Widths, weights and appropriate naming

Daniel Veneklaas

I was wondering about larger type families with a multitude of widths and weights, and how people go about naming them. What happens with the apparent weight across widths is that condescended styles end up looking heavier and extended styles end up looking lighter when using the same mathematical weight. I assume this optical illusion is due to the changing size of counters.

So I see two options, either weights are optically adjusted across widths to have a consistent weight, or the weight is kept the same and the name is changed accordingly based on its optical weight. 

The other issue is that designing with very narrow display styles you have to increase the point size to for it to look like the same size as the regular widths. So taking this into account do you name the ultra condensed styles with this scaled application in mind? What’s everyone’s thoughts on this? 

Kent Lew

either weights are optically adjusted across widths to have a consistent weight,

This is what I will typically do.

Thomas Phinney

Kent Lew said:

either weights are optically adjusted across widths to have a consistent weight,

This is what I will typically do.

I think that is the norm in professional type design. You can see this in just about any family that has sufficient range of styles and is well crafted.

Ben Blom

Below is a table of the uprights of my Aspira family. The width of left stem of the “m” in the sixth row (Bold) is, from left to right (UPM = 3333): 402, 412, 422, 432, 442, 452, 463.

AbrahamLee

@Ben Blom: Why UPM=3333? Just curious.

Chris Lozos

I had the same question for Ben.  I always use a factor of 8

Jasper de Waard

AbrahamLee said:

@Ben Blom: Why UPM=3333? Just curious.

Not really a complete explanation, I suppose, but: http://www.durotype.com/Info/3idx_Precision.htm

The number 3333 still seems a little arbitrary to me. 

Chris Lozos

That tells you about precision but not about rounding errors. Would not 3333 guarantee rounding errors every time? Why not 3200?

Ben Blom

See here for the reason why I use a UPM of 3333.

Daniel Veneklaas

I read that when Univers was first released the strokes in the regular weight across widths were identical, Frutiger saying it was more a "matter of logic than harmony".

Another question that arose is the purpose for having different widths be optically equal weight in the first place. I can't think of many instances where different widths are used together in situations where it's required to appear "the same". When different widths and weights are used in typography it's usually for the sake of hierarchy where contrast is key.

It doesn't seem uncommon that variations that come after an initial typeface release such as Futura Condensed have much to do with the original in regards to weight. Is it more important to logically arrange a family or to design the weight most appropriate for the width and it's intended use?

@Ben Blom Just out of curiosity if you don't mind me asking, how much of Aspira was drawn from scratch and how much was interpolated? 

Mark Simonson

Weight is a description of type color, not stroke thickness. Normally, the meaning of the weight names shouldn't change depending on width in a family.

Theunis de Jong

I am sure to have seen a solid reasoning a while ago on why weights should not increase linearly, but rather exponentially?

AbrahamLee

It’s because weights differ more, percentage-wise, at the light end than the heavy end. Choosing a more exponential-ish progression tends to make the relative change from one weight to its neighbors more optically different, regardless of the weight. However, that makes the lighter end appear to differ less and makes the change in weight more obvious on the heavy end. I like to prefer a progression that starts more linear and ends more exponential in order to make the weight changes more distinct across all weights.

Ultimately, trust your eyes. They are the best judge of a good progression of weights.

Ben Blom

Just out of curiosity if you don't mind me asking, how much of Aspira was drawn from scratch and how much was interpolated?

See here.

Ben Blom

Abraham Lee: Ultimately, trust your eyes. They are the best judge of a good progression of weights.

Well... yes and no. Only use your eyes when u are deciding which weight progression logic to choose. Once you have decided which logic or system for such a progression you like best, just use that system for all your future font families. I simply do not believe that the design of a font, is relevant for the choice of its weight progression logic.

