Fixing Signature Plastics SHIFT key legends digitally

[zslane]

I'm moving the discussion that began in 7bit's Round 5a group buy thread over here where it belongs.

I'll begin by showing some images of what I have so far. I invite commentary.

First up is a scan of my Dolch DSA 2.75u shift key:

Dolch DSA 2.75u shift key (original scan)

dolch_scan.jpg (20.45 KiB) Viewed 8373 times

Next is the same keycap scan, but with my digital version of the legend with the same F T ligature:

Digital legend with original glyph separation (gs = 1)

dolch_kd100.jpg (20.31 KiB) Viewed 8373 times

Here is the digital legend where the glyph separation between the F and T is uniform with the separation between all the other letters in the legend:

Digital legend with uniform glyph separation

dolch_kdu.jpg (20.3 KiB) Viewed 8373 times

Here, the glyph separation is 75% of the original glyph separation:

Digital legend with 75% glyph separation (gs = 0.75)

dolch_kd75.jpg (20.34 KiB) Viewed 8373 times

Here the glyph separation is 50% of the original:

Digital legend with 50% glyph separation (gs = 0.5)

dolch_kd50.jpg (20.36 KiB) Viewed 8373 times

Here the glyph separation is 25% of the original:

Digital legend with 25% glyph separation (gs = 0.25)

dolch_kd25.jpg (20.32 KiB) Viewed 8327 times

And last, no glyph separation at all:

Digital legend with no glyph separation (gs = 0)

dolch_kd0.jpg (20.28 KiB) Viewed 8327 times

[zslane]

For comparison, I also scanned the RETURN key and compared the glyph separation between the E and T with the glyph separation between the F and T on the shift key. Guess what, they are identical.

[Muirium]

Guess what: that's not kerning. That's its absence!


https://en.wikipedia.org/wiki/Kerning

[chzel]

Well, in this case kerning is impossible, what we need is F and T almost touching each other.
zslane can you do one with 25% separation? 50% still seems too far away.

[Muirium]

Try zero separation. A true ligature!

[7bit]

This looks pretty good to me.

I'm afraid Melissa will tell you that this is not possible, because the gap is too close, but then there is such a little island in the messed up ¥ key, that I don't believe there is such limitation.

[zslane]

I've updated the original post with 25% and 0% separation.

[chzel]

In my eyes the 25% is the best, but I miiight be biased

[zslane]

I'm afraid Melissa will tell you that this is not possible, because the gap is too close, but then there is such a little island in the messed up ¥ key, that I don't believe there is such limitation.

If so, then the benefit of Muirium's suggestion becomes readily apparent. There is no gap problem; the F and the T become a single shape.

[Elk]

But now the 'S' pops out

Le can of worms...

[zslane]

Most of this is just optical illusion, of course. If you measure the space between the E and the T on the RETURN key, it is the same as the space between the F and the T on the shift key. It only looks funny to our brains because the shift key's T has nothing to its right to help its spacing appear balanced.

I don't know the typography rules for kerning trailing (uppercase) Ts, and in the absence of a clear standard I'm wondering if we'll be able to reach community consensus given the subjective nature of the perception of the separation.

[mtl]

50% looks good. Where's the poll?

[Muirium]

25% is the best looking. But I'd like to see what happens to it once it's passed through SP!

Optical illusion isn't quite right. What you're talking about is negative space. A key foundation of typography.

If it looks right: it is right.

[zslane]

If SP's double-shot mold process does indeed impose a minimum space between legend elements, as 7bit indicates, then our choices will be severely limited.

I refer to it as optical illusion in the sense that letters may only appear closer or further apart than they actually are.

[cswanic]

It sounds like what we are strictly discussing is visual representation of the word SHIFT correct?

So this discussion isn't about the Row 4 SHIFT key?

Zslane, would you be able to add a bit more to your first post on the objectives being sought. I am very interested in your potential with correcting any graphic fixes to SP so I want to be able to have more of an understanding on the thread. I have been following the topic discussion you bring to the Round5a design and I am keenly interested in what kits you decided to go with and more commentary on the thoughts to making this kit and others like it.

[zslane]

Let me first clarify that the "row 4 problem" is a completely separate issue that we, the mechanical keyboard loving public, have no influence over. Signature Plastics is supposedly working on new row 4 molds for 2.25u and 2.75u SA-family keys, but they have no clear timetable for them (this based on a recent e-mail exchange with Melissa at SP).

Fixing the space between the F and the T for their double-shot keycaps is what I am investigating digitally. Since I don't have an SA-family shift key to scan, I am starting with the DSA-family shift key from my Dolch set. That keycap will, in effect, be our test case for this approach. If all goes well, the double-shot DSA shift keys will have "corrected" legend molds and a DSA-based GB could ostensibly be organized, starting with the Dolch colors.

I can't do the same thing for SA-family shift keys until I have a scan to work from, or alternatively, conclusive confirmation that the SA-family shift key glyphs are 100% identical to those on my DSA Dolch shift keys (allowing me to use one vector definition for both). In any case, if I manage to create a corresponding vector file for the SA family, then in theory a GB for corrected shift keys could be organized around whichever colors manage to collect the necessary MOQ.

I've never worked with SP before, and have only corresponded with Melissa a couple of times, so I have no feel for how long this process will take. In particular, the iterative process of sending a vector file, getting a prototype made and inspected (by me), corrections made, wash, rinse, repeat, until an approved mold is produced will take an amount of time I can't begin to predict. It would be great if new SA legends could make it through this process in time for 7bit's Round 6, but it's way too early to even hope for that. We don't even know yet if SP can succesfully make what we are asking for.

