Dear Sarah

Christian Robertson's picture

I put this study together this afternoon. It was good to get out the ink and work on paper for once. I've been meaning to play with cursive contextual alternates for a little while. The good news: no kerning! The bad news: getting all the connectors just right. I'm not sure if this works yet; I'm still against handwriting fonts in general, but they're definitely cooler with some alternates mixed in, and with the connectors done right. I think to do it right I would need to mix in a bunch more alternates, but it's a start, and I like where it's going so far. The letters are pretty low res; my scsi scanner won't work with the powerbook, so I resorted to the digicam : (

application/pdfcontextual alternates
dear_sarah.pdf (40.7 k)

kris's picture

I think it is one of the better handwriting
types I have seen! The contextuals work
well, I hate seeing two consecutive glyphs
that look the same in a supposedly random
type. You are missing a dot on your 'j' I think.


eomine's picture

Looks pretty good indeed.
What about making the curves softer? (Just a thought).

Christian Robertson's picture

You're right that the resolution is pretty low. I think I will have to go in and add some more points. In the mean time, I've added even more 'splotchiness'. I added some ink blots in an OpenType feature. I also added one set of caps. (I'll be adding a bunch of alternates for the caps too).

The trick is setting up the OpenType stuff to add as much randomness as possible. Right now I've only got five different ink blots, which appear fairly frequently. I think if I add a bunch more, then make each of them less frequent, the effect will be better.

application/pdfCaps/Ink Blots
dear_sarah_caps.pdf (23.2 k)

eomine's picture

Great stuff, Christian.

ideagent's picture

Yes. Great stuff!

Christian: In addition to the alternate caps, it would be nice if there was an alternate swash "d". It may disrupt your randomness because it is so prominent.

Christian Robertson's picture

More stuff. I'm up to around 400 characters so far. This sample doesn't have everything, but it's a bunch. I didn't intend to go this far with this one. I was just working on ideas on how to work the connectors with opentype. At any rate, I've manually added a little bit more resolution to most of the letters. It will take quite a bit more before I'm happy with the digitization. I need to get a scanner (I recently gave my old scsii beast to a thrift store). My 3MP CoolPix isn't cutting it here. It still doesn't seem random enough. I am working out some more fancy OT stuff to randomize the letters. For example, the ll ligature seems to repeat too often. I might mix some of the alternate l's in there somehow. The ff combination also seems repetitive. Maybe it's just becuase I've been looking at it too long. The reality is that the font will probably never be used for more than one sentence anyway, now that I think about it.

One question for the OT techies out there: at what point to OT instructions start to bog down the software. I would like to do some contextual stuff to make sure that no two letters are ever the same within three characters or so. It will take a lot of instructions to really get it right. I haven't noticed any slow down yet, but it seems like if it had to parse a couple hundred lines of code for every character it would really start to bog down the host software.

application/pdfMore work
small_cap.pdf (65.6 k)

iota's picture

Very nice.

I wish i had handwriting like that. :-(

Grant Hutchinson's picture

Hey Christian. This is rocking so hard, it hurts.

The calligraphic doodly bits are especially nice. Have you given any more thought to the contextual spatter and splotchiness features that we chatted about?

Christian Robertson's picture

Tiff: I have been trying to work out a double f solution. I'm not quite satisfied with what I've got. You are one step ahead of me on the non-connected characters. I'm working on a set of leading strokes that can attach to the first letter in a word, as well as a set of characters sans connecter. There are certain characters that work well without a following connector: an occasional x or t--s's can be nice too. If I do get around to adding the unconnected series, I'll tie the left connecterless to the right connecterless with OT.

I have been thinking about some more ink spatter and I've settled on using "spatter alternates." Since the current implementation of OT doesn't allow for inserted characters, I'll have to attach the spatters to different characters. What I'm doing so far is placing the spatters so that they will end up to the far left or right of the spatter alternate (or even show up on another line). This way, the spatters will be connected to random characters. The trick is planning the spatters so they look good where they happen to land.

Of course, I'm still adding more alternate caps. I've got three different SC sets. I'm starting to resign myself to the fact that I can't get the randomness perfect with OT. The typographer will have to add a few alternates here and there to polish it off, but at least they'll have plenty to choose from. (See the small caps in the last sample: I've added alternates by hand.)

aquatoad's picture

It's looking pretty darn random. I just came up with a theory. It seems like the glyphs that suffer the most for lack of randomization are the ascenders/descenders. The l jumped out straight away.

Be careful of making your outlines too detailed. I don't think they are bad now. Maybe a little smoother so it's not quite so toothy when it's 400pt. But I wouldn't give the outlines any more nodes.

A set of fancy signoffs could be fun, Sincerely, With Deepest Affection, Best Regards, Love, etc. How bout a digitally remastered version? Only kidding :-)

You are pushing me to experiment with OT. Rock solid work!


pablohoney77's picture

hey christian, mind if i ask a question? how are you using OT to generate randomness? i've been trying to figger it out for a bit and haven't had any luck so far. mind sharing what yer doing with the rest of us?

pablohoney77's picture

hey thnx for sharing that christian, i've been tryin to generate my own "randomness," but i don't quite have a firm grasp on OT programming yet. i think looking at this might help me out, actually. keep up the good work, it's lookin beautiful!

johnbutler's picture

Very, very impressive. Christian, it is so cool to go back through these forums and watch a brand new high-quality type library unfold inside less than five years. You are truly talented. Your designs just keep getting better and better.

I worked with Erik van Blokland (Letterror) a few years ago on something similar for the OT implementation of his Kosmik design. Every glyph (including accents and punctuation!) had three variants. We overloaded the Contextual Alternates feature to simulate ideal flipping behavior as best we could. But as you widen the "search radius" the feature code grows geometrically in size. We learned yet another way to crash InDesign.

I am hoping that the good folks at Adobe and MS will do the necessary programming to add true flipping capability to their layout engines and support the RAND feature, described here in the OT spec. Then Contextual Alternates can be freed up to do its real job in fonts (like yours) where you do have truly contextual alternates like your loopy d. ("pendant")

It would not hurt for a font developer like yourself to drop a line to the Powers That Be at Adobe and second these requests to the InDesign/CoolType programming teams. Especially with one of these PDF samples of your font in action.

hrant's picture

Pretty impressive! A superbly relevant use of OpenType.

The question is:
What is the real difference between the first and third paragraphs?


kris's picture

Looks very good Christian. Excellent work. The speciman design is also very fitting.


pablohoney77's picture

stunningly sumptuous.
i can only hope my digitization of my Granny's handwriting is half as beautiful.

johnbutler's picture

This Sarah is one lucky gal!

hrant's picture

Thanks for the comparison. I actually did a Mad magazine style double fold deal on the previous PDF to get a good look.

I must say I think the swash alternate scheme seems more worthwhile than the contextual one, although the former would still make a big difference for display setting. But for display setting I think you need the outlines less coarse?

Anyway, I don't want to seam nasty: this is really great stuff!


hankzane's picture

Christian, I hope you don't get it the wrong way, but. I think the highlight of your script typeface are the inkblots and the absent copyright sign (at least in the sample). Do you mind if I borrow them?

hrant's picture

Sergej, check out "Ornaments Ink And Pen" here:


neuroman's picture

WOW! That RAND Feature is pretty interesting...the only thing that bothers me is that I don't have any clue how to implement it...I'm using FontLab (as you can assume)

fontosaurus's picture

Oh my god. Planning on selling this? Where, when and how much? I'd buy a copy.

hankzane's picture

Hrant, I don't know how to open the files once I unpack them on a Windows system. Any judging by the screenshot the inkblots look rather phony, so it's not a big deal. Christian's inkblots look as if they weren't supposed to be made; they look real. Having said that I don't remember last time I saw any real inkblots, so perhaps I should keep my mouth shut.

hrant's picture

BTW, I just discovered that the ink blots font on that "A4" web site is lifted directly from an old metal font. Not too surprising, considering many of the other fonts on that site are lifted too...

I found out by casually browsing through a long-ignored book I have of vintage type specimens: "130 Alphabets & other signs, #1", J Rothenstein and M Gooding eds. A couple of other "interesting" precedents, shall we say, revealed in that book can be found on pages 105 and 136...


jefferson's picture

What a gorgeous typeface. This is several cuts above the various script faces I've seen. Sign me up as a future customer. Gorgeous work.

Grant Hutchinson's picture

Dan Bailey:

aquatoad's picture

Ready, Steady, Go!

Congratulations Christian! Another triumph.

Grant: Surely the Divergently sample can do justice to all the heartburn that went into programming a smart v-e substitution. Especially in the Pro version! BTW, Stock photos on cake: Someone get into the sauce during the long Canadian winter? So random it's excellent.