A long time ago, before I created my first font family, I experimented with Lucas De Groot’s weight progression logic. I didn’t like the outcome. Then I tried various weighted averages of the results of a De-Groot-progression, and the results of a linear progression. With this, I used my eyes as judge to choose a progression logic. In the end, I chose a 50%/50% weighted average of the results of De Groot, and the results of a linear progression. This corresponds to the yellow line in the first chart of this post.

Since then, I reuse the chosen weight progression logic for all my new font families. So I don’t reinvent the wheel for something I decided in the past.

[Deleted User]

The user and all related content has been deleted.

PabloImpallari

There is at least one standard naming scheme than you can follow, and that CSS coders and Browsers are supposed to be familiar, at: https://www.w3.org/TR/css-fonts-3/#propdef-font-weight and https://www.w3.org/TR/css-fonts-3/#propdef-font-stretch

I do not like it very much, but for theoretical "easy of use" I've renamed one of my families in the past.

My original naming scheme (that has the nicety that each subfamily folder sorts alphabetical when browsing the files in the hard disc) can be seen at http://www.impallari.com/testing/encode/index.php

The final release, following the CSS spec's can be seen at https://fonts.google.com/?query=encode

But as they say, "The nice thing about standards is that there are so many of them to choose from."

Thomas Phinney

Ben Blom said:

I simply do not believe that the design of a font, is relevant for the choice of its weight progression logic.

I am quite sure I believe that the design of a font is relevant for the choice of its weight progression logic.

In particular, the “weight progression” is being defined solely in terms of the vertical stem thickness. However, I believe that (1) internal white space and (2) horizontal stem thickness both contribute to the perceived boldness, and that these may progress differently for different typefaces as weight increases.

 So I would not *necessarily* always use the exact same weight progression across all typefaces.

Ben Blom

Thomas Phinney: I am quite sure I believe that the design of a font is relevant for the choice of its weight progression logic. 

To support this, you say:

Thomas Phinney: In particular, the “weight progression” is being defined solely in terms of the vertical stem thickness.

By whom? For sure not by me. “Weight progression logic” as I use it, is the same as “interpolation logic”. And when interpolating, for instance, a thin and a heavy “m”, all aspects of those two m’s are being interpolated—not only their vertical stem thickness. For the calculation of linear interpolation values, the vertical stem thickness of the thin and a heavy “m” is not being used. For the calculation of “De Groot” interpolation values, the vertical stem thickness of the thin and a heavy “m” is being used. But again, when using the interpolation values which result from such a calculation, every aspect of the involved glyphs is being interpolated—not only their vertical stem thickness.

So, when interpolating a thin and a heavy “m”, the design of these two m’s determines what is being interpolated. (If both the thin and heavy “m” would have the same internal white space and the same horizontal stem thickness, then these aspects would be the same in all interpolated versions of that “m”. But really, then the designer of that thin and heavy “m”, would be a very incompetent designer.)

Thomas, I am afraid that your reasoning to conclude that the design of a font is relevant for its weight progression, is based on a misunderstanding.

Thomas Phinney

Of course all aspects are being interpolated. I may have worded that poorly. But yes, I understand how interpolation works. (I did my Master's thesis on extreme form change in multiple master fonts.)

My point, however poorly expressed, is that even with linear interpolation from a reasonably designated "light" to "black" master, the visual boldness progression may differ depending on the design, in at least *some* cases.

Perhaps you will argue that the cases I have seen all involve poorly designed fonts. Yet as there is quite a bit of disagreement about what the correct standard weight curve is, to start with, I am amused that you can not only insist that not only is there exactly one perfect curve for the weight progression for all typefaces, but think that for me to believe otherwise “is based on a misunderstanding.”

Ben Blom

Thomas, perhaps I misunderstood you when suggesting that your reasoning is based on a misunderstanding. However, it seems again that you misunderstood what I wrote before.

