What are you thoughts on the new MT3-inspired MTNU and URSA designs?

EDIT: just realized Andreas already started a thread last month that I somehow missed.

I don’t know if MT3-inspired is the best description for Matt3o’s newly announced MTNU profile and Andreas’ URSA profile, but it’s my attempt to get at what I find intriguing about these recent developments in keycap design. I love MT3, it’s been a one-of-a-kind for me, so I’m both intrigued with the development of two very similar profiles — one by MT3’s chief creator — and curious what others in the community think and expect.

If you’re wondering what on earth I’m talking about, for context:

  • Matt3o has decided to give sculpted spherical profiles a second shot, this time aiming for a design that might be more suited to more general use on a wider range of keyboards compared to the more eccentric MT3 (that’s my attempt at summarizing my takeaway, see his blog post here.)
  • Andreas reacted to the above announcement by sharing his own designs for yet another spherical, sculpted profile, which is also similar to MT3/MTNU and also aims to respond to specific needs and tastes that MT3 hasn’t turned out to be as suited for.

There are elements of both designs that intrigue me, especially the fact that both are looking to solve the “sharp edges” that some find uncomfortable with MT3 — including me, on occasion — albeit each in their own way. I think they could both have the potential to improve my favorite profile by making it a little smoother and more comfortable.


Hello again :smiley:

I’m very happy for Matt3o to work with GMK as a manufacturer. Coincidentally, I only live two hours away from them, maybe I should give them a call. :smiley:

Seriously, though, I think Matt3o is doing great work with MTNU. And I can’t say it often enough, how open and helpful a person he is.

Regarding smoothness…

With URSA, I iterated numerously to be as smooth as possible, but still maintain some ‘edge’ to the touch. I had it very smooth to the point where it was too smooth. To me, it’s night and day different to MT3 ABS, even more so to my PBT set.

Thank you, Matt, for sparking the discussion.


Thanks for taking the time to answer!

I suppose the challenge is to improve smoothness without losing the lip that is a hallmark of the vintage terminal keycaps that I think inspired URSA and which give the distinct feel of a delicious deep dish.

What’s interesting to me is that you and Matt3o are using techniques that are more subtle than just rounding out the edge — in your case, fine-tuning things like row angles, or in the case of MTNU, embracing a fully spherical dish, with a curved top. Really interesting!

1 Like

You Sir are a keen observer :smiley:

Warning: Nerdy things ahead!

Everything in URSA is ‘super-continuous’, if I can describe it that way. The design is far more than simply adding a fillet. All flow is continuous and smooth in a way I’m not sure any fillet option provides.

I’ve spent weeks obsessing over how the highlight travels across the edges of the key profile and tumbling objects in the 3D view port to judge the surface smoothness.

The goal was to make it look and feel like it had been hand sculpted, just like keycaps of the past.

I think my approach to this is quite unique in the sense that I do not use Nurbs modelling to begin with, but rather quad based subdivision modelling. This process allows me to use algorithms that apply to a much larger data set than any manageable Nurbs surface/solid. Since it is quad based, the data can be converted to Nurbs patches rather easily.


Thanks for explaining, always fun to get a window into the technicalities of the creative process.

You definitely went a little over my head at first, the only experience I have with modelling is some simple architectural work in Google Sketchup almost a decade ago. So I did a little searching, and I think I can now vaguely grasp why you’re excited about the potential of subdivision modelling when it comes to smoothness. TIL!

One thing I’m not sure I understand is: what do you mean by fillet option? Are you referring to NURBS?

Also, do you happen to have any insight into the kind of modelling techniques that are used for other keycap sets?

Previously, I wrote a whole tirade about why I dislike MT3 keycaps.

Now, looking at the currently running group buy for MTNU, I am happy to see that many of the points of critique about MT3 have been resolved:

  • The legends are cleaner and much larger, on par now with AFSA.
  • There is a dealer for Europe.
  • Like AFSA, the keycap height has been much reduced compared to SA, whilst maintaining (most of) the cylindrical profile. This is the most significant improvement for typists who prefer low typing angles.

However, I am still sceptical about the inclination angle of R1, which has been much reduced compared to SA. This reduces legend visibility and probably also typing comfort. As with MT3, Matt3o persists in sinning against this. Nonetheless, The proof of the pudding will be in the eating.

I do like the crossed zero legend. Having a twisted brain and with the letter O and the digit 0 being located next to one another, I sometimes mix those up. That will no longer happen with these legends:


The F10 key and the numpad zero are similarly styled.

1 Like

URSA: Normal and shift legends side by side confuses my brain with epileptic sparks between its hemispheres… :brain: :zap:


Indeed, I changed them back.

1 Like

I think it comes down to costs. Matt3o mentioned this on multiple occasions.

However, I’m not convinced it’s the best compromise, as it impacts usability. Btw, I share your experiences with MT3, and that expierence sparked me to create URSA.


My thoughts after having used MTNU for awhile:

Acoustics: Cherry profile sounds better than MTNU.

I think the difference is likely in keycap cavity size.

Feel: Cherry profile feels more comfortable than MTNU.

While much better than SA profile, my fingers get tired faster with MTNU.

Look: MTNU wins here. No question.


what was the MTNU tester board environment

A bunch of 60% boards with FR4 plate. All sounds more satisfying with Cherry PBT keycaps than MTNU, not unlike the way SA profile sounds deeper than Cherry profile.


Yeah Cherry profile is just so hard to beat IMHO. I’ve tried numerous different profiles out & I always end up going back to good old Cherry profile caps.

1 Like

I suspect Matt3o’s upcoming PBS keycap profile will fare better than MTNU soundwise. Thicker walls (1.7mm!).