A faithful sans version of Garamond?

Catharsis's picture

I've recently developed the urge to doodle a monolinear sans-serif rendition of Garamond that tries to stay as faithful to the character of the serif source as possible. Now, that seems like a pretty obvious thing to do, and I expected to find a number of fonts online that had pursued exactly that design concept. However, with my very limited overview of the font world, I didn't find any. I suppose Gill Sans Light comes close to that to first order, but upon closer inspection it has a lot of quirks.

Are there any obvious candidates for that kind of font that I'm not aware of, or is it a concept worth pursuing with a new font project?

For a test drive, I made a few letters in stroked path mode in FontForge. I quite like the looks of it. I've added a shot of EB Garamond below it for direct comparison:


What do you think? Is it worth pursuing?

If so, I suppose it's unhealthy to stick with the stroked-path mode, and I should switch over to outline mode as soon as possible. However, I'm having trouble with the "Expand Path" functionality of FontForge; it invariably causes the application to crash. That would considerably increase the amount of work it would cost me to achieve a nice consistent stroke strength. Do you know of any easy solutions to that problem?

Cheers

Stephen Coles's picture

Now, that seems like a pretty obvious thing to do, and I expected to find a number of fonts online that had pursued exactly that design concept. However, with my very limited overview of the font world, I didn't find any.

There are at least three: Claude Sans, Marian 1554, and Aragon Sans. That doesn’t mean there isn’t room for other interpretations, of course. You would also want to steer clear of Today Sans and Cronos, but if you haven’t seen those maybe it’s best not to look at all so you won’t be in danger of being subconsciously influenced. I think your concept of following the structure of Garamond is a good one. Just don’t trace the model too strictly — use your own ideas to make it work as a sans.

William Berkson's picture

The category of 'humanist' sans generally go in this direction by basing their lower case on the oval, rather than the circle. One of the earlier ones, inspired by Jenson rather than Garamond, is Legacy sans.

I think that one can't capture the whole look of Garamond in a sans. It's impossible first because chopping off serifs doesn't work—you have to change relative widths of characters to get a good rhythm. But the other thing is that most sans are more monoline, and old style serifs have a lot of thick-thin contrast. Particularly with old style, this means that the curves in the interior are different from the exterior. Here you have followed mainly exterior curves, though not exactly. And that means that the whole has a different feel. For example, the o and d are quite different in Garamond and your sans.

None of this is to say you can't come up with something great. It's just that you will have to decide what you want to capture of Garamond, and what you are willing to give up. And then you as you progress you will find that to have integrity and coherence in your design, it will start dictating its own rules, independently from Garamond.

I think you already have a challenge along the line of coherence in that your o is rather geometric, and the circular part of the d is not. Also your a has inevitably changed the relationship of the spaces within the bottom and top parts of the letter, and now looks a bit awkward.

hrant's picture

Worry about how it would be useful more than how to derive it.

FWIW the first such font was Syntax.

hhp

Catharsis's picture

Thanks for the replies, everyone. I don't think I'll make this my main project at the moment, but I'll keep it around in case more inspiration (or boredom) strikes.

@ Stephen: I had a look anyway, and I'm glad I did. ;o) Pretty much all of those fonts seem to take a notably different route. They generally have a certain "calligraphic" feel to me due to their modulated strokes (except for Marian, which is really more of a slab serif than a sans). Also, from what I can tell they make a lot of concessions to the common wisdom of sans design — large x-heights, for example. My impression is that a simple, truly monolinear (or at least optically so) sans that sticks to the proportions of the original Garamond is still an unfilled niche, though whether anyone has a use for that is another question entirely...

@ William: Very good points, thanks. I'm aware that I can't fully reproduce Garamond in this medium, but I'd at least like to provoke a sense of recognition — "hey, that's Garamond!" — at first sight. As for the overall curve philosophy; true, I'd have to make a decision about that soon. My gut feeling is to go for a slightly diagonalized |o| rather than a circularized |d|. It's true that I've mostly followed outer curves so far; that came from the fact that I wanted the letters to retain their original "girth". The Garamond |G| looks quite wide, and tracing its spine resulted in a narrower look (I tried it). I'm not quite sure what you mean about the |a|; its bowl is also smaller than the space above it in the original... and I rather like its current shape.

@ Hrant: The main use I envision would be for display in large sizes, or maybe for short passages at medium sizes. I think my monolinear and conservative approach would break down at heavy weights (perhaps even at Regular), so I might only produce Light and Hairline versions. Whether or not there's a market for that, I have no idea. Think there is?

@ Eliason: Thanks for the link. I especially like the Nota font mentioned in that thread. I'll put in on my list of fonts to buy once I start to make money with my own (still waiting on MyFonts to get back to me!), but its tittles feel a bit misaligned to me, so it's probably not going to end up on the top.

Catharsis's picture

Also, any suggestions on how to convert a path into an outline in FontForge, without crashing the application?

hrant's picture

Try installing the FontLab plug in. ;-)

Whether or not there's a market for that, I have no idea. Think there is?

You don't have to do it just for the market - you could do it for culture too. What I'm personally not a fan of is doing it for kicks.

hhp

Catharsis's picture

I figured out the problem with FontForge: I wasn't using the latest version... that's what you get when you offer Mac users the choice between a pre-compiled old version and a newer version that you have to compile yourself. ;o)

I've finished the lowercase letters now, just to give myself an overview of where this would be going. In all cases, I've tweaked the curves of the sans such as to give me as close an optical match to EB Garamond as I could get, rather than just tracing the inner, outer, or median curves.

I quite like the results for some letters (especially |t f a e g|), while others are downright ugly (e.g., |x|). The overall appearance is not very cohesive, though.



riccard0's picture

I'd say the results are interesting enough to pursue.

Catharsis's picture

Alright, I finished a first version (including language support and ligatures) in three weights. I'd have to adjust the spacing of some accent marks in the Bold face if this were getting serious (it's just stroke expansion for now), but for the standard characters it works. No kerning yet.

Yousa like?

Catharsis's picture

Alright, there was quite a bit of cleaning up to do. FontForge had a maddening propensity to screwing up one or two random curves in the character set when I tried to add extrema to the expanded paths. For now, I've just generated the fonts without 100% accurate extrema (the points as they come out of the expanding routine seem close enough) and without rounding to integer values (since that makes the lighter weights all wobbly). They seem to work well in PowerPoint.

I've settled on five weights so far: Hairline (5), Light (10), Regular (20), Medium (30), Bold (40). The numbers refer to stroke thickness, where 1000 is the font's total height.

I'm thinking of calling the font Eau de Garamond, since it is the spirit of Garamond distilled down to its pure, unembellished, æthereal essence. What do you think?

Ryan Maelhorn's picture

I don't think the rounded ends of the strokes work for this. Shear them off into straight lines. The e still looks a bit akward. I do see potential here though.

Catharsis's picture

I see what you mean. It's going to take a lot of manual work, though. I can use the square cap / miter bend expansion method as a starting point, but then I'll have to clean up countless intersections and clip extreme points. Ah well, I guess it's still much less work that one would normally have for a non-monolinear font. Maybe I should try to clean up the Bold and Hairline weights first and see whether the others will interpolate...?

I've made the eye of the |e| a bit larger and smoothed the curves a bit; is it less awkward now? I'm afraid a little awkwardness will have to remain if I'm to stay true to the original. I bet |e| wasn't Claude's favorite letter.

Ryan Maelhorn's picture

I see what you mean. It's going to take a lot of manual work, though. I can use the square cap / miter bend expansion method as a starting point, but then I'll have to clean up countless intersections and clip extreme points.

Welcome to being a type designer. Hey at least you don't have to carve them in steel at actual size.

Also, I wouldn't worry at all about making extra weights until you have one absolutely nailed. Take whatever weight you think is the "normal" or "regular" weight, delete all the rest. Then work on that weight until it's perfect, until there's not one thing you want to change about it. Only after that, start working on the extra weights.

Ryan Maelhorn's picture

"I've made the eye of the |e| a bit larger and smoothed the curves a bit; is it less awkward now? I'm afraid a little awkwardness will have to remain if I'm to stay true to the original. I bet |e| wasn't Claude's favorite letter."

Yes it looks better, but it's hard to tell as it is a way bigger pic. Can you render out a pic of it at the size it is in your previous pic where you showed all 5 weights?

To me the akwardness isn't in the counter per se, but in the curve in general at about the 1 o'clock position. It just looks a bit sharp, a bit cramped there.

Catharsis's picture

I've renamed the previous "Bold" weight at 40 units to "Regular", which is closer to the truth, and used that as my reference. I have the impression that it shows flaws in the curves most clearly.

How does it look?

ftp://ftp.mpia.de/pub/thalmann/eau/inventory.pdf

Catharsis's picture

Meanwhile, I have a Thin version of the same quality as that Regular. I might not go any lighter than this, unlike my original plan.

ftp://ftp.mpia.de/pub/thalmann/eau/inventory_light.pdf

I'm fine with redoing some characters in all weights if there should be some problems left in there. I just felt like having a second full-fledged weight to toy around with for now.

I really like how that |@| turned out. I might have to make an italic for this font. :)

Catharsis's picture

I seem to have reached a stable state with the letter shapes. Four weights are complete now. I think I'm going to submit them for iKerning soon — I figure any edits I'm going to make from now on will be minor tweaks. I'll also get started on the italics soon.

Character inventories for the four weights:

ftp://ftp.mpia.de/pub/thalmann/eau/

And some running text:

William Berkson's picture

Your overall letter shapes have got some nice asymmetry into them, which is working in mass, but you have some seriously wonky curves and weights that need fixing, in my opinion, before you get to kerning. For example (I am looking at regular only) the shoulder of your 'n' is thinner than the join, the opposite of normal, and I don't think looks good big. Your 'e' has some awkward flattening between 12 and 1:30, and the vertical stem is thinner than the top horizontal, which again I question. In general you are not following normal latin stress—thick vertical, thin horizontal—or old style, which is thinner in circular curves at 4:30 and 10:30, and thicker at 1:30 and 7:30. In a sans reducing these stresses is normal, but if you are going to reverse them it needs to be systematically and with a specific goal in mind, which I'm not feeling here.

eliason's picture

Particularly as your weights get heavier, you need more generous letterspacing.

Catharsis's picture

@ William: Thanks for catching that! I was actually going for true monolinear, but it seems FontForge is not being very exact about it. If I'm going to have to hand-tune all letters to have consistent weights, I might as well add in some optical correction as well (particularly at the joins, as you mention). That's going to be a lot of extra work — and yeah, I know it's part of type design, but I was hoping for a quick and easy job here. I'm also wondering whether this is simply growing over my head in terms of skill, since I barely see these flaws that stick out to you. Perhaps I should leave the subtleties of the sans to the experts and return to making flashy display fonts.

The thickness variability seems to be less of a problem to me in the light weights — perhaps I should just focus on finishing and releasing those? I do think that's where the font concept works best (see below for a demo of the Thin weight). I guess I'll still have to work on the curves though, as in the |e| that you pointed out.

@ Eliason: Yes, that's right. I was hoping I could get that done as part of the iKern process, but I might as well do some preparatory work myself.

Ryan Maelhorn's picture

This is starting to look very nice. You still need to break fully away from the 'strokiness' by which i mean it still looks very rigidly monoline.

And again I think you are putting the cart way before the horse by even thinking about multiple weights at this point.

William Berkson's picture

Christian, quick and easy aren't adjectives associated with type design! I think you will regret releasing this prematurely, as it does have some things going for it, and it's not going to look professional if you push it out the door prematurely.

It took me two years working on type constantly before my eyes learned to see what was wrong. That something was wrong I could sense from the beginning, but I couldn't see what and how to correct it, and had to just keep trying variations until I hit on something. Then one day I woke up and went down to the computer and looked at the screen, and it hit me: hey, I know how to do this! Then I could *see* what was wrong and how to correct it. I came with no art background, so it might have taken me longer than for someone with more of a background, or someone with more raw talent. But there is definitely a learning curve for the eye and brain, which ramps up pattern recognition as you keep working on it, until you get 'type eyes'.

Once you get over that hump, things are a lot faster, and some type designers are astonishingly quick. But I think Matthew Carter said that watching him design type would be like watching ice freeze in the refrigerator.

hrant's picture

I came with no art background

Not a bug, a feature.

hhp

Catharsis's picture

Thanks for your support, this is really helpful. Eau should be pure and essential, so sloppy strokes and stunted curves have no place in it. I'll keep distilling it until it lives up to its name.

@ William: I don't have any full-time years to dedicate to this — my day job is pretty demanding, and I don't have any art background either. So I'll have to learn quickly or pick another hobby with a shorter reward cycle... However, I feel I've learnt a lot in the past few months already, thanks to the extremely helpful advice and feedback on this forum, so while that trend continues, I'll stick with it and see how far it will take me.

However, I have to make a few decisions before I invest more work:

- What range of weights do I need in the end? I know I'm supposed to perfect one weight first, but the choice of that weight is tricky. Currently, I'm leaning towards focusing on very light weights, maybe even just up to the "Regular" I've shown above. So maybe I should pick the Light or even the Thin as my work weight. I don't suppose this will ever become a workhorse sans, but if it should go even halfways in that direction, it will need a few much heavier weights, and that will require completely rethinking all letter shapes as space gets crowded, which I'm currently not planning to do. Do you think there's a market for a font like this with light weights only? It seems to work for Miso at least...

- Contrast: I agree that the letters should not have counter-intuitive stroke stresses. However, how much contrast is actually needed? Is it OK to aim for monolinear and just lighten up the junctions a bit for optical correction, or am I required to some contrast even in unbroken curves, such as the |o|?

In the meantime, I'll focus my efforts on the |e| and maybe also the |a| of the Regular and try to tune them to "professional" quality. Hopefully that will teach me enough to do the same thing to the other letters.

hrant's picture

Why rush a "hobby"?

I know I'm supposed to perfect one weight first

I don't see it that way - the various weight inform each other.

Number of weights: I like four.

hhp

William Berkson's picture

double post

William Berkson's picture

On polishing this. First, I would read the short monograph The Stroke, by Gerrit Noordzij, which is originally in Dutch (I see you are Dutch), about the history of scripts. And then try doing Garamond with a broad marker. It's impossible, because type departs from natural strokes, but trying will give you the idea of the basis of old style letters, and what the eye expects. It will give you an idea of how type typically departs also. When you design something, you are playing against that expectation of broad pen 'rules'. Eg if you don't follow the bolder down-stroke, lighter up-stroke rule, and make them all uniform parts of the W will look fat. You will see where the joins are automatically thinned and thick, etc. You can break the rules, but if you do it should be systematically, for a purpose.

A second thing is to familiarize yourself with the optical illusions that are always at play in type design. A good summary is here, but you need to nagivate to type design/basics/illusions.

These two will only take a few hours, and will up your game very quickly. The second thing is, to get an idea of professional quality curves, compare your glyphs to those of say Frutiger's self-named face or his Avenir. He is a real master, and remember you are competing with him! When I say compare, I mean paste one over the other and visa versa in Fontlab.

Hrant is right that different weights will inform one another. But if you are eager to get a first version out, then I would suggest just getting the regular very polished. You can always do a 2.0 version later...

ps I don't know about Font Forge, but I suspect it is an obstacle. If you try one of the other font editors out there, it may help.

Ryan Maelhorn's picture

I don't seem to recall anybody suggesting you make your strokes sloppy, lol. They should be more varied however. Check out Avant Garde, a monoline font done right. At first glance one thinks, "yes this is all monoline, all the strokes on everypart of it are the same width exactly." But then, as one looks deeper, you find this is not really the case. Also, check out Caspari, which throws its weight around deliciously.

Ryan Maelhorn's picture

Perhaps I should leave the subtleties of the sans to the experts and return to making flashy display fonts.

If you're really serious about that, I'd be glad to continue this on for you. I can't promise you'll love the results as taste is an individual thing. I also can't promise you it will be done in a short amount of time. But that said, if you're serious, I'd love to tear my teeth into this thing.

Catharsis's picture

@ Ryan: "I don't seem to recall anybody suggesting you make your strokes sloppy, lol." — Misunderstanding here. My current lines are sloppy, so they clash with the font's concept. I'm all for adding some stroke modulation to clean things up.

