Thanks for the feedback! I’ve got mixed feelings on some of this, so I’ll start with my rationale: I want there to (ideally) be visual consistency across all icon and text modifiers, such that the stroke widths are (broadly) the same in all places. When it comes to how to treat special symbols with this rule in place, you are left with some difficult choices at points: should the glyph be larger to compensate for its (relative) small on-screen size? What if this means that it ends up looking very ‘thin’ as a result compared to its peers? I’ve opted to try and keep things somewhat similar to the originals. There are some glyphs that aren’t even consistent within the typeface though, for example the @ symbol in Gotham looks pretty bad by default, and wholly unsuitable for unmodified use here..................
In my view, it's more important that things feel consistent and aligned
perceptually than
actually. Sometimes the two are the same, but sometimes they differ. As you probably know, fonts often contain various 'tricks' that make things
seem properly aligned by purposely
misaligning them. Like the way the gap between the letters AV is artificially shortened to the point that the V actually starts before the A ends - it's out of whack with the spacing of other letters, but it just feels the most natural to the eye this way. Or how fonts will often artificially lower the end points of their curves, so that the bottom curve of a capital U sits a little bit lower than the bottom of a capital H - the U will often actually feel 'wrong' and too high if it's placed 'correctly' in line with the H.
I would put the line weight of that Gotham @ into the same category. I think the original that you screenshotted looks quite good - at least at that large size. There's something very visually and psychologically dominating about surrounding a letter with a near-complete circle, so I think it's smart of the designers to de-emphasise the circle by thinning the line weight. This balances the two elements and lets the eye continue to be drawn to the most important part of the glyph (the "a" in the middle) rather than being drawn outwards. It also helps neutralise the sillhouette of the glyph by making the "a" still feels like it 'owns' the glyph: so that it feels like 'a small
a with a circle around it' rather than 'a big circle with an
a inside'.
If they had kept the line thickness the same in the circle and the "a", they would have had to significantly increase the padding between the two so the "a" could still breathe, which probably would have overblown the overall size of the glyph. This is the dillemma you talked about. I'm sure that the original thin line weight would be no good at the tiny size of a keycap though, so you're right to thicken it. However, I don't think you should have thickened it to 100% match the "a": the symbol kinda can't breathe now. It looks very dense. Still, I think your v2 is a great improvement on the v1. The three-quarter circle looks much more elegant, and the overall size and placement seems a lot better. The new * is good too.
By the way, I'll add that I'm being purposely nitpicky, for the sake of trying to provide sharp design feedback. I've just now looked at the @ on my own Signature Plastics DSA keyboard and noticed that it looks kinda crap, with the hook of the "a" coming out at what seems like a wonky angle. Obviously I'd never scrutinised the @ on my own keyboard as closely as I did the poor @ in your design renders
The only other feedback I can think of at the moment is that the \| look uncomfortably close to me - like they're one long shape that's been slightly broken. I'd maybe shorten the \ a little or tilt it a little more, and/or move the | a little to the right, to give the 2 elements some more visual separation.
I also imagine that many people would prefer to have the traditional "win" or "command" key instead of "super" or "meta", so since you're already supplying alternatives in the base kit (1.5 ctrl, 1u super, 7u space, etc.) maybe you'd want to throw in a 1.25 win/command too.
Forgive the suggestion if this is already something you do, but it might be useful to do the old designer's trick where you assess your design while looking at it in the mirror (or flipping it horizontally in Photoshop). It helps the brain perceive a design as a whole by stripping away some of the familiarity of individual elements. This would probably be particularly true of a typographic project like this, where it'd be hard to look at abstract qualities like balance, weight, and negative space when your brain is constantly trying to
read the letters. One of my favourite examples of this trick is to flip Australia. I don't know about you, but to me, the normal Australia looks balanced and more or less horizontal, while the flipped Australia looks crazy skewed and tilted (even though it's only been flipped horizontally)
