

I immediately downloaded, printed and framed your picture.


Besides all these very informative replies, the idea is that the digital model is as good as possible, so that you have a base to judge the quality of the output after production and maybe even respond to it. A large error of the digital model plus a large error of the production process is likely a greater error, than having only a large production error.

I think if you listen to all this someone could conclude you don‘t need to go for digital perfection because you won‘t see it later anyway. Whats visible or whats not, is not predictable, so you want a greater resolution as required. Its that simple.


Going back to the original question- Class A just means visible surfaces of a product. Class B is surfaced or modeled not visible to the user. When injection molded- these will typically be on two different halves of the tool. Cars have certain standards for surface continuity and tolerances for this, so do products. Most consumer products are on a small scale where highlight changes are masked with textured and acid etched surfaces. I don’t buy that every blend has to be G2 and above- that’s up to the designer. Its a false badge of honor that usually doesnt translate how you think it will into production. A G1 tangent fillet is perfectly appropriate for most human-scale- non-shiny products. In fact, if you want a product to look machined and precise you want to go for a G0 or G1 surface transition.


The Dualshock 4 game controller is the perfect example of a badly made decision to use G1 instead of G2 for the transitional surfaces. Even though its body has a matte surface finish (and painted shiny finish for some of the limited edition versions), the G1 transitions are super obvious and make the controller look cheap (quality-wise).

Under certain angle the highlights expose the G1 nature of the transitions and make them appear as a separate portion of the model. Notice the right side handle next to the right analog stick. I owned a PS4 several years ago and was shocked how those G1 transitions looked way worse in person than on photos.


1 Like

I think your vision might be biased. I, for instance, never noticed nor was ever bothered by any of things you noted multiples times in this thread. I even had a hard time identifying the things you pointed out by reading your text and looking at the pictures.

I never really payed that much attention to my controller, so I never related these surface continuity problems with cheap quality. On the contrary, by looking at it and holding it, it seemed like a great quality improvement from PS3 to PS4.

Looking at the things you point out, they don’t come to me as shocking evidences of bad quality at all. I can even say that I like the controller the way it is, or at least, I don’t dislike its current state, formally speaking. Functionally speaking it’s a piece of garbage that breaks and fails constantly. Sony is synonym to planned obsolescence, but that is another matter.

But I think that asserting that: ≠ perfect continuity = bad quality is a bit too much. Maybe you are obsessed with continuity by working all day with it and seeing something not 100% continuous annoys you, it happens… :man_shrugging:

1 Like

As I mentioned early, the G1 transitional surfaces of the Dualshock 4 controller also have a negative impact on the ergonomics. :smiley: The area where the middle finger is placed, in particular. Should it had a twice as smaller radius, it would be much, much more comfortable for holding it. It didn’t followed the basic rule that a similar transitional surface meant to be in a direct contact with human fingers must have a bit smaller radius than the radius of the finger itself. It’s a mistake also present in some of the cheap power tools such like drills, jug saws etc.

Dualshock 4 looks and feels considerably better than Dualshock 3, because the latter was very uncomfortable and fragile. DS3 also have extremely badly designed triggers at the back that featured springs that felt unnatural when pressed. These springs also broke easily, because they consisted just a single coil, leading to a premature metal fatique. I used to fix multiple numbers of these back in the day as a side job. However, despite DS4 being better than DS3, the former is still much less comfortable than the Xbox One controller.

1 Like

I was simply talking from an aesthetic standpoint. I did notice your ergonomics argument, but I didn’t feel it was necessary to say that I never felt the discomfort you talk about either. I never placed my middle finger there, but below the controller instead as support.
Like so: