RE: [css-fonts] proposal needed for synthesizing oblique fonts in vertical text

> From: John Daggett [mailto:jdaggett@mozilla.com]
> 
> > > From the use cases and examples you've described, authors want some
> > > form of obliquing (斜体、shatai). That's *not* italics
> >
> > Yes, I've been discussing about how to synthesize oblique. If you were
> > not, what were you discussing about?
> 
> From the use cases and examples you've given, what you appear to be
> talking about is the use case for Japanese obliquing (斜体、shatai).  Yet what you're
> proposing is a modification to the behavior of
> 'font-style: italic'.  What I and others have been saying is that Japanese shatai should not
> be equated with italics, they are different features and the usage patterns are quite distinct,
> even though they overlap because some form of obliquing may be involved.
> 
> To put it another way, when an author says "use an italic font" they are not saying "oblique
> this text", even though the end result may be the same sometimes.

Hm, I'm not sure if you and I have the same opinion or not.

I completely agree with you that Japanese obliquing (shatai) is out of scope for Fonts Level 3. We discussed this in Tucson, you mentioned that glyph-level transformation is a nice to consider in future levels. I fully support the idea.

What I'm discussing is, since Fonts Level 3 says "can be synthesized," I would like single consistent synthesizing behavior defined. In order to discuss what behavior is most reasonable for that purpose, we're studying Japanese obliquing (shatai) use cases. So use cases I have are Japanese obliquing (shatai.)

Is this different from what you're talking about?


> > 1. The font-style property defines[1]:
> > >> If no italic or oblique faces is available, an oblique face can by
> > >> synthesized by rendering the normal face with a sloping
> > >> transformation applied
> > 2. I requested to define in which direction the sloping transformation is, especially for
> East Asian text in vertical flow.
> > 3. In Feb F2F, you wanted to leave it as UA dependent.
> > 4. You updated ED[2] to the new wording as below:
> > >> If no italic or oblique face is available, oblique faces can be
> > >> synthesized by rendering non-obliqued faces with a right sloping
> > >> transformation applied. When synthesizing these faces, the
> > >> transformation should apply to all glyphs in the same way and not
> > >> vary based on codepoint or due to horizontal or vertical line
> > >> orientation.
> > 5. From this, I understood you agreed to define how to synthesize oblique.
> > 6. But I didn't understand how it looks in vertical flow from the text, so asked for
> clarification.
> > 7. You answered you want #2 of my picture[3], or #1 of your picture[4].
> > 8. I said I want #3 of my picture[3], or #2 of your picture[4].
> 
> That's a fair summary.  I think my viewpoint has evolved from simply thinking the change
> you're proposing wasn't really important since italics are rarely used in vertical text to the
> realization that the mechanism you're proposing is a very poor match for the select set of
> use cases that exist in practice and that the correct parameterization of Japanese obliquing
> is complex enough that it warrants a separate property.

Completely agreed, although, I started to worry that what you say "rarely" may be a little different from mine. In certain categories of text, they're often seen, some uses Japanese oblique, some are Italic Latin text in vertical flow. Viewing it from a whole set of Japanese text, it's rare, but real use cases do exist. Do we agree on this point? If this is the point we disagree, we can discuss further on this. It'd be much easier to discuss on it rather than on what we do now.


> Is your thinking that tweaking the behavior of 'font-style: italic' is a way of implementing
> "good enough" Japanese obliquing?

As said above, my objective is to define a single consistent behavior of what currently written in Fonts Level 3; how it "can be synthesized." We're agreeing on this point, right?


> In my mind, the look and usage of a font face with real italics should not differ from one
> made via synthetic obliquing, regardless of the writing mode and text orientation.

Agreed. That's one technical issue of the option #3. But as you must be aware of, it only occurs when a) Latin fonts with Italic face built-in is used as the first font in the font list, b) text-orienttion:upright is applied, and c) ASCII code points are used instead of full-width code points. It's really rare, I hope you agree on this point.


> Given
> that Japanese obliquing is an effect typically only applied to single lines of text, using CSS
> transforms offers authors better control, the obliquing angle can be adjusted to either
> slanted down to the left (左下がり) or slanted down to the right (右下がり).

CSS transforms only helps one-liner, while authors want to set against spans that can across multiple lines, so it can't help.

Anyone wanting more than single behavior have to wait for Fonts Level 4 or future levels to get glyph-level transforms. There's a consensus in the Japanese discussion for this.


> You've proposed that 'font-style: italic' cause "slanted down to the right" obliquing but
> according to Taro [1], "slanted down to the left"
> is the more common usage and those are the examples that show up in documentation
> explaining InDesign's shatai controls [2].  Likewise, your formulation of 'font-style: italic'
> would not support the Harry Potter use case you cited since it also slants down to the left:
> 
> http://lists.w3.org/Archives/Public/www-archive/2013May/att-0032/harrypotter-7-193.p

> ng

Please refer to my 2nd blog post[3] for the discussion between up or down. We had a deep discussion on this. Many thought #1 is better at first point but finally concluded that #3 is better.


> I think we should resolve on the behavior currently in the spec and try to tackle Japanese
> obliquing at the next level.

Agree to "tackle Japanese obliquing at the next level," but disagree to resolve on the behavior currently in the spec (which is #2 if I understand correctly,) for issues I wrote in the blog post[3].


> [1] Taro Yamamoto's comments on synthetic italics in vertical text
> http://lists.w3.org/Archives/Public/www-style/2013May/0331.html

> 
> [2] examples of Japanese text obliquing effects
> http://lists.w3.org/Archives/Public/www-style/2013May/0334.html


[3] http://koji.ec/archives/32


/koji

Received on Wednesday, 22 May 2013 11:58:29 UTC