At Thu, 20 Oct 2011 08:57:28 -0500, Jon Nelson wrote:

On Thu, Oct 20, 2011 at 8:44 AM, Takashi Iwai wrote:

...

At Thu, 20 Oct 2011 08:26:07 -0500, Jon Nelson wrote:

...

On Thu, Oct 20, 2011 at 8:19 AM, Takashi Iwai wrote:

...

At Thu, 20 Oct 2011 14:50:47 +0200, Petr Gajdos wrote:

...

On Thu, Oct 20, 2011 at 02:07:52PM +0200, Takashi Iwai wrote:

...

> Here is an counterexample of my previous examples (again left hintfull > and right hintslight).

Hmm, a wrong attachment?

Yes :-D.

Yeah, in this case, the glyph are slightly shrunk vertically, so looks uglier indeed.

OTOH, the hintfull has still a better visibility than hintslight.  For example, look at the letter "m" in the right side (check it via xmag or such).  The whole glyph is filled with gray pixels.  So if this letter alone is picked up without texts, you can't say what it is.

...

...

> causes the small font in the search input on opensuse.org render > "better" for me.

would be a good compromise.  (I guess we can change the style even in the point size condition?)

I hope yes, see conf.avail/20-unhint-small-vera.conf and test on pixelsize in it.

Good, we have many choices :)

I have two concerns - on my laptop, I had to go back to light because full was fuzzy. Perhaps it's because I'm using DejaVu Sans?  As has been amply demonstrated here, hinting /does/ appear to be working in 11.4 at least and what works for one person might not work for another.

First of all, you need to define the verb "work" in this context. It's about the visibility / clearness of each font glyph, or it's about how well the original form is kept, or overall readability in small size, etc?

If hintfull gives fuzzier (= less clear) output than hintslight, as I mentioned, something must be wrong.  It's a bug either in the font design or in the render.  In such a case, we need to fix it or give a workaround for this font, i.e. font-specific config.  And, this is not what works for one person but not for others.  It must be reproducible on every machine.  It's no art.  It's just a program.

Let me see if I can supply some samples showing where, in some cases, full is not as nice as slight.

Yes, the visible examples would be good for more objective discussions. Whether it's "nice" or not is subjective and unmeasurable, so we can't argue in such a way. More better way to compare is needed.

Don't forget about these other items:

- for LCDs, "RGB" usually works best

Do you mean about subpixel rendering? Changing this would be easy, but I guess we can't enable it as default.

- use embedded bitmaps for no? At which point size is the cutoff?

Really case-by-case. Some fonts have pretty limited amount of embedded bitmaps while some have many and good enough.

- use the autohinter now that the patent has expired?

I thought it was already enabled? Maybe missing something?

In my case, with DejaVu sans mono (what I use in konsole), 'slight' is clearly superior to 'full' at the font sizes I use. As the font size increases, 'full' looks nicer.

As said, please define what is "nice" :) thanks, Takashi -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org