I've never liked Avant Garde, but more from an overall shape perspective. It sure does look extremely monolinear. Caspari is very pretty, looks like the next evolutionary step from Thesis. (Actually, on second thought, its |a| is too narrow. Pity.)

Catharsis's picture

@ William: Thanks for your suggestions! I'll look around for The Stroke. I'm actually Swiss, only temporarily living in the Netherlands, so I'll go for a translation... As for understanding the old-style stress pattern, I think I do have that one down; after all, I made a sort-of oldstyle font from scratch (Octant). It's just that I didn't think Eau would want to have any stress to begin with.

I'm aware of that "illusions" website. It's very illuminating, though it mostly concerns itself with serif fonts. For instance, do caps still need to be heavier than lowercase in this case? Do you still need overshoot for round bottoms on the base line if your stems don't have any serifs (and therefore look rather flimsy themselves)? Looking at my Eau example above, I tend to think yes, my |o| is slightly too small.

Catharsis's picture

@ Ryan again: "If you're really serious about that, I'd be glad to continue this on for you." —
Good to know! I'll keep that in mind. I'll give it a try myself first, though. I don't yet know where my skill ceiling is. I'd certainly prefer to finish this under my own power if possible.

William Berkson's picture

Yes, all of the illusions apply to sans serif fonts. When I first saw this section of Briem, I was skeptical. Now when I look it's all crashingly obvious that it's correct. That's what I mean by the eyes learning, which I experienced. What you will experience before your eyes are tuned to seeing the o as too small etc. is that when you do what he says the font will seem more polished and 'right'.

As far as stress, yes if you want it to look monoline, and that means not actually being monoline, because of the illusions at joins, that verticals seem lighter than horizontals the same thickness, etc.

One of the things about type design that still blows me away is that everything has to be different by mathematical measurement than the way it looks to the eye. *Every* character is affected. For example, I couldn't believe I had to spend time 'tuning' the plus sign so that the horizontal and vertical look even, the same.

Personally, I hate avant garde. The caps used with all the ligatures can work well, but the lower case is just ugly, with its too-wide c and e. But I know when it comes to 'ugly' it's a matter of taste.

Ryan Maelhorn's picture

Catharsis, I would tend to agree that the 'caps must be heavier than lowercase' ideal is a bit overstated, certainly when it comes to a pure monoline font. Why go through all that trouble to make all the strokes in the lowercase 'appear' to be all the same weight, and then distract away from that effect with a thicker upper case?

However, you definitely need overshoot in pretty much every font, regardless of if it's a sans or a serif. The only exception I can think to this rule would be fonts based on old quartz displays. Think of the "fonts" that were built into VCRs and digital watches. However this is probably mostly a technological restraint, they would have had to use more crystals to do any kind of overshoot.

William Berkson's picture

Caps being heavier than lower case is a complicated subject. Partly, you just need to make them a little heavier because they have bigger internal spaces. So to at least match the 'color' (grey density) of the lower case, they need to be a little heavier. Old style fonts had caps a lot heavier, and with more space between them and the lower case than between lower case characters.

But whether to really overweight them is a design decision, not one of basic craft. The argument for that is that the caps are a different alphabet, and they are used as markers of beginning of sentences, etc., so having them heavy they serve more as markers. On the other hand, they disrupt the color and flow of the text more. In sans, the tendency these days seems to be to make the cap height relatively low, and keep the weight of them more matching. Personally I'm not crazy about this, but it's really arguable either way, and depends on the design.

Catharsis's picture

I've tried tuning my |n o e| for a start. My letters have grown a bit larger (and thus lighter) due to overshoot, and I noticed smoothing out the curves tended to regularize them a bit too much, so I had to put the |e|'s characteristic rightward trend back in afterwards... I like the effect overall, though it still feels extremely subtle to me at this point.

According to Briem, I'm supposed to make the |o| heavier than the stems of the |n|, but I couldn't really do that without hurting my eyes. My impression is that only makes sense when you have a high-contrast |o| where the stroke really tapers away from the middle instantly, which is not the case in a sans like this.

Is this going in the right direction, and how far do I still have to go on these letters...? Just to get an idea of the magnitude of the work ahead of me.

