LiveBlog of ATypI 2011 Day One

Usual disclaimer for liveblogging: These are informal notes taken by Dave Crossland at the ATypI 2011, and may or may not be similar to what was said by the people who spoke at ATypI on these topics. Probably if something here is incorrect its because Dave mistyped it or misunderstood, and if anyone wants corrections, they should email him immediately – or post a comment!

If you enjoy the details below, please join ATypI πŸ™‚

Raph Levien (G): Spirals in the Cloud

Why make a new font editor? GWF wants lots of fonts πŸ™‚

Spirals? Beziers get lumpy, especially when refining an existing curve; a bezier segment is optimised for mathematical computation, but isn’t what the eye sees, we see total contours. Interpolating splines are similar to real wooden ‘splines’ where you pin the wood with ‘whales’, these big heavy pins.

My spirals are based on Euler spirals, rather than elasticas, which is the maths behind real wooden splines.

This is the fundamental shape: [image of the S curve]. A beautiful curve, with beautiful maths. Its not a physically accurate simulation of an elastic strip, but its great for drawing, and it dates back to Euler’s time in the 18th C and many great mathematicians of the time knew about it.

Its more traditional than Beziers; Ikarus was unveiled at ATypI 36 years ago, and many people feel this is more accurate to their ideas than Beziers. Its been totally displaced in production, but some developers still consider it a good tool. It has a ‘biarc’ interpolating spline; you need a point for every 15 degree of curve. I took examples from the Ikarus paper and redrew with Spiro splines and found I needed half as many points to get the same degree of smoothness in a curve.

So here’s a demo. Its very unfinished; I hoped to get it more to release by now but that hasn’t happened, but development is ongoing.

You have regular polygon points, then the spiral points. You can see the whole curve moving to meet the points. Its not local editing; the maths in the curves guarantees smoothness. Its difficult to get G2 continuity with Bezier curves, but always happens here.

Something else I looked at was straight to curve transitions. This can be hard for new type designers to get to high quality; you get a visual bump if you are not careful about the straight to curve transition. Its similar to road design; if you have an off-ramp, you don’t want a jarring drive. The software I made to calculate these curves is used in the G Car project, and it stops the car having sudden changes in the steering wheel πŸ™‚ Human perception research backs up that smoother curvature is better πŸ™‚

So here’s an attempt to draw a Thorn live πŸ™‚

Here’s the regular spiral points, and now selecting some points we can set them to the ‘straight to curve’ points.

Another thing these splines are good for is bracketed serifs; using 2 consecutive straight to curve points.

One of the long term goals is to return to optical variations; this was tried before with METAFONT, Multiple Masters, TrueType GX… and these historical tools are powerful and there are great fonts made with variation.
But ordinary designers may not use that kind of technology.

This isn’t a finished part of the tool, but here’s an ‘a’ with an interpolatable number of points, and linear interpolation between the control points – and the spiral curves mean the outline stays smooth. Bezier interpolation has no guarantee of smoothness at all. The ‘s’ is more extreme, since the high curvature inside in very heavy type and low curvature in light type is a big difference, and yet this works very well.

So, ‘spirals in the cloud’, whats the cloud thing? Its big at google πŸ˜‰ The idea of this being a WEB BASED font editor is that we can do collaboration. In GWF we email ZIP files of fonts around, that’s not ideal. It could be like Docs, with people sat around the world, working together to improve fonts, live. This will initially be useful for people to send in extra glyphs, and many people want GWF fonts to have more weights.

So here’s the Ubuntu font loaded into the editor, and its using Bezier splines. The goal is to take existing glyphs and new Spiral glyphs and merge them πŸ™‚

Here’s the Nanum Myeongjo font. This is all running in Chrome, and even with a large font, its very responsive.

Its implemented in G Web Toolkit. GWT is used to make a lot of front ends at G, including the GWF directory today. We can write the tool in Java, it is compiled to Javascript and runs in the browser. JS is now very competitive with native apps, even multitouch is in HTML5 now, and there’s a LOT of engineering to make the HTML5 canvas (the underlying drawing engine) really good, and today it often uses accelerated graphics cards.

The back end is java G App Engine. Many G products use this, and if you get a million users, its no problem.

And this is not just a HTML5 app; the vision is of loosely coupled blocks. Existing editors can read and write the cloud fonts, it can hook into auto-hinting tools… but the tool doesn’t tackle hinting problems.

ttfautohint is interesting as it will make autohinting useful if funded. Hinting is less important in the future, as screen hardware resolution goes up, and Mac and DirectWrite become more widespread, but it is needed today.

Q: File formats? Transform Spirals to Beziers?

A: Yeah, there’s some simple scripts that are highly optimized, I don’t have tools to convert Beziers to Spiros but it is possible with high accuracy.

Q: Can you edit typical font metadata?

A: I haven’t developed that part. I plan to open source it soon, I hoped to do that for today, but it will be soon, and its an important part of the font creation process, so if people want to add that, that is great!

Q: Whats the roadmap?

A: I want to get it out there in a usable form, so you can create a font, a total cycle. Its an open road map, driven by user interest and what people end up using it for. If you’re interested, talk to me after πŸ™‚

Dawn Shaikh (G) and Steve Matteson (Monotype): Typefaces for Mobile OS

Steve: G needed something for mobiles, that worked with the freetype rasteriser. The same vectors can look different depending on the rasteriser used, you know how Photoshop has 5 settings for rendering type, and this is similar to switching platforms and browsers.

Custom type designed for specific uses is a long time tradition. Goudy said, ‘if one type is more suitable than another, there must be some type MOST suitable’. When I start a new type, I want to consider the voice; different mobile vendors sharing a platform will have different voices.

We couldn’t be too cute or too techno. My initial sketches added little bounces to the shapes. We started with too much personality. We also tried something totally different, more techno. I went back to earlier drawings and Droid Sans grew out of this.

Studies with the UX team, to decide how many weights to have. On a phone you want a hierarchy of information. We made many weights but a small filesize footprint on devices meant that we had to cut it down to just 2 weights, and auto obliques rather than true italics.

We made a lot of mock ups, and the designers did a lot with coloring of text to get the effect of a range of weights.

We also needed a monospace for programmers. Times New Roman is small and spindly on screens, Century has thicker strokes but closed forms that aren’t great on screen, but it is thicker. Georgia is the gold standard in serif screen type.

I did sketches of the medium forms, and we then added a true italic, so it became a 4 style family. WGL-4 was the initial character set, and then it was extended to Cyrillic and Vietnamese, from 600 to 800 characters. There is a Droid CJK with 40,000 characters. One of Wired’s editors wrote ‘the fonts look Gy’ so that’s good πŸ™‚

Droid Sans is intended to reach across languages. Here’s some common fonts. This Arabic appears very small, the Hebrew too heavy. For branding, it is so crucial to hold the brand together to have the typefaces all speak with one voice.

Pascal Zoghbi couldn’t make it to ATypI personally, but he did the Arabic. This is the Kufi style, and then the Naskh style. He experimented with white space on the baseline, and decided that a solid base was best for mobiles. This is the final Kufi style that is shipping in phones. The UX team in the Middle East explained that we couldn’t only have Kufi as its not preferred for reading, so we made a Naskh style; here’s a diagram from Pascal about how pen size relates to the letter forms.

We wanted a large ‘X height’-like sizing in the forms, so that users don’t have to zoom as soon as they land on a web page. Here’s a diagram of the features that marry the Latin and Arabic letter forms.

Dawn: I work with Chrome OS as a user experience researcher, focused on fonts and on-screen reading. I wasn’t involved in this project until it was brought to the international stage. It became a full time project for me, to determine when the fonts were ‘good enough’ to ship. I said ‘its good enough when 100% of the most vocal stakeholders agree it is good enough’ – sometimes there’s a huge group, sometimes its just 3 people.

I had to work really fast! For researchers its not the way we like to work, but Arabic was done first, but the project was soon to launch, and we had no tools built (as Kevin Larson’s group has used) so it was all qualitative testing. I know only a little Arabic, Spanish and Hindi so it was challenging: users were on 4 continents, across timezones…

All the methods were ‘stakeholder reviews’. We did 1:1 interviews with native speakers, who were mostly Grs who spoke the language and loved typography and wanted to make the fonts better. Sometimes I got external participants, even flying to NYC as there is much cultural diversity there. We would get the fonts onto phones to test with.

When we worked with Thai users, the 1:1 interviews told us that the font was perfect, but no one liked it. In a focus group, once one person started critiquing, we were there for 3 hours. So culturally, what works varies.

And finally, we did 100% email reviews by the most committed stakeholders. I assembled key questions from my previous work, and took a group of 40 stakeholders. Some wrote 5 page documents! 36 did 4 reviews across the whole process which was great! If the group is committed, the email reviews are fast and effective.

Over the last year this Review process is one I feel works for any kind of font: Monotype sends several font samples, with fonts that really work and are good, with a few that are less good, and I get feedback from stakeholders and send that to Steve’s team at Monotype. They design the fonts, and I get more feedback. That process is repeated throughout the creation of the fonts. As it becomes a final design, we try for 100% stakeholder sign off.

Lessons? Being involved early is best. Committed stakeholders are key. If I go to external stakeholders, it takes a lot longer and costs real cash, so I try to keep stakeholders happy and engaged. It is an iterative process, and lots of communication from the rough idea is key.

Eben Q: To what extend is the feedback in this process available to other type designers?

Dawn A: We work so fast, I didn’t consider making it public. It surely would be valuable, and we learned a lot across many languages. I’ll see πŸ™‚

Q: Steve, how did you use this feedback?

Steve A: I took it with a grain of salt at first, eg with one script, a comment was like ‘Its good, but it looks like a short fat man.’ Well, what to do with that? Is ‘it’ the font? ‘it’ a specific glyph? Ambiguous feedback was tricky. And some feedback you can see is subjective; as a type designer you have an intuition about whats useful, and some other comments were very true and good feedback. It depends on my familiarity with the script, and how specific the feedback is. Like a lot of things, as we work in other regions of the world, its harder to find experts. These scripts haven’t been designed for mobiles before, so how to bring the script to Chrome and the new age is often an open question.

Frank Steitiya (URW): Fonts for a Globalizing World

[missed most of this]

URW’s non latin fonts are almost all made by native readers, and our own tools make it possible to them to do so. Nimbus Sans Global πŸ™‚

Frank Blokland (DTL): Parametrized Type Design

This has overlap with my talk tomorrow. Today I want to talk about parameterised type design.

A large part of the design can be automated and WILL be automated. we need resources and a good defitinition of what happens. a grapheme is a graphic represenation of spoken language. movable type in te 15thC, letter shapes written in italic were fied and the invetion a few centries early, we would today use gothic type, texturas πŸ™‚

What are we dealing witH? when writing, making graphemes, you must udnerstand the system. this is my diagram, starting with a script, then the writing system, then graphemes, there are grpaheme subseystms like capital, lowercase, cursive, and then each has a harmonic/propertional/relational/rhytmic system subparts.

first, harmonic system complrises of 1 harmonic models. yo know v and w and k and x y and z and s are from a different system, from captials, and combied with other letters that are not related. it looks harmonised but its not. that you need kerning pairs shows they are not prfectly combined.

second, a proportional system. in a rotunda system you see this clearly. there is a reltion betweeen horizontal and vertical shapes.

third, relational system, contrast of pen, and this effects stroke width. x height and pen width are also related. same contrast but lower x height makes the type heavier in weight.

4th, the rhytmic system, the relation of stroke thickness and interletter spacing. i think garamond and jenson already had a letter spacing system.

5, formalisation. taking the pen strokes and making them into formal type convetnions, and, 6

6, the idiom. you can make basic stuff on the computer and you make specific and unique things on that design. like rembrand did in the past, they had others make the base canvas and they made the FINAL stuff and signed it.

type design is that simple!

how much can the computer do for you? oyou go out, take a beer, come home, finish it, and sign it off

programming is expensive though, so its 10-20 years away, but i tihnk we can do this πŸ™‚

My LetterModel is similar to sketches by Edawrad Johnson on a blackboard in 1930.

A skeleton shape, applying a broadnib to it? Totally wrong. here’s this with 15′ and 30′ nibs. 15′ isn’t “right”, the arch doesnt connect to the stem. Its that the broadnib itself defines the skeleton shape! those 2 degrees create different skeleonts.

Capitals can be done this way… you can have a model on which you apply the pen effect and it works okay for caps. the skeleon form, the latin model is BASED on the broadnib.

the first roman type was based on written letterforms, and you can see the move from writing to roman type. this is a geomtrics miniscule. you can say yuo dont like it, but its readable, and 100% what you get from a broadnib.

the first step in formalisation is to replace the feet with triangular serifs. you cant find this in renaissance times. then ‘split button serifs’ are added at the top. next, you see the jenson from 1470, and here’s a book from that period, the wide, round letters have more space. the ‘e’ bar has become horizontal, some letters are a little more compressed. this is first seen in griffo/manutious’ hypnomachia (?) and these are the proportions that most people consider standard.

We can use the LetterModel to place over the Paragon Romaine by Van de Keere, based on Garamond, and see how these shapes relate to the model. If you are using the same elements, you can let teh computer do this work πŸ™‚ If you change the ‘a’ then you know the change should be upated in the ‘e’

I showed this to a famous dutch type design. he said he never worked in that way. but this can lead to new type πŸ™‚

I also found relations between letters; the ‘m’ width and cap height. the golden section rectangle. ‘n’ square.

Also i found grids in Jensons work.

I found spacing is quite simple, since you can find the rhym based on space ‘a’ from center of letter ‘n’ to x bbox, and look for that unit in the letter, and its from left side of left stem of ‘n’ to left side of right stem. If you have a big screen, sure, but these guys worked at 40pt and they didn’t see the details as we do, they saw overall rhythm. so you can space type without looking; this is the fencepost rythm, no optical correction, and then a little bit, like ‘ib’ should have a kern. 3rd one, i refined a little more, and this is not optical. finally there is the slimback adobe jenson. This usggests to me that the casters in the 16/17thC used a system of standardised widths.

Cpatials are totally different. Spacing is based on the ‘H”. This is DTL Haarlem. HHHOHHH.

But you have to combine them with the lowercase. can do them on the ‘mmmmm’ grid (where stems are aligned)

So I wondered if you can make a spacing system of this? roman type is related to Jenson’s type in some way, so can you measure your tyepface and used a fixed system to space it? This is SO simple, my 84 year old mother can understand it πŸ™‚ This is the sapcing I did in 10 minutes. We will make a LetterFitter program to do this in a second πŸ™‚ A lot of what we consider ‘optical’ can be explained in this way.

Finally, idiom. this is the HARD part. you can get the computer to do a lot of things, but this is the hard part. perhaps you can even encode a designer’s idiom, ‘unger type’ or ‘X type’. if you cant describe it, it doesnt exist πŸ˜‰ if you are a lecturer, and you dont describe what is happening, its not good teaching.

so we have some tools to implement this system, and eventually we will get a complete design system. and thats it πŸ™‚

Adam Twardoch (Polish Mafia): OpenType Mastering

[5 mins late]

Mac Terminal defaults dont match the icon πŸ˜‰

Theres few tools; FontLab, Fontogrpaher, and terminal tools like TTX.

We can get rid of the terminal, and replace it with the DTL OTMaster developed by DTL and URW++. A useful, I found, companion to font production workflws. FontLab studio, Fontographer, UFO tools, others. DTL tool is useful. Windows and Mac versions are available. Its a GUI tool. We have 2 fonts open here, a tree of the OT tables, and we can take a look. we can see the glyphs, dive into the tables like glyph table and see what glyphs are built of. A good place to start is to work side by side with a browser pointing to the OT spec on and the glyf table, you have various flags and for them the detailed explanation of what is in sight can be seen on the OT spec. when i tried to read the OT spec, i couldnt make sense of it. but this is a way to VISUALISE it. you can read the spec, see what is in a font, and it starts to make more sense. so you can learn omre about opentype and how it works inside. i recommend familiarising yourself with the OT spec with the OTMaster tool on a high quality font. The NAME table is useufl.

OT layout tables, you can see what is inside. Kerning can exist in 2 ways, GPOS as a kern feature, or as a TrueType ‘kern’ table that is ‘old style’. these can coexist as different apps may not use the GPOS kerning. if it has both, the old kern table will work as a fallback (but its limited in size)

Mark attachemnt previews, and glyph viewer so you can see the glyphs, and also previewing rendering with gridfitting (truetype hinting) and greyscaling with FreeType.

OTF consistency checker, also checks TTF, and this is a great lint tool. I found a problem in a font I made 2 days ago with this.

The simple workflow is to turn a VFN into an OTF with FontLab. You might have a UFO imported to FLS and into OTF. What if you make chagnes? Easy way is to change the original UFO and do these compilation steps. But these steps from UFO to VFB to OTF are probably not automatic. You may tweak some aspects, use other tools like OTF to VOLT to OTF again. Maybe you forget the UFO and make chages directly to the hinted VFB?

A useful solution is to make small changes that do not effect the design directly to the previously generated OTF. You might even take that OTF back into a VFB or even UFO. Since you can at any step in the workflow, you can even look the workflow. VFB -> OTF -> VFB -> OTF … And this can save a lot of time.

Fontographer is a design oriented tool, opposite to OTMaster, which has no design tools but allows detailed modifications. FLS is in the middle, with features similar to OTMaster and also it offers design tools.


XITSMath.otf in OTMaster. A math font with a large number of glyphs, similar to the new MS Office Math engine, the OpenType MATH table, a MS extension to OT.

There is a kind of glyph editor, with layers, remove overlaps, etc. I can import a new glyph as an EPS.

XITS is a libre (OFL) font so we can make such changes.

I recently mastered the Lato fonts, as we’re preparing a Pro release. When I opened my final TTF, I checked it in the ‘Consistency Checker’ and i saw that the extended latin glyps that were made for the font but then removed for the OFL version, still had the ‘latin range’ flags set even though the glyphs were now missing. Similarly, the OT GPOS and GSUB features, I could export the FEA file from OTMaster, then import it into the OFL Lato, and I can see that it has filtered the Pro FEA file to just the features that are supported by the glyphs in the font. So I can have a monster FEA file to cover everything, and make a Latin and Latin-Extended font and OTMaster has this intelligence to filter the OT features to the characters extant in the font.

Pablo Q: Why did you have class kerning?

Adam: That wil work as well. the syntax is, well, it looks different, but that is class kerning.

Q: can you type FEA code in OTMaster?

Adam: You use a text editor and import, mostly. DTL OTmaster is available from with Fontographer and Fontographer.

Q: How to maintain a sane source file?

Adam: What I do, PS VFB, TT VFB, then XY hinting, then autohint, then best versions

Dave Q: you mentioned ttfautohint, could you explain that?

Adam: its freetype, it works well for bolder faces, for lighter faces we use fontlab autohinting, and if your stems are thin, you should be more careful. if the stems are thick, and there TTFAutohint is nice. Werner is accepting donations, and I think its a good idea. FontLab Autohinter is good but not always the best choice. ttfautohint can be better, in some cases its worse. check the output πŸ™‚ These 3 tools, FLS, OTMaster and ttfautohint are all good tools.

Dawn Shaikh, Marc Tobias Kunisch (G): The subtle & peculiar lessons we learned from G Web Fonts users

[10 mins late]

Dawn: Users wanted ‘try before you buy’ and they wanted to easily compare fonts side by side. they would come up with lots of workarounds. they wanted suggestions about what good fonts are. if the suggestion is too subltle they miss it and if to blatant they resent it. they want a beautiful website to showcase the fonts. they knew examples. they wanted good organization. fast and intuitive filtering, not too complciated and not too simple. Users wanted powerful search, and excellent rendering on all platforms, browsers. They wanted an easy way to download the fonts. An easy way to call multiple web fonts at one time.

There were things users didnt care about; detailsed stats on font usge was liked by type designers but users didnt care. popularity and newly added were good for users. designer inforamtion wasnt so good, only real type geeks but most users ignored it or found it distracting. our sample was US centric, and they didnt care about extended latin. font licenses, users didnt look too closely unless they were an org concerned about lawsuits.

G, these are libre fonts, and users were concerned about poor quality, missing characters, not sure if it was curated.

Users didnt like tagging much, and didnt understand who puts the tags there. Is it old fashioned to me or my grandma? One user said it was noisy.

Most people tried G Web Fonts as its free, and they stck around because the fonts are good an deasy to implement. many users had trial accounts they could rattle off, and they would bounce around web font services. users were not sure how to determine the value of web fonts. its unclear what their ROI is and how they can measure it. they didnt know how to pass the expense on to clients. they had concerns about recurring expenses. if they pay $100 this year, will it be $150 next year? If they give a company their CC details, they dont want to be auotmatically billed.

Users are more willing to pay for fonts they can downolad. Its valuable to use it on the web and in any other ways.

There were mixed views about the pay/font vs pay/subscription models. 50/50, some users felt TypeKit style model was best, mostly designers, where they want to change their mind of use a few fonts, and that pay per font was limiting. but then develoeprs thought they would pick the font, not change it, and paying a one off fee was fine. designers wanted to give their clients options. so tahts the initial research we did, not Toby will talk about turning this into action.

Toby: I was in charge of the relaunch. First, we did a workshop with everyone, engineers, type people, UX people, and a facilitator, and spent a half day wriing ideas on postits and posting them on a wall, then half a day arranging them, then into a spradhseet and raing them in terms of impact and feasability. we made a large diagram with X for impact and Y for feasabilty. so the top right are things with impact that are possible to do:

Font colelction, future proof UI, custom text preview, font selection mechanimc, advanced font specimn, organize by headline and body copy, and font combinations. those were the targets. then we did a ‘design sprint’ although it was more of a concept sprint. lots of paper prototyping and whiteboards, this was for a week, and the last few days we were met with the product manager and engineers, so we could discuss what to do.

this is the general idea from that week, ‘choose review use’ and then the current version of the site. while creating the site we used ‘RITE’ method, a game design team at Microsoft came up with it (its on wikipedia) and it was very rewarding to see the changes working better and better as you go πŸ™‚

We also did task based tesing, asking users to find what font was used in a graphic (lato:100) and it could be tough when we came up with a clever idea and then users didnt see it or guess how it worked.

For suggested usage, we found that people who knew about type got cross with suggestions that a font was only for text use, and so we just made teh fonts display in 2 modes, headline and text, and have teh default the recommended use.

The filters, people think of a ‘font’ as what is actually a ‘font fmaily’. so selecting styles within families was tricky. we had users add families and then at the final step ask which styles they actually want from teh families.

We launched, and you can see the satisfaction survey bar charts now.

Q: users dont care about tags?

Dawn: they want to SEE the fonts. if you ask them to saearch for a wedding font, they will on any site, filter a little, and then scroll the fonts. the tags, any sites that had them, i had to ask users what tags are, users always skipped over them.

Frederik Berlaen (Dutch Mafia): The missing UFO editor

I live and work in Gent as a father, designer, teacher and programmer. I’m here to talk about the missing UFO editor, RoboFont.

To introduce myself, I did Kallicultator, a final project for T&M in Den Haag, it generates contrast around a skeleton, its not finished, and FontConstructor is type design from another direction, elements that can be pieced together. A playful tool to explain how glyphs are mades from similar parts. Its a gratis program.

RoundedUFO is a tool initiated by Erik van Blokland, and UFOs have round corners.

UFOStretch is a tool for making eg smallcaps

RoboFont will be released later today, I hope: its a nice story.

Erik said, Robothon 2009, “The tools you choose, influence your creative process”

This encourages me to make better tools and make better skills in programming to make better tools.

What is it? RoboFont is a UFO editor. UFO is defined at, its a folder with XML files, you have the font as text you can edit in many apps. There is prepolator, superpolator, metrics machine, area 51, a few tools from me. apps that do one thing, bound together by the UFO format.

One thing missing! AFter Robothon 2009, there was no app to generate a UFO from scratch. These apps all use UFO as their native format. They do one task only. They only correct contours, do OT features, or generate a binary file. They do one thing, but very well.

Petr van Blokland messaged me on iChat, thats how it started.

RoboFont is to manage your design process, draw black and white shapes, edit UFO data, and scrpiting is key. its important to my design process. its really a platform, a second desktop for your type work. I think your desk is good, your phsyical desk is different if you are doing a headline type or a revival and so should your editor from project to project.

What will it not do? No knowledge; i don’t like when an app designs for me. It doesn’t know naything about anything, I think designers should make the decisons, and the power to design their fonts. It contains no hinting, so far, many users ask about this but its a spcialist thing and I think its ok to leave that to experts. you can do scripting, there are deep hooks to the data object, so you can embed a hinting tool πŸ™‚ it does have basic kerning too. Kerning is a mastering job to me, not drawing; you may do some metrics but not the whole thing.


Its Lion! Start a new project, you have a charset template (no ‘encodings’ files, so 90s…) and you have a Features text editor, a font info metadata similar to Area 51, a space cetner for metrics, and a kerning tool that wiill launch metrics machine if you have it, or a very basic metrics UI if you dont. Since there’s no encoding, you can sort the font table’s sorting and so on. There’s also a simple toolbar icon to launch in any UFO tools.

There’s a boolean finder view, it accepts regular expressions, and you can save these ‘sets’. there’s a table view of the metrics and glyph data.

Then there’s the GLYPH view; you can select points, multiple handles at the same time, undo obviously, there’s a pen bezier tool, you can measure very easily (and this is important for my design process) and these rulers stay on the canvas.

Remove Overlap is tough, the Adobe FDK tool is a mess, and i worked wth Tal to make a new Remove Overlap. Thanks Tal!

We have layers, its UFOv2.

You can drag and drop in a raster image, and apply basic scaling/rotating, and contrast/etc transforms to make tracing it easier.

Space CEnter has a main text input, and then left and right text input that insert that into the main text. zoomable, drag the metrics. can double click to edit the glyph. multiline, waterfall, ‘x height cut’ that hides all ascenders and descadners, there’s a ‘beam’ style metrics you can drag and drop up and down the glyphs.

There’s a single window mode, and the preview/layer/transform tools are nicely done.

We have curve error detection that is set in preferences, so if you point is ‘off by one’ it will highlight the spline in turquoise. we can show live bitmap previews. we have guides, that can rotate, guides can show intersection measurments, you can set how magnetric each guide is. You can label points, and these labels are shown on canvas.

A lot of things came from RoboFog, the uber type design application.

You have a simple python REPL, and then there are python extensions. python scripts to run in the tool, and then ‘extensin builder’ like Sfari has, to build plugins so RoboFont can be extended.

These extensions are files and when you run them in the Finder, they instantly appear in the app. Eg a shape tool. Here’s a RoundingUFO extension that adds that functional to this app.

Erik made an arabic extension to do arabic kerning. This took a day to make!

So, that quote, really sums it up, “The tools you choose, influence your creative process”

Claus Q: How much does it cost?

A: €400. A fair price. Every time you use it, you are saving money! Its due out tonight.

Q: Encoding?

A: you just write the glyph names, and you set their unicode value.

Q: can it autodetect the unicode value from the name?

A: no, thats knowledge. you can name it any way you want. I dont like the adobe short names, they dont make sense to me, and i cant remember numbers.

Q: Can you implement kalliculator as a plugin?

A: Sure, there’s nothing hidden or magic, you can see everything and write any plugins.

Ted Harrison (FontLab): Protect your fonts with EPAR

EPAR is about education of the user. Background and definitoins.

EEULA, started 6-7 years ago, embedding font license ifnormation inside the font to make it accessible to users, so they know there is al icense and they can easily figure out what they are allowed and are not allwed to do, and act approprietate. we hope that invvled UPGRADING their font license.

EPAR is the implementation of this idea as a SFNT table. It took a while to figure out what the fields should be, and we managed to try it out. the first thing with Paratype was an online version, to add EPAR to an OpenType font.

We then got this into Fontographer 5.1, released a few months ago. Demo!

Go Element, Font Info, click Advanced, and then Licensing, and you see all the EPAR fields. Each field can be ‘N/A’ or “Limited” or “unlimited”.

[DC: There’s lots of nasty restrictions on time, number of uses, format conversions, places it can be used, embedding, modifications, sharing, license transfer, compression, bundling – and telling users URLs for the license, an upgrade url, and there’s a password to hopelessly try and restrict people from changing these DRM flags. This is a pretty comprehensive list of ways fonts can restrict people, and this would be a pretty beastly DRM system if anyone ever implemented it…]

When I open a font with an EPAR in Fog5.1, it opens a popup modal display that disappears on click that explains the EPAR restrictions in english. there’s also “a nag screen” that is displayed when users open a glyph editing window and modifications are not allowed.

Next steps?

EPAR table import/expoert so you can populate an EPAR with your standard license once. a ‘portable feature’ for your whole catalog

Font Agent Pro, used by big corporate cusotmers who want to track their font licenses, and when EPAR is integreated, it means they can easily see what they are allowed to do. Eg in a corporate merger, they can easily see how many licenses they need for all their new employees.

