A real change weight feature

John Hudson's picture

After hearing for so many years how much better Fontographer's change weight feature was than FontLab's emboldening, I downloaded the new Fontographer 5 demo to try it out. Clearly, the outline transformation algorithm used by Fontographer is superior in that there is much less distortion and a much cleaner result, but still... this is a weight change feature programmed by someone who has never designed a typeface and knows nothing about letterforms. Any weight change function that is only capable of adding the same amount of weight in both the x and y direction is completely useless. Any weight change feature that only offers the options to preserve existing height and/or width is similarly useless.

The interface for Karsten Luecke's Glyph Tweaker script for FontLab Studio shows what a real weight change feature should provide:


.
First, there are independent controls of height and width that not only allow one to fix those, if one wishes, but more importantly enable one to intelligently adjust them as part of the outline transformation. The height control is particularly clever, because it enables one to identify a particular outline feature height and adjust everything else relative to this. So, for instance, you might adjust or fix the x-height, and not run into the silly situation produced by the Fontographer weight change feature in which, if overall height is fixed, the x-height of glyphs that start as the same end up different:

The ability to independently change the x and y stem weight, rather than adding/subtracting a global amount obviously makes much more sense given that type is very seldom truly monolinear.

Now, the KLTF Glyph Tweaker is not perfect, in large part because it relies on some FontLab Studio outline tranformation functions that sometimes results in outline kinks that require a lot of manual cleanup. But if something like Karsten's tweak UI could be married to Fontographer's cleaner outline transformation algorithm, then perhaps we'd be approaching a real weight change feature.

I say approaching, because ultimately a weight change feature should also understand diagonals and be informable about design topology, e.g. by enabling the user to identify typical hairline or bowl-to-stem weight thicknesses and have these independently adjusted too, regardless of whether they are horizontal, vertical or diagonal features.

Khaled Hosny's picture

Have you ever tried FontForge's change weight feature?

John Hudson's picture

No, because I've never succeeded in getting FontForge to work properly even on Linux. But looking at the documentation is appears that, apart from being able to isolate serifs if they are flat, this feature suffers from the same limitations as Fontographer.

oldnick's picture

If this is a feature request for FL6, can I add my two-cents' worth for MORE ZOOM? DTL's Bezier Master, as I recall, can zoom to display a 4x4 unit grid, and CorelDraw can zoom to over 400,000%...

John Hudson's picture

Hey, go start your own feature request thread! :)

Actually feature requests per se should be posted on the FLS wish list forum. I posted these here because I'm interested in gathering ideas on what would constitute a better weight change feature, in some respects independent of specific software.

Arno Enslin's picture

Any weight change function that is only capable of adding the same amount of weight in both the x and y direction is completely useless.

This is partly untrue. The weight change function of Fontographer is extremely useful, if a font prints too light. In many typefaces the contrast is too high in my opinion. So it absolutely doesn’t matter, that the function reduces the contrast, if you make a font bolder. But you cannot simply create a bold weight with it, that’s true.

riccard0's picture

gathering ideas on what would constitute a better weight change feature, in some respects independent of specific software.

http://typophile.com/node/65456

anagnost's picture

Khaled Hosny

FontForge's Change Weight is in fact very much like emboldening in FontLab. But FontForge also has a "Change Glyph" tool, which provides the same features as the Glyph Tweaker script (as far as I can conlude from John's screenshot) and a bit more. In particular, it is possible to manipulate alignment zones, thus changing vertical glyph proportions. There is also a diagonal width control. Of course the result should always be corrected manually, but anyway this tool has already saved me many hours of work.

Khaled Hosny's picture

@ anagnost:

I didn't use either, I was just interested in opinion or perceived limitations, since intent to use it in my current Arabic font project.

dberlow's picture

>what would constitute a better weight change feature,

Variations. I mean, the stuff in these tools is for slow-motion users like us: actual users need this on the fly, as perhaps you can now see? ;)

Cheers!

JanekZ's picture

Steven Moye in "Fontographer" (p. 152 and following) gives a way to embolden effectively a serif face. I made a draft with Times Italic. Here is the result:


Left: original, Right: emboldening by 20 em units.
I suppose the crude effect is quite budding. (to be honest I introduced one improvement ;-)

John Hudson's picture

David,

TT variations are lost in the mists of time now. Pretty much all the web links to GX typography are dead. You and Dave Opstad might be the only mortals who remember how they worked.

