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

Koji Ishii wrote:

> > Your proposal introduces an inconsistency when 'text-orientation:
> > upright' is used, real italics and synthetic italics will differ
> > in how they display, one will slant right, the other down.  Case
> > (3) in the illustration above results from setting 'font-family:
> > Arial', the real italic face of Arial is used but for the Japanese
> > text the italics are synthesized.
> 
> Good point, I missed this case, and I agree this is an issue for (2).

This is why I'm saying I don't think there's a "correct" behavior,
only options that all have undesirable side effects.

> > But as everyone is saying, there really isn't a use case for
> > vertical italic Japanese text runs
>
> The use is less common, but use case does exist. "Less common" is
> not equal to "there really isn't a use case." It's used in Harry
> Potter. It's used in several books in light-novel style[2].

Koji, could you post *examples* of these?  The only thing you've
posted are testcases which makes it difficult to assess what the
underlying use case is.  Are these really *italics* or the use of
Japanese obliquing (斜体、shatai)?

> > and proposal (2) *does* introduce an inconsistency into vertical
> > runs of Latin italics, the synthesized italics will differ from
> > real italics.
>
> I agree on this point. But issues of your proposal are listed in my
> blog post[1]; there are much more common and severe issues. Upright
> non-full-width Latin characters in Italic are really rare. Your
> proposal breaks dashes in simple Latin text in sideways. You need to
> create a list of common code points not to slant. Don't you agree,
> if we chose the least problematic case, it'd be (2)?

What is the use case of italic dashes in Japanese vertical text runs?!?!?

The "severe" issues you refer to aren't realistic usage at all, I
don't think we should spend lots of time fretting about feature
combinations that never occur in practice.

In particular, I think it's not a good idea to try to achieve Japanese
obliquing text effects by shoehorning a very limited version of these
effects into the behavior of 'font-style'.  We both agree that it
causes an inconsistency for vertical text runs of Latin italics and
that's something I can imagine non-Japanese authors actually using.

In his comments on synthetic italics, Taro makes a similar distinction [1]:

    On the other hand, Shatai is a compression transformation
    that can be applied in different possible angles and with
    different factors to each EM body with respect to the center
    point of the EM box, and if the baseline-alignment option is
    set, the affected glyphs are rotated, so that the direction
    of the progression of the upper and bottom (or left and
    right) sides of the type body matches that of the line.

    Does the Shatai operation have anything to do with Western
    roman glyphs and its dimensional properties? It doesn't. The
    difference between Shatai and Slanting is a fundamental
    difference, and the difference is not only in the appearance
    of the results from the two operations, but in its underlying
    operational model. This means that the two things are totally
    different things.
      
    If I summarize these considerations:

    1. It is nonsense to apply "italicizing" to Japanese characters and lines.
    2. It is nonsense to apply Slanting to Japanese characters and lines.
    3. It is nonsense to confuse "italicizing" and Slanting.
    4. It is nonsense to apply Shatai to Latin alphabet characters and lines.
    5. Although some people may feel strange, seeing a typical
       Japanese vertical line with Shatai, because the left side
       of each type body is positioned lower than the right side,
       this is very reasonable according to the tradition of
       Shatai transformation in Japanese typesetting. (By
       explicitly adjusting the Shatai transformation parameters,
       it is not impossible to make the right side down, and the
       left side upper, but it is unusual).

I should also point out here that the crude version of Japanese
obliquing that you're proposing to add to 'font-style' can already be
achieved using a geometric shear operation on the line via 'transform:
skewy()'.  That's how I made my examples! :P This approach allows
users vary the angle and to make the glyphs slope "down to the left"
rather than "down to the right". According to Taro's mail, this is in
fact the more common usage of Japanese obliquing.

Regards,

John Daggett

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

Received on Wednesday, 15 May 2013 00:37:44 UTC