[potatowire]

I think this is a very worthwhile effort, and I'm glad you're keeping the press on. Let me know what I can do to help. I have some SA theoretically on its way to me, and I will scan the SHIF T, if you aren't already set by then.

[zslane]

Thanks, potatowire, I may very well need your assistance. For the work above, I scanned the keycap at 1200dpi with the cap in the center of the scan bed. Making sure the keycap was as straight as possible along the horizontal was tricky; I used a CD jewel case as an alignment object.

You can crop the scan to just the legend or to the entire keycap, it's up to you. With a scan of the whole keycap I can provide mock-ups like you see above, but obviously it will big a bigger file. I also would prefer a PNG rather than a JPEG; lossless compression vs. lossy is always better here.

[7bit]

But now the 'S' pops out

Le can of worms...

Right.

I think 50% or maybe 25 are looking best. Just give it a try. Tell them to do them with the small gap and that they should produce them to see what happens, no matter if they feel that we might not be pleased with the result.

[zslane]

So here are some measurements to give context to the spacing/gap discussion. They are all approximations accurate to about 0.001" at best.

• The space between the F and the T in SP's current shift key legends is ~0.027"
• The space between each of the other letters S-H-I-F averages ~0.023"
• The space between the F and the T in my 75% sample is ~0.020"
• The space between the F and the T in my 50% sample is ~0.014"
• The space between the F and the T in my 25% sample is ~0.007"

Today I received a PDF from Melissa describing their requirements for custom double-shot molding. That document says:

On any 2-shot legend, there needs to be a minimum space of 0.020" to 0.035" between characters and lines...

That makes the 75% spacing the narrowest spacing we should expect to have any hope of success with. And with their margin of error, an F-T gap of 0.020" in my vector file could still end up being closer to 0.025" when it comes out of the mold, and yield no noticeable improvement at all. I could instead send them the version with 50% spacing (0.014") and hope that it actually comes out closer to 0.018" or 0.020", but I'm not so sure that throwing money at the 25% version would be productive use of funds.

Thoughts?

[mtl]

What if you spaced everything out more instead of bringing the T in?

[zslane]

I imagine SP can do that without my help.

What do people here think of the idea of spreading the other letters further apart? That might make the shift keys look like they come from a different keycap set.

[chzel]

I thought the concept was to make an FT "symbol" and use that instead of the actual individual letters...
Spreading them out would look totally out of place...

[zslane]

Do you mean the true ligature version where the F and the T are merged together with no space in between? That one, I feel, has the best chance to succeed, but I'm not sure everyone likes the look of that. Most folks seem to prefer some small amount of space between them, in which case we're potentially running up against a manufacturing tolerance limit with anything under 0.020" separation, and even 0.020" seems to be pushing it.

[7bit]

So here are some measurements to give context to the spacing/gap discussion. They are all approximations accurate to about 0.001" at best.

• The space between the F and the T in SP's current shift key legends is ~0.027"
• The space between each of the other letters S-H-I-F averages ~0.023"
• The space between the F and the T in my 75% sample is ~0.020"
• The space between the F and the T in my 50% sample is ~0.014"
• The space between the F and the T in my 25% sample is ~0.007"

Today I received a PDF from Melissa describing their requirements for custom double-shot molding. That document says:

On any 2-shot legend, there needs to be a minimum space of 0.020" to 0.035" between characters and lines...

That makes the 75% spacing the narrowest spacing we should expect to have any hope of success with. And with their margin of error, an F-T gap of 0.020" in my vector file could still end up being closer to 0.025" when it comes out of the mold, and yield no noticeable improvement at all. I could instead send them the version with 50% spacing (0.014") and hope that it actually comes out closer to 0.018" or 0.020", but I'm not so sure that throwing money at the 25% version would be productive use of funds.

Thoughts?

Submit the 50% version.

How much does she want for it, plus some 1.25 units sample keys in whatever color (any of Round 5a)?

[chzel]

I meant the 25 or 50% but treated as one "symbol".

[Muirium]

SP's doubleshot rules are inconsistent, as 7bit already described.

If we could add a new glyph to their Gorton typeface, it could indeed be one "symbol" which happens to be the 25% or 50% spaced FT above. (I much prefer 25%, but there's squabbling already!) It would still be a ligature, but not touching.

But all of this depends on what SP can be bothered with. And will always require iteration.

[potatowire]

I meant the 25 or 50% but treated as one "symbol".

As I understand it, this is the way it would have to happen.

[zslane]

Perhaps I'm daft, so please bear with me.

In a vector file like I have, there are five independent closed paths. One for each letter. There are no "symbols". Just closed paths (possibly with an "island" situated within, but we don't have that in this particular case) and a distance between each of them. Unless the F and the T are merged into a single closed path, they will always be two separate shapes, and the space between them must be at least 0.020".

So are we talking about merging the F and the T as in the 0% version I showed above, but with the top (merged) crossbar stretched out horizontally such that the vertical stem of the T is in the same position as the 25% or 50% version?

[Muirium]

You're taking SP's specs too literally. There is wiggle room. They won't tell you about it, but their fabulous knack for errors exists within it. Basically, they aren't as competent as they say or think. They don't fully understand their own equipment. They manufacture things without that point whatever lower limit. Sometimes even on purpose!