Big/small:


hrant's picture

No, curves do have to be thicker than straights to look equal. Your "o" there is clearly lighter (partly because it's so wide - which I think is actually good for readability).

hhp

Ryan Maelhorn's picture

It sure does look extremely monolinear.

Ahhh, but you see, that is the whole point! It only "looks" monolinear!


AvantGarde Book. All units in em.

hrant's picture

It's still a horrid excuse of a font.

hhp

Catharsis's picture

@ Hrant: It doesn't look "clearly lighter" to me at all... sure, there's a big counter, but as far as the lines go, I just don't see it. I might just have hit my skill ceiling.

Then again:

@ Ryan: Nice example! I particularly find it interesting that the curves in |g| are not heavier (in fact, they're lighter on the left side!) than the bars in |t|... is that just because |g| is a dense letter, or does the same thing happen in |o|?

William Berkson's picture

"horrid excuse of a font." :)

But the caps, when used with the alternates can work together. I forget who said that the only place Avant Garde ever looked good was in the nameplate on the magazine. The font was expanded from that. Jeremy Mickel did an impressive and clever job of creating Avant Garde style alternates in huge variety, so that every three letter monogram could be constructed with overlapping letters, like in the original name plate of the magazine. It's called Sobriquet He actually improved the horrid C and G by having everything thin and carrying the strokes less far. It still only works with overlaps.

I definitely disagree on Christian's latest o above. It's too wide. The o an n should have interior counters of about equal area. Otherwise you get uneven color. Christian, look at the *counters* in Garamond. It's those more than the exterior outline that give it its look. The problems with your earlier e didn't relate to its asymmetrical shape, which was good. It was the bumpy interior border of the arch at the top, and having the vertical arch thinner than the horizontal arch.

About thickening the vertical arch of the o compared to the horizontal arch. Classically, the vertical arch is fatter than the vertical on the n. But that's really a design decision. However, I think you're just wrong about it working only in a high-contrast font. Look at Myriad. I just checked the o. Vertical arch 89, horizontal arch 66. The vertical is 35% fatter, but it doesn't look high contrast!

Also *always* look at your letters in context of other letters in your font, if you don't already.

Ryan Maelhorn's picture

It's still a horrid excuse of a font.

Ludicrous.

William Berkson's picture

Ryan, hatred for Avant Garde is pretty widespread among designers. It's not just Hrant and me. I know a lot of people love it. But they tend to be young and, er-um, inexperienced. It doesn't wear well.

hrant's picture

The o an n should have interior counters of about equal area. Otherwise you get uneven color.

Which is the wellspring of readability. :-) The equality you speak of I considering valid only in the realm of display type.

On the other hand if this is supposed to be based on Garamond then that's a strong design constraint too.

hhp

Ryan Maelhorn's picture

This is the o, Catharsis:

Ryan Maelhorn's picture

Ryan, hatred for Avant Garde is pretty widespread among designers.

Typedesigners maybe. Avant Garde Bold is being used in advertising everywhere right now, everywhere. It's like the current Helvetica, which I'm sure almost all typedesigners hate too, but they would be short sighted for that.

PabloImpallari's picture

Catharsis, this looks very interesting. Keep working on it!

The only thing I will like to add, is that all the weights are still a little bit light.
For a quick reference, this are the stem weights that I use (more or less) as guidelines:
Light: About 40 or 50 units
Regular: About 70 or 80 units
Bold: About 120 or 130 units

The need to compensate the rounded stems is not so apparent on the lighter weights, like the ones you are working on right now. But they will become much more evident as your font gets bolder.

Ryan Maelhorn's picture

Partly, you just need to make them a little heavier because they have bigger internal spaces.

Well said.

Catharsis's picture

@ Hrant, Ryan: The |o| of Garamond does look wide to me. It might have to do with the suprisingly generous spacing. Here's a comparison of Eau and EB Garamond. I seem to have increased the ratio of intra-letter to inter-letter space, but the |o| doesn't strike me as out of place. As Hrant mentioned, faithful recreation of Garamond's character is my main objective.

@ Pablo: I know my weights are very light. The reason is that I don't see this concept working anymore once the lines get so bold they have to squeeze to fit into the x-height. I didn't intend this as a book font. I wouldn't mind being proven wrong, though.

Syndicate content Syndicate content