We will make a DLL so all font aware applications will be font license aware. we’re making a Read-EPAR DLL, as part of the Font Agent project, and anyone can license that like Adobe licenses the FDK.

Finally, the original EEULAA idea of ‘font personalisation’ where private information about the licensee is inserted into fonts. THis is scheduled for the next version of EPAR.

We have a EULA SIG, tomorrow at 2pm.

Q: when users upgrade, how do they all get the EPAR upgarde in their files?

A: the EPAR table is portable. the last field, the EULA part has a ‘remote’ URL where you keep their EPAR table. so you can sit the importable EPAR table at a URL, and every time they use the font, their system will check the URL and if that URL is updated, their files will be updated. you just change your EPAR file.

Q: Would you be better talking to font and OS vendors? what if the user doesnt use font manager software?

Ted: we hope to, but its one step at a time.

Adam: if a few font managers adopt it, then a few OS people might adopt it.

Atilla Korap (Monotype): You said you want a revolution…

Jobs said there is a beatles song for any moment πŸ˜‰

Revolution is a big word, the French revolution, a political revolution about their society. Then there are revolutions of the mind, paradigm shifts, and technical revolutions. those seem more interesting to me – the industrial revolution with steam engines and machine factories. typesetting was part of that.

the latest revolution is the digital revolution. we have digitized typefaces for a long time, and now we are able to provide web fonts for all browsers.

this has a long history, and i looked from the start, HTML1, who were involved and what tools. CSS1 gave you control over typography, but only supported system fonts.

CSS2 introduced this, it was removed in 2.1 and put back in CSS3.

Neither TrueDoc (PFR) or EOT took off by 2001, and everyone was worse off. “You can’t always get what you want” is the stones πŸ˜‰

Web fonts was a lost cause, no one asked for them or were offered.

Then out of nowhere in 2007, WebKit announced that it would support @font-face with regular TrueType fonts. No conversion, encryption or compression. The positive reactions from users and web designers, it became a ‘thing’ and after that a lot happened fast.

The type “industry” reacted to this, and there’s a big discussion in the W3C forums, and someone suggested reviving EOT. Browser developers and W3C wanted an open standard, EOT was proprietary, and so some people suggested URL binding is DRM. So Ascender suggest ‘eot lite’ without all that, but this also wasn’t appreciated. the discussion was tough and went up and down, ideas of ‘partial fonts’ that browsers could assemble, or obfuscate OpenType fonts so that they can’t be installed, a compressed font format ZOT, and finally, Erik and Tal made WOFF with Jonathan Kew. A solution to all the questions.

Daniel Rhatigan, Monotype: Web fonts

Summary: Type designers are working hard to make web typography look good, but its really up to web publishers to ensure what people see is good.

Web fonts is a shifting landscape. customers expect vendors to take care of eerything. i made a document for customers to understand using type on the web. or on screen.

not all browsers support all formats, not all vendors support all formars….

rendering is a key issue. people fall into a full continuum from old school typographers to print people to newbies. often people think of fonts as clean outlines. but there’s a software rendering environemtn and then hardware too that translate those typeforms.

“Its rough out there”, and we have a responsibilty to you to make good type, and you have a resposiblity to your audience to make it great for people.

Type looks different on macs and windows. I show a side by side screenshot of the web font awards mac sfarai and windows 7 firefox, and then on MacOS safarai, iPad safari, and iphone 4 sfari, you can see there is a lot of difference.

Type choice is a key thing for web fonts. Most fonts were designed for other purposes! all software issues aside, there are classic book faces not intended for this coarse environment. Since the DTP revolution its been easy to choose many fonts but its not always true they are great choices.

First thing people have to grapple with on the web, after a basic set of core fonts with similar proportions, is vertial proportions. The internal proportions mean the same point size between helvetica, goudy old style (small) and caecila (big) varies optically.

This also effects the interline spacing; this is all set the same size and set solid, and frutiger, gill sans, garamond and neue helvetica are all very different amounts of text and feels.

you see that designers and developers are at odds about this issue; because they were not anticipating the current environment.

Eg this New Yorker page using ‘MAD MEN’ and the A bar is aliased badly. but its okay in this page layout.

frutiger vs helvetica, frutiger has open apertures and loose spacing. a lot of work has gone into making both work well on screen, but the basic design of frutiger means it works better on screen. most people dont look at this level of detail but it matters.

Choose families wisely. you must look at the linked styles within the family, the italic and bold may be too narrow or too dark on screen. this means specifying work is harder, to ensure the reader has the best experience getting through it. Our big multiweight families are selling well because you can choose which weight you want to get the right pixel density.

Moving beyond type choice, the designers and developers setting the text is the next thing.

Setting type in relative units is okay with verdana and georgia, but mixing helvetica with gill sans or bembo that have a small x height, you have to bump up the x height so they look similar, but in proportional units this will throw off the line height. so you want to specifiy px units to get a predictable outcome.

If in doubt, be more generous with space and size means things are a little easier to read.

Everyone is racing to make web fonts look good, hinting existing fonts to look better, revising outlines and spacing to look clearer at small sizes, designing new fonts for screen display.

Speak up and tell us what you want!

I’ve made some test pages at and you can email me at dan (a) about them

Christopher Slye: CFF on the web

OTF is diffferent to TTF, I’m not going to smack on TTF, but OTF does have benefits for web typography. TTF is huge, OTF is much smaller, and WOFF-TTF is smaller than that, but WOFF-OTF is even smaller!

Simple idea:

TT: dumb rasteriser, smart fonts.

OT: smart rasteriser, dumb fonts.

TT fonts have INSTRUCTIONS. OT have HINTS. “Hinting” vs hinting. Accurate, its instructions, and this is important distinction.

Type2 charstring format docs say ‘hint are just hint, recommendations, for an intelligent rasteriser’

Can TT instructions behave more like hints to accomodate a wider variety of screens?