So, how did they work, relative to the question of weight change? My understanding of the variations legend is that, like MM axis masters, someone needed to design at least two outlines, from which variations could be interpolated and extrapolated. Very useful, as you say, for on-the-fly, user-controlled weight change. But what I'm talking about is deriving a second weight from a single weight in a way that is more intelligent than what Fontographer and FontLab currently offer.

dberlow's picture

>TT variations are lost in the mists of time now.

Lol! if you take even a brief walk in the excrement storm of web type, it'll clear that mist for ya! And remember, close your mouth if possible. :)

D.O. was not really the developer of cvar and gvar. That would be Mike Reed. So... that's three of us. Then there's EvB who wrote an app that makes them, and I know more than a few Superpolator users, don't you?

>So, how did they work, relative to the question of weight change?

Check it out! What do I look like the weightipedia?

Cheers!

John Hudson's picture

Sure, I know Superpolator users, and I've gone and I tracked down the Apple gvar table spec. And this all indicates that variations, like other interpolation mechanisms, rely on designer definition of two or more weights. Unlike MM, the variations approach involves specifying point moves from a default weight, rather than creation of two end-of-axis outlines, but its still an interpolation mechanism.

So variations are tangential to what I'm talking about in this thread, which is tools for creating -- or at least getting a head start on creating -- secondary weights from an individual outline, whether those secondary weights end up being defined as individual fonts, MM axis masters, or tuple variation data.

dberlow's picture

>....variations are tangential to what I'm talking about...

ooooooohhhhhhh. So what is a "real change weight feature" to you?

John Hudson's picture

One that enables selection of a bunch of glyphs and specification of what should happen to them during a weight transformation, in such a way that features of the glyphs that are distinct in terms of stroke contrast are transformed distinctly, e.g. as a minimum, are not subject to universal increase/decrease in the x and y direction.

In other words, exactly what I described in the first message in this thread.

blank's picture

One that enables selection of a bunch of glyphs and specification of what should happen to them during a weight transformation…

That sounds like the job description of a junior type designer.

Ray Larabie's picture

Other than less fucked up vertices, I'd the ability to retain vertical dimensions while allowing horizontal dimensions to change . . . or vice versa.

Weight change on an angle would be useful. For example, if I were to rotate a shape 45 degrees, apply vertical weight but no horizontal weight and rotate back 45 degrees you get kind of a "nib" effect. I'd like that but without the rotation. Eh?

blank's picture

It’s just occurred to me that if LivePen works well once they incorporate it into FontLab we may not have to worry about other weight change tricks.

Mark Simonson's picture

It is possible to transform glyphs before making a weight change, and then transform them back afterwards, to get non-uniform weight change effects. Not that I've ever used it, but it should work.

JanekZ's picture

It is what I did in the sample above. It works?

Ray Larabie's picture

@Mark, it works but it throws the registration off so you have to leave an original on the mask layer and line it up manually.

dberlow's picture

Mark >It is possible to transform glyphs before making a weight change, and then transform them back afterwards

It is, and this is often a way. The problem with all the transformations, including change weight when the point structures are not compromised, is that they don't "know" the difference between one x thing and the next, the difference between one y thing and the next, much less the connecting things between the things, (going by Stone's terminology of letter parts).

I just finished 10 of these boldenings, boldening for web text boldness which is more than most. They all followed a similar pattern of; scale, (slant some), and then grab the right subset of the contour points, from a ball terminal, up to the whole x-ht, scale them tiny amounts, tweak multitudes of points 2-5 units to complete the gross coordinates, before micro-tweak for shape.

We also recently did 4 other boldenings using lost-in-the-mist technology and these were different. There is a moment to select values for the axes while thinking about going to the beach, and then a moment to process while mixing ice cocoa, and then it's right to micro-tweaking shapes. Yes, you have to make the decisions up front on Ultra and Thin, (telling the software about the extreme whereabouts of all the x and y and other things), but then you can go to the beach and phone in your weights.

Cheers!

Syndicate content Syndicate content