How extreme do you make your light and bold masters

Adam Ladd

I'm wondering others experience on this:

Of course, every typeface can have different needs, but for something like a relatively standard sans serif family with multiple weights, I'm curious how extreme of a difference in stem width/weight you use for light and bold masters (I call them light and bold just for distinction in the working file, you may call them something else).

I know some may use 3 or 4 masters (light, regular, bold, etc.) but I've so far found that I can interpolate all intermediate weights pretty well from just having a light and bold (plus, in Glyphs, a few brace layers for /e /a /s as needed) and being careful about point placement.

I draw my Light master somewhere in the 20-30 stem width range and my Bold somewhere in the 180-200 stem width range so that it doesn't take much to extrapolate a Thin and Black. The awkward part is that this is a pretty extreme visual difference in weight to switch back and forth between as I'm drawing all my glyphs. I try to avoid having a Regular master to keep things a little simpler, but I have to rely on previews of the intermediate weights as I go along.

Setting the stem widths for the masters listed above, I can generally add or subtract 10 - 20 units for extrapolated weights like Thin and Black with pretty solid results. Beyond that though and the points can get a little wild.

What's your setup look like: 2 masters? More masters? What stem widths do you work from for your extreme masters?

Vasil Stanev

No cutting corners for me. As light as it goes, as fat as it can carry, until glyphs like $, Dcroat,a, g etc. break down. Obviously according to the Impallari formula. I like to give the clients options. With sans, in practice the bolder weights are more sought after.

In the book weight, I usually make the vertical thickness of the lc, in points, the same as the horizontal thickness of the UC. Then I work out the other weights, compensating and discreetly cheating where there is need. Repeat for italics, synchronize the design, check, fine-tune, finish.
3 masters for better control. hairline, book, ultra. After I got the first two, I generate the third and retouch it, then I go back to synchronize. Boldening effects are also my friend for the grunt work.

Hrant Հրանդ Փափազեան Papazian

Since 2000, I make the Regular, then the extremes (generally by hand), then I interpolate a new Regular and tweak the extremes to make the interpolated Regular as close as feasible to the original, then I dump (well, archive) the original. It sounds extreme (pardon the pun) but has nicely set me up for the var-font age by helping me know how to design good extremes only.

{I just realized I misread the title of this post. Sorry.}

Adam Ladd

My "Light" master is more like an ExtraLight or almost Thin weight as I draw it, and the "Bold" is more like an ExtraBold or Black weight... so they're pretty extreme. Just doing a lot of previewing of mid-weight instances is needed to make sure my points are placed well in the extremes to generate good results for all the weights.

A Regular master would be helpful, for me I've found not all glyphs need the control/tweaks via a Regular, so the brace layer in Glyphs has worked well for the few glyphs that do need some tweaking.

Craig Eliason

For a sans serif I like making the Light master monoline, 10-unit strokes. Often I can get most of the way there drawing a skeletal shape with a single open contour and then Offset Curve with Make Stroke checked to turn it into a proper contour. The Bold master is more-or-less as bold as I can make it for busy glyphs like Vasil mentioned. I find it useful for really understanding my design to have one master that is entirely about skeleton and another that is mostly about areas of color.

The thinnest instances I generate are usually a bit thicker than the Light master, so they pick up a subtle bit of modulation/optical correction. 

Max Phillips

Doesn't it depend on what sort of type you're drawing? For most sanses, slabs, and other relatively monoline faces, I think it can be useful to try to go right down to a hairline skeleton for the lightest weight. For garaldes, didones, frakturs, and other genres that depend to some extent on contrast between thick and thin, it might be less helpful to go significantly below text weight.

One thing I've started trying with sanses is drawing a 'regular' weight, then deriving a hairline weight from the centerlines of the regular's strokes, and then, when I've got those more or less where I want them, extrapolating a rough bold. Then I take this extrapolated bold, which looks like pure hell, and sculpt a proper bold out of it, as if each of the wonky raw glyphs were a lump of clay. But the extrapolation does automatically fill all the glyph slots with something approximately the right size and shape to work on, and with the node structure, anchors, rough kerning, and (very) rough sidebearings already in place. Now and then you even get a glyph where the raw extrapolation actually looks pretty good and just needs polishing. And studying the way the extrapolations go wrong can help me understand the proportions of my regular and hairline better.

I'm about to publish an eight-weight grotesque I did this way, and this method got me to some bold weights I'm happy with a whole lot faster than expected.

Michael Jarboe

It really depends on the design as you've stated, but in the case you’ve mentioned I use three masters which include Regular and then maximized extremes in either direction. If needed, intermediate masters with only key glyphs can be inserted for greater contrast adjustment.