Grant Hutchinson's picture

> Grant: Surely the Divergently sample can do justice to all the heartburn...

I know, I know. It pains me to see that as well. Due to a tight product schedule, we weren't able to get a custom display up for Dear Sarah as we intended to. However, I did attach Christian's marvelous PDF specimen to the page. The main display will get updated shortly to better represent this face.

eric_shew's picture

I'm truly sorry to raise a technical issue in this forum, but I'm frustrated and don't know where to go.

I installed Dear Sarah because I wanted to see the chaos built into the typeface at work. I'm using a PC that's brand new (less than two weeks old), Illustrator and Indesign CS, and Suitcase 9.2 - yet none of the open type features seem to work at all. When I type it out, I get no variations in letterforms.

If anyone has any ideas as to why this is happening please email me. Thank you very much.

marcox's picture

Eric, maybe this will help:

In InDesign, go to the fly-out menu in the Characer palette. Look for the OpenType option, and turn on things like Contextual Alternates, Titling Alternates, Swash, etc.

Becks19's picture

Great work on this font Christian.

I'm hoping that someone can help me. I've purchased the OpenType version of the font from Veer and I'm having trouble with it in Photoshop CS.

All of the options in the type menu are greyed out for Dear Sarah, while Adobe Caslon Pro has the ability to check/uncheck ligatures, ordinals, etc.

Please help...

paul d hunt's picture

have you contacted veer directly? from my experience they're very helpful in solving issues such as this.

Stephen Coles's picture

Really nice, as usual, Christian. The spacing is remarkably even
except maybe between the two 'l's in "village".

There are some spots where things get pretty dark: 'e' in "let",
the smaller 's'. That could be considered splotchy, but I think it
works well if your intention is to mimic handwriting as close as possible.

Eduardo is right about the curves. Digitizing can make some
pretty harsh sharp angles at points where it should be smooth.

Miss Tiffany's picture

I can't tell, but will you add an 'a' with a longer entry stroke and an 'a' without any entry stroke? Perhaps to offer a little more random disconnectedness?

This is really very nice.

Where's the double ff lig? hehe.

Miss Tiffany's picture

Great news, Christian. This is really growing on me, the more I look at it. And hearing how much work you are putting into ... and taking advantage of OT. Cool! What about a zillion different ampersands?

Christian Robertson's picture

It's not true randomness. The OT code looks something like this:

@random_set_1 = [a d g j l m n o c];
@random_set_2 = [k e i n p q t v z];
@letter = [a b c d e f g h i j k l m n o p q r s t u v w x y z];
@letter_alt = [a.alt b.alt c.alt d.alt e.alt f.alt g.alt h.alt i.alt j.alt k.alt l.alt m.alt n.alt o.alt p.alt q.alt r.alt s.alt t.alt u.alt v.alt w.alt x.alt y.alt z.alt];

feature calt {
sub @random_set_1 @letter @letter' by @letter_alt;
sub @random_set_2 @letter @letter' by @letter_alt2;
} calt;

If you put in enough of these ugly hacks it starts to appear random. You would also want to make sure that no two letters were ever the same right next to each other. There are only so many of those that happen regularly in English, so I have added specific substitution rules for each of those cases (ee ff gg ll mm nn oo, etc.) Also notice that I have placed an @letter between the @random character. This makes the pattern even more difficult to spot. You might notice the funky h after every t, but you wouldn't notice the funky h after t-anycharacter, (unless you were freaking crazy).

In the case of this font, the fact that there are a different set of characters to connect to the o b v w also helps the font appear more random. The randomness feature above needs to be adjusted to get the connectors right for this one, which makes it more complicated. To be honest, I haven't completely figured it out.

Christian Robertson's picture

Thanks, John. I wondered how you guys had done the OT flipper. I guess I was right in thinking there was a limit to how many swaps InD could make. Cool that you guys actually found it.

Christian Robertson's picture

Ok. I've been working on this one again. I reworked the caps, and added a series of non-connecting characters. I'm up to 502 characters now, but I'm getting closer. Here is the most recent pdf:

dear_sarah_34.pdf (186.8 k)

Christian Robertson's picture

Hrant: The differences between the first and the third paragraphs are subtle. Here is another pdf with the text side by side. The left has the alternates enabled, while the right does not.

dear_sarah_alternates.pdf (43.8 k)

Miss Tiffany's picture

Just received the VEER catalogue. Congratulations Christian.

a few days late ... but happy for you nonetheless.

Syndicate content Syndicate content