Well, thats what CAREFULLY HINTED CFF fonts look like in modern browsers. basic TT conversions look good too.

Adobe Garamond is a classic from 1986, and this has very even stems. This is MSIE6 with CFF converted to TT and then EOT.

Does the evolution of screen technology to justify a large investment in TT developemnt? I hear a lot of anxiety about TT hinting.

Does the hinting of most TrueType web fonts really look that good? If you look at the most popular fonts on FontSquirrel, they are not hinted well. They might be better with CFF.

Some personal theories I want to throw out; TT procedure hitns means its more vulnerable to hacks and exploits; its a programming langauge. Some peopel worry about it, although I don’t know of examples. CFF Native efficency is less work for the client than simple compression.

Conclusion, consider both CFF and TT and choose the one that works best for that situation, and support the improvement of CFF renderers.

Chris Lilly Q: Where are we with those CFF rasterizers improving? is it just starting, or winding down?

A: People should care about it more, and there are all these devices. a lot of embedded hardware can do a lot better with CFF.

Tom P: Apple dont care about TT hints, Mac OS and iOS renders them the same! And DirectWrite is now nice with CFF, and thats the future for Microsoft platofrms.

Q: what about old IE versions?

A: CFF in EOT will work in MSIE8 i heard! Im not sure about MSIE7. FreeType CFF rendering isnt so great, and its something adobe would like to see improved.

Vlad Q: to build a good rasteriser, is it publicly specd?

A: I think its there publicly and adobe would help.

Victor Gaultney: ScriptSource

If you are a type designer asked to extend your type to an unfamiliar script?

Many places are out there, but some conflict and may not be correct.

Where first? πŸ™‚

A huge amount of 42 million results. what are some familiar ones?

Omniglot, a static site, quite limited.

Wikipedia, a big one, with lots of oddly linked pages. it seems comprehensive, but may be inconsistent. today there’s 3 pages listing what languages use Cyrillics, and they dont agree at the moment

say you want to know what glyphs to support for the languages your users are using? unicode site doesnt tell you what languages need which characters, and you might not be able to afford them all.

unicode gives you static data, useful but limited, and not a guide to design.

decodeunicode is a great site on specific glyphs, and contains a LOT of stuff. but its focused on characters.

theres the diacritics project website.

typophile, a great community forum with many wise designers and technical experts, sharing a lot of expertise, but while there are some wiki pages, most of the know how is locked up in old forum threads.

what you want is something that pulls the best bits of all these sites.

I work for SIL International is a not for profit, faith based charity, serving langauge communities around the world, doing literacy, linguistic research, transltaion and software – including fonts. And I do sessional lecturing at UoReading.

The above didn’t meet our needs, so we made ScriptSource. This is designed to be a dynamic, collaborative website for meeting the needs of type designers dealing with the writing systems of the world.

Eg, look at Beria, an african script used by 169,000 people. eg, punjabi written in the gurmukhi script, what glyphs are needed. we can drill down to a character. we can see its unicode STATUS – ScriptScource can contain things not yet in unicode.

Its also useful for latin designers; Latin Extended-A is there.

There’s a “Add to ScriptScource” button on all pages, which allows anyone to contribute – submissions are moderated!

You can link to 3rd party sources, you can add to entries with various media, and you can register needs that are not yet met.

scripts + languages = writing systems, which lead to character sets, which are parts of scripts. this model is authoratitive. Ethnologue, ISO 639-3 for languages, ISO 15924 for scripts, Unicode and CLDR (common locale data repository, an industry wide, vetted source of locale information, run by unicode consortium) and the SIL PUA.

Roadmap: Comments, discussions, topics, FEEDS and APIS! please talk to me about that! πŸ™‚

The lac of organized informatin is a problem for our communityy, and we hope you will conttribute your own amazing knowledge to this resource for the world πŸ™‚

lorp Q: state of API?

A: RESTful JSON πŸ™‚

Dave Q: what is the license for the core data?

A: it depends on the license used by the information providers. we dont impose a license.

Q: who else funds this?

A: 6 people over last few years at SIL did this. our long term vision is to get it run by an academic consortium, and there was interest by they wanted to see it concrete.

Georg Seifert (German Mafia): Smart Font Creation App for Designers

I am a type designer, Graublau Sans and Slab, and Olivegreen Mono; I love extensive and playful cursive forms. Rosa Stencil. Fraktur Rosenrot, unfinished project to match the latin evolution to blackletter; I recently did a typeface and pictograms for the Berlin airport! πŸ™‚

Glyphs is my font editor. I want to HELP the designer as much as possible, and hide a lot of the technical stuff that we have to deal with these days. I encourage users to give up control to gain speed. Its about DESIGNING. Removing technical stuff.

I added keyboard shortcuts for many things, and I have a nice ‘reconnect splines’ tool to add back overlaps that are usually removed.

I have Petr van Blokland’s idea of moving bezier handles incrementally.

I have a ‘cursive’ way of slanting that isn’t a simple slant.

I have a diacritic ‘cloud’ preview

the glyph table is versatile, with coloring, export flags

There is python script support, eg this ‘beowolferize’ script.

It has small caps autokerning

This is Jan Gerner’s arabic, released at FontShop. the anchor positions work similarly to the latin diacritic clouds.

there is a Mutliple Master style interpolation. set up masters, then set up instances, and you can see different instances in the glyph editing window via the layers palette. There is a cats cradle view of the masters stacked, so you can see that each point is interpolating correclty.

There are timestamped/named layer versions, a poor designers versioning system. There’s also a dimensions notepad to remind you what stem distances are.

there’s a measuring tool

Q: PC versions?

A: No, its based on Cocao so no chance.

Q: Price?

A: €239 via paypal, $299 on Apple App Store

Q: no support for TT outlines?

A: no, its so rare that designers use them. i generate OTFs and they can be converted elsewhere. I see that as post production.