I like to draw all masters at the most extreme points, avoiding extrapolation since I’d rather have full control throughout the entire process. It's really helpful and more efficient to plan and make these decisions before getting too far. Working through and cleaning up automation of any kind is more unpleasent to me than any extra work required to plan out and carefully draw all extremes together from the start. I'll experiment with various extrapolations if the design is in an relatively early, rough stage though.

The book Size-specific adjustments to type designs has a section discussing masters. I've always loved the quote by Christian Schwartz stating “You have designed enough masters when you can cover all of the zones without any of them looking wrong – too clumsy, too brittle, too tight.”

Hrant Հրանդ Փափազեան Papazian

To actually answer the question :-/ to me extreme weights are great for display fonts, but are against the nature of text fonts, throwing off users in terms of usability. So for the latter I go very mellow; my bolds look like demis. This is something Otl Aicher professed as well.

Nick Shinn

I use the same method as Max.

The beauty of extrapolating between Hairline and Regular is that one gets a degree of precise subtlety in the Extra Light and Light weights that is very hard to attain by just drawing those weights directly.

Thomas Phinney

Any time I extrapolate, I instantiate the result as a new master so I can tweak it directly.

The only exception is in newfangled workflows without corner masters... but these are tricky to get right.

Adam Ladd

One thing I've started trying with sanses is drawing a 'regular' weight, then deriving a hairline weight from the centerlines of the regular's strokes

@Max Phillips when you say derive a hairline from the regular, do you mean you're drawing a new hairline master, using the regular as a guide; or nudging/moving the points from the regular to create the hairline; or using a script/filter to generate the centerline? 

Chris Lozos

...drawing a 'regular' weight, then deriving a hairline weight from the centerlines of the regular's strokes

I don't know what Max means. For me, I just draw a single path outline in the mask layer, then stroke it.  I usually need a UPM of at least 2000 to get a cleanly stroked path.

Adam Ladd

Thanks for sharing your method @Chris Lozos, that's helpful.

Christian Thalmann

I tried to cover the entire weight spectrum of a sans with only 2.5 masters (Hairline, Bold, and bracket layers for controlling optical compensations where necessary) for my Quinoa and Eau de Garamond families. While it works pretty well for that weight range, the Bold that comes out is pretty scrawny, since it has to retain the overall morphology of the intended Regular. I ended up drawing a Black master for Quinoa as well, and might do the same for Eau if I ever get around to it. All things considered, it's probably more efficient to start with a three-master approach to begin with.

Chris Lozos

I also need a black master to overcome the corrections needed for weights where the counter is dwarfed by the mass of the form.  I feel more comfortable with hairline, regular, bold and, black masters.  For me, the corner is turned with at least 4.

Max Phillips

@Adam Ladd I'm afraid I do much of this deriving in the dread, much-hated Illustrator. Sometimes you can use the Blend tool to average the outer and inner contours of a glyph, and then scale that new path to fit your cap- or x-height. Sometimes you have to do a little surgery on the paths/contours before you can average. Sometimes you just drawn the damn thing by eye.

Chris Lozos

best way is to...

just drawn the damn thing by eye.

James Puckett

It depends some on the style of letters being drawn. When I did Antarctican Headline, a ten weight typeface, I had four masters. The lightest was a thin weight with 20 unit strokes to a cap height of 700; I wish I had done 4 unit hairlines The ultrablack weight has 250 unit strokes. So those are some pretty huge extremes. I ended up with four masters to give me control over the horizontal strokes; thin, book, bold, and ultrablack. Book was interpolated from thin and bold then cleaned up manually. Ultrablack was extrapolated from book and bold and then cleaned up manually.

With my latest release, Barteldes, horizontals are all fine lines with little variation across the design. So I was able to go from extralight to bold with only two masters.

Cory Maylett

It depends on the typeface, but I've had good luck creating lights and x-bolds, then interpolating a regular from them, which I'll tweak as needed before interpolating other in-between weights. Hairlines and ultras/blacks often require too many design deviations for them to serve as masters, so I've been mostly handling them separately.

Adam Ladd

Thanks for sharing your methods @Max Phillips. Yeah, was wondering if you had a found a "trick" that worked best or sped things up to generate the centerline, but like most things, it probably just takes time.

Adam Ladd

Appreciate that insight @James Puckett. I'm working with 2 masters right now: Light = 20 units and Bold = 185. Might extrapolate/cleanup a hairline too, we'll see. So far so good, but do need some control for the horizontal strokes also (Brace layers for a few glyphs seem to be helping so far).

For Antarctican Headline, did you space and kern those 2 masters first before creating the other 2... or waited and spaced/kerned all 4 together?

James Puckett

Adam Ladd said:

For Antarctican Headline, did you space and kern those 2 masters first before creating the other 2... or waited and spaced/kerned all 4 together?

I kerned thin and ultrablack, interpolated kerning for regular, cleaned that up, and repeated for bold.

Adam Ladd

Gotcha, thanks for sharing @James Puckett