Thomas Phinney: as there is quite a bit of disagreement about what the correct standard weight curve is, to start with, I am amused that you can not only insist that not only is there exactly one perfect curve for the weight progression for all typefaces

If you suggest that I said there is exactly one perfect curve for the weight progression of all typefaces—then you misunderstood me. I only suggested that if a designer chooses a specific weight progression for a typeface, then this designer can reuse this same weight progression for his/her other typefaces (as long as this designer continues to prefer this weight progression).

Thomas Phinney: even with linear interpolation from a reasonably designated “light” to “black” master, the visual boldness progression may differ depending on the design, in at least *some* cases.

I am not sure what this “visual boldness” means. I guess it is related to someone’s perception. However, I was not talking about someone’s perception, but about interpolation values, like this linear series of values: 0%, 25%, 50%, 75%, 100% (the masters are 0% and 100%). When reusing such a series of interpolation values, the (technical, objective) “weight change” between the interpolation results—or the relative weight of one result compared to the other results—will always be the same (whatever the design of the font involved).

Chris Lozos

To me, "visual boldness" takes in to account stroke weight, counter width to stroke weight differential, and spacing.

Thomas Phinney

Chris Lozos said:

To me, "visual boldness" takes in to account stroke weight, counter width to stroke weight differential, and spacing.

Exactly. The progression of these things may differ between typefaces....

My own tests show that this can affect what I perceive as the ideal weight curve.

The whole notion of a weight curve is perception-based in the first place. So while the numbers are always the numbers and invariant, that does not mean they are always “right.”

Ben Blom

Thomas Phinney: The whole notion of a weight curve is perception-based in the first place.

Thomas, here you confirm that you misunderstood me. I discussed a weight progression like a linear progression or a “De Groot” progression—which is about numbers (interpolation values), not about perception. You responded to my post as if you were discussing the same thing—but you were not. So—one big misunderstanding.

Thomas Phinney

The entire point of the numbers is to generate something that “looks right.” Which numbers are “correct” is dependent on perception. So, yes, a misunderstanding, but not on my part.

Ben Blom

Thomas Phinney: The entire point of the numbers is to generate something that “looks right.” Which numbers are “correct” is dependent on perception.

Of course, in the end, the choice of a weight progression logic is based on perception (and also on the preference of the designer), as I said in my original post. Does the fact that this choice is partly based on perception, lead to another conclusion regarding the reusability of a chosen weight progression logic?

What is this thing that should “look right” we are talking about here? Answer: The “weight change” or “weight progression” between the interpolation results—or the relative weight of one result compared to the other results. So, this is not about absolute weights, but about the difference in weight between different interpolation results (i.e. relative weights).

Now, does the “visual interpretation” by someone’s eyes of objective relative weights (i.e. interpolation values), into perceived relative weights, change these relative weights? In other words: Do those perceived relative weights differ from those objective relative weights? No, because the glyphs which weight is visually compared, are from the same typeface, so the influence of the design of these glyphs on their perceived relative weight, is the same for all those glyphs.

The perceived absolute weight of a specific interpolation value within a font, may look different from the perceived absolute weight of the same interpolation value within another font—because of the difference in design between those fonts. However, within any font, the perceived relative weight of the same interpolation value will be the same, i.e., will have the same perceived relative weight compared to the perceived relative weight of the other interpolation results of that font.

So, if a designer chooses—based on what he sees and likes—a specific weight progression logic for a typeface, then he can reuse this weight progression logic in his other typefaces (as long as his preference for this weight progression stays the same).

Daniel Veneklaas

PabloImpallari said:

There is at least one standard naming scheme than you can follow, and that CSS coders and Browsers are supposed to be familiar, at: https://www.w3.org/TR/css-fonts-3/#propdef-font-weight and https://www.w3.org/TR/css-fonts-3/#propdef-font-stretch

I find it interesting how a set of standards has been attempted to be established, but I assume in this case by the web dev community and not the type community.

Surely all users benefit from a standard set of naming conventions, but it probably has to come about organically to really stick.