[OTVar] Baseline curvature axis?

Belleve Invis

This is for the text being placed on a curve.
When a glyph is placed on a curve, locally each glyph is placed on a circular arc (in either inside or outside), so we could register its signed curvature as an OTVar axis.
The value assignment would look like this:

Belleve Invis

Also the vertical version

Peter Constable

Interesting: it could provide a way for connecting scripts to remain connected when set on a path without badly distorting the text.

But it would probably also be very hard to implement, and it's not clear how much it would be used / what the felt need is.

John Hudson

Is the assumption that the curve is always an arc of a circle? Text-to-curve is a subset of text-to-path, and a curve to which text is applied might not be an arc of a circle, or might be a complex path made up of segments of different curvature.

Belleve Invis

John Hudson said:

Is the assumption that the curve is always an arc of a circle? Text-to-curve is a subset of text-to-path, and a curve to which text is applied might not be an arc of a circle, or might be a complex path made up of segments of different curvature.

"locally arc". Which means that a complex curve could be broken into a series of arcs approximately.

Hrant Հրանդ Փափազեան Papazian

I had this idea back in the MM days. Ideal for cartography... and trendy logos. Check out the "Z" in this example, for starters:
https://www.flickr.com/photos/jfsampsonphotos/29175540013

This is the type of thing that would validate the existence of variable fonts, versus the mere repackaging of weight/width/yawn.

It would be great to see Illustrator et alia hand over the curvature of the path to the font, and the font choosing its instance accordingly. Probably too dreamy, although not impossible.

Christian Thalmann

That would be a lot of work, given how non-linear that transformation would be.

Hrant H. Papazian said:

This is the type of thing that would validate the existence of variable fonts, versus the mere repackaging of weight/width/yawn.

Now I want to make a typeface with a yawn axis.

Hrant Հրանդ Փափազեան Papazian

A yaw axis.

Jens Kutilek

Am I understanding this correctly? The typesetting application would place the glyphs on a circle/curve, and pass the local curvature value to the ‘curv’ axis so the font can apply an optical correction to the glyphs, e.g. make the glyphs narrower at the baseline?



Then the glyph variations should be mostly linear. Maybe an axis mapping (avar table) would be necessary to make sure the desired angles are applied when the glyphs are spaced very tightly.

Jens Kutilek

A font with circular compensation was presented at TypoLabs, and some people seemed to be confused about whether the font itself would contain the displacement and rotation of the glyphs to place them on a circle. The demo fonts on the website don't contain this axis though. The effect can be seen in the coin samples.

Hrant Հրանդ Փափազեան Papazian

Jens Kutilek said:

The typesetting application would place the glyphs on a circle/curve, and pass the local curvature value to the ‘curv’ axis so the font can apply an optical correction to the glyphs

Yes. Please.

Besides cartography and hip logos, such an axis would also be super useful in animation, VR...

And another cool example:

Jens Kutilek

Hrant, I know that you meant it that way, but not sure if Belleve also intended it this way, and Christian’s comment about the non-linear transformation can be interpreted so that he also understood it differently.

Hrant Հրանդ Փափազեան Papazian

Jens in action:

https://twitter.com/jenskutilek/status/984925498122924032

Khaled Hosny

Jens Kutilek said:

Then the glyph variations should be mostly linear. Maybe an axis mapping (avar table) would be necessary to make sure the desired angles are applied when the glyphs are spaced very tightly.

It would be complex in cursive scripts like Arabic (which would benefit most from such axes) as one would need to change the curvature of the horizontal stems as well.

Hrant Հրանդ Փափազեան Papazian

@Khaled Hosny I agree Arabic would indeed benefit more than Latin. But Latin also has horizontals, which generally could avoid being bent only within a very narrow range.

https://twitter.com/NickSherman/status/984946558448820225

Belleve Invis

Jens Kutilek said:

Hrant, I know that you meant it that way, but not sure if Belleve also intended it this way, and Christian’s comment about the non-linear transformation can be interpreted so that he also understood it differently.

The axis' definition is the curvature with a scale (1/em for 1000).
However, using AVAR or whatever new table proposal (like XVAR or my ZVAR) could be used to simplify the working process.

Hrant Հրանդ Փափազեան Papazian

A very nice specimen, captured in Fargo (ND) by Nikki Villagomez:

Theunis de Jong

Hrant, incredibly nice indeed. I wonder if it's possible to put a date on this design. Could it be true 1920s Art Deco? (In which case I might get arrested for trying to pry it loose with a crowbar...)

Wikipedia tells me the city of Fargo is no stranger to Art Deco, as it sports a contemporary theatre in that style.

Simon Cozens

Never really saw the need for this axis until today.

Hrant Հրանդ Փափազեան Papazian

@Theunis de Jong Let me ask Nikki. BTW it looks like it says "JACK YOUNG ASSOCIATES" and "MILWAUKEE WIS" along the edges...