TrueType instructor information

Hey there!

Very much looking forward to the new handbook especially to get to understand the power of the new TT inspector. I’m hoping where needs be to use it in conjunction with Werner’s built in TTFAutohinter

So I got hold of @DunwichTypeFounders Rhodium Libre and was excited to take a look at the TT hinting information. I guess I’m after more information. ie. the difference between stem hints and align points. I don’t understand the point of cycling through pt sizes if you can only create one set of hints. In the s glyph of Rhodium the hint info is represented by purple lines, seemingly everywhere else it’s light blue. Are the alignment buttons preferential snapping directions? Basically I’d love all and any information. Any external links that might explain the glyphs methodology better also welcome.

Lastly - if that wasn’t enough! - do you have any recommended protocol regarding testing? I find the load to InDesign and zoom in/out advice a tad unscientific. The best I can come up with thus far is uploading somewhere and testing via browserstack.


PS. James if you are about I would love to learn more about your process. My understanding is that by manually adding hinting information TTFAutohinter is not applied. Since you’ve put hints on every letter I ask myself if it fell short somehow?

that is not possible. It is quite complicated to have two kinds of TrueType hinting in one file. TTFautohint might have support for it but as the hinting strategies are quite different will most likely produce uneven results.

They are very close. Stems are kind of like aligns but will stick to a stem value.

Set up a virtual machine and run a Chrome and internet Explorer in Windows. There is no way around that. It works much better as it sounds. You export in an shared folder so you just need to change to the VM and reload (that should be possible with a script, to reload the page every second or so).

Hey Georg —

Thanks for getting back to me.

I was wondering if you thought using BrowserStack as a VM was a bad idea or just a bit slower than running your own?

With CFF hinting it was recommended to only hint one master which was then applied to the others. Can I take it this is not the case with TT instructions and that each master requires its own? How does this effect interpolations?

I notice the red DirectWrite background doesn’t change as I move between masters?

Lastly (lastly) I noticed this in your blog on the TT update.

Could you explain the purpose of those green lines? They seem to come out of the interpolation option but bugger me if i know what it means.


I never used BrowerStack. And I don’t know how you can feed it your HTML. Uploading the fonts all the time would be time consuming. The point is that the VM can read the fonts directly form the export folder on the Mac. So if you export and switch to the VM they are already there.

You need to hint only the first master for TTF, too.

It shows instances, not masters.

That is the representation of the TT interpolation command.

If you’re hinting with a target size in mind it makes sense to work with that point size. In Rhodium I targeted 10px for Latin and 14px for Roman. I’ve also reviewed various spots at multiple sizes to see how the hints behave. This is very helpful when hints do weird stuff and create black blobs that cannot be understood at actual size. That can happen a lot in Devanagari.

IIRC That’s a triple hint—three hinted stems that maintain even white spaces between them. Georg, there should probably be a more obvious way to identify these. Maybe a “3” could appear in the hint info box when it is selected.

QA on the Latin hints is easy because most of those Glyphs have simple shapes and unicode points. But the Devanagari has over 350 unencoded glyphs, which have up to thirteen white/black transitions. To proof them I would have to either create new software or dump my brute force Indesign proofs to HTML and start exporting fonts and refreshing the browser, scrolling back down, etc. Nobody is paying me to do the hinting, so I’ve left the manual hinting out of the release fonts until Glyphs implements a pixel preview.

Interpolation attempts to position the touched (middle) node evenly between the other two hints. I’m finding that while it was very useful hinting for grayscale in the x direction, it doesn’t work too well with DirectWrite in the Y direction, so I’ve used it sparingly in Rhodium.

@GeorgSeifert @DunwichTypeFounders Thanks to you both for taking the time to reply.

That seems peculiar. I would imagine it’s rare that a single size would be so dominant though of course it helps to aim there or thereabouts.

Is a pixel preview scheduled?

When used to set running text fonts are used in a narrow range of sizes. Devanagari text tends to be set between 14 and 18 pixels. So I targeted 14 pixels, the worst case scenario, and then reviewed the larger sizes after.

I don’t really know if to speak about a target size is useful. Hinting is meant to produce output on various sizes. There are always sizes that look better then other but you need to make sure that the fonts looks good at all of them. There are devices that may have 150% resolution so to get the same physical height, you need to set it in 21px.

Glyphs tries to adjust stems and zones that should give a even output. I’m working on an option to set that manually and to provide better preview.

That’s true. But as I noted, the target is really the worst case scenario size. Because if the hints work well at the smallest size they will probably work well at larger sizes, the exception being things like the spine of S and the crossbar of E, which are going to jump up and down at different sizes.

Is it planned to be able to adjust hints for different pt sizes in Glyphs?

Two of the most commonly repeating characters - the jumps being a dagger to the heart! But such is the medium (and life)

What exactly do you mean? There are Deltas and I hope that I can implement them somehow.