Q: do you do anything if you are distributing OTFs?

A: I ship to customers what Glyphs exports (its wrapping adobe FDK makeotf)

Evening Event

John Berry: Thanks to all our local hosts. This space is great! ATypI has brought type people together since 1957 and larger events recently have been great. 57 was metal type and early phototype. The business has changed a lot, people on the bus now chat about fonts, and there’s a lot of ignorance too. ATypI is educational. many kinds of writing systems that need new fonts, and making things all work together in a global system of communication. We’re also here for our own education and to enjoy where we are. I’d like to introduce the President of Iceland.

H.E. Olafur Ragnar Grimsson: Thanks. Welcome to this extraordinary building, one of the largest artworks in the world that happens to have a concert hall inside. A fantastic location, where design, inspiration, nature and mathematics is on display. An old school friend is a mathematical and perhaps design brain behind the artwork. Its true that even if you are a great artist, its difficult to work alone. People have to come together to create an extraordinary piece of art. This is the first time I’ve been asked to make a speech about a single letter. Some speculation is building about if I will run again for President. Delivering such a speech, may fuel such speculation, because for respect for the alphabet I should not step down until I am finished. πŸ˜‰ I’m not sure if a head of state has delivered a speech, dedicated to a single letter. Our country has enjoyed a good design community lately, in design, textile, industrial and graphic – we have seen extraordinary progress. A young generation taking a strong interest. We’ve also seen type design become a fascinating field for our designers. Greek temples and roman coins to iPad and Kindle. … Your conference brings a new aspect to this arena, there is a prevalance of English, so characters not used in English are something of a curiosity. But they are a key piece of linguistic and national identity. The languages of the alphabet of small communities need to fight for their place. In the year 2000 we wanted to celebrate the 2,000 year old discovery

Microsoft refused to make Windows available to users less than 1,000,000 speakers. I wrote to Microsoft to say it would be a pity if Windows didn’t support a language used in North America 2,000 years ago. It demonstrated, not language imperialism, but insensitivity to small communities with different traditions, who have to fight for a right to use their own language in a world dominated by globalisation. We see an umlout, then we think it is German, Hungarian, Swedish, Danish… there are a limited number of nations who could claim that letter. When we see a german S we know its german. But an eth and a thorn, these are unique letters, and belong only to the old icelandic language. we are proud we have these letters. it reminds us that we share it with the anglo saxon and the old english heritage. its not solely icelandic, and we are still part of the culture and alphabet from the nordic countries, across iceland and greenland to america and into russia too, 1,000 years ago. when the vikings replaced their ruins with the roman alphabet.

for a society to keep exclusive and beautiful letters like eth and thorn, the language is the foundation of the nation, can be a mixed blessing. traveling around the world, when we use a computer in a hotel room, the keyboard doesnt recognise the existance of these letters. when no one understands what we are talking about, we talk to people back home in a way that seems silly, with our letters missing. in global work, it can be awkward for officials and interpreters to deal with these letters that they ay have never seen before. a meet highly educated people in top positions can’t understand why we use such ‘old’ and ‘outdated’ letters.

but its not just a part of our language. its a contribution to our humanity’s choir of languages. i hope you will broaden coverage of your type designs to this language. we can’t take modern progress to old letters; ask your host to bring an icelandic computer dictionary. we translated computer concepts, where most languages use the english term. these letters are integral to our modern computer usage. i hope you have a fascinating time, that this building inspires you, and you will not just stay in this box but go outside and

The process of creation is still going on, and here the forces of nature – earthwquake, nrew islands, glacial floods, volcanoes – you get a sense from early youth that the environemtn is in flux and creation is new. many nations have the book of genesis’ idea of creation that the almighty made the world in 6 days and rested because the work was finished, is a strange notion. here in iceland, we know, that not true πŸ˜‰ the process of creation is still going on, and the almighty is still having a fantastic time continuing the creation.

Gunnlaugur SE Briem: Keynote presentation

John Berry: This talk has no useful information. He is from iceland although he lives in the USA now.

Gunnlaugur: Here are 2 observations that are not original: life is wonderful, and we are doomed. Love and cherish while we can. People who work with the alphabet see the world in a narrow keyhole. The Vinland map is fine, but $300,000 is a lot. This roman headstone is fine ,but we look at the spacing of the R. In ‘Utopia’, family size is strictly regulated. we are intereste din peple with a written culture. mesopotimia gets a metion for inventing writing, and had strong mail men πŸ˜‰ Egyptians invented a fine alphabet. The heiroglyph for writing became phenetician, and eventually the M.

Mayans also had signs for writing. Xian missionaries nearly burned every book, a few examples survived.

There are greek ideas of mathematical proportions that are pretty, they dont always match nature as they are famous for.

Only 4 or 5 manuscripts survive from ancient rome. The trajan column was made after resource stripping what is now romania. htey mined gold πŸ˜‰

The roman empire had a rich tradition. early cursives attempted to write captials fast, and the D became d, the ‘f’ has a top serif because its an entry stroke. The tiber valley brought material for roman buildings.

I use Font Studio on Mac os 9

line length, the hypnomachia has a line length of 60 chars. gutenberg has 2 narrow cols that need a lot of hyphens.

computers often dont place the right quote marks, or hyphens are not case sensitive.

there are books with rules of typography, the chicago manual of style, the elements of style,

This entry was posted in Knowledge. Bookmark the permalink. Post a comment or leave a trackback: Trackback URL.

Post a Comment

Your email is never published nor shared. Required fields are marked *

To create code blocks or other preformatted text, indent by four spaces:

    This will be displayed in a monospaced font. The first four 
    spaces will be stripped off, but all other whitespace
    will be preserved.
    Markdown is turned off in code blocks:
     [This is not a link](

To create not a block, but an inline code span, use backticks:

Here is some inline `code`.

For more help see

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>


  • What we do

    Understanding Fonts is a type design training business. If you'd like an event in your college or city, let Dave know: