Have you tried FontLab? ;-)

You can embed bitmaps in TrueType-flavored OpenType fonts using FontLab Studio.

A.

I know what he means. I have instructed TrueType fonts using Visual TrueType and FontLab. The interfaces are a joke, and the process is backwards.

The GUI for writing instructions should be nothing more than a pixel grid — a simple point-and-click bitmap editor. So all you have to do is place the pixels where you want them, and the software writes TrueType instructions and places deltas which ensure that the rendered type exactly matches the bitmap you drew. This would be good software.

Nobody makes good software anymore because it's too hard.

The whole point of TrueType hinting is to use an instruction set to control rasterisation because editing bitmaps by one-by-one takes too long. A hinted font is a little machine for automating bitmap production. So a hinting UI that is basically a bitmap editor would defeat the purpose for which TT hinting was invented.

Yes, hinting is defining repetitive rules so that most individual cases are solved automatically, plus adding some exceptions to precisely control how some (few) individual cases are solved. Editing bitmaps is solving all the individual cases one-by-one. You can still do it — right there, in FontLab Studio, by editing the embedded bitmaps.

But if you edit the bitmaps, you just do this: edit a black-and-white pixel matrix for monochrome rendering devices. With hinting, you do much more: you establish rules how a glyph should behave in various rendering situations: different color modes, different point sizes, different resolutions.

In other words, when you're hinting, you're making decisions about multiple outputs for each glyph at once. A user interface where you turn pixels on and off would not work because in such a user interface, you only make decisions for one output at a time. But again, you can do this — through embedded bitmaps.

Best,
Adam

I can't help but feel that if hinting were automated to match a user-provided bitmap at one or two sizes, the hinting results would be greatly improved for most other sizes as well, with far less work.

Also, some display rasterizers appear to use hinting data for "smooth" greyscale rendering, which of course wouldn't be influenced by embedded bitmaps.

"Will hinting die before it kills me?"
Funny, I have wondered about this at roughly ten year intervals, but only three times, so it's not a great sampling.

"The whole thing is just too deliberately arcane for me."
I am curious as to what parts of this underused cutting edge you consider deliberately arcane?

"...if hinting were automated to match a user-provided bitmap at one or two sizes, the hinting results would be greatly improved..."
I guess you mean the results would be the same, but that the abilities of bitmap designers would be enhanced.

Auto-tracing your bitmap would be easier than trying to write a bitmap to hinted outline machine, I think.
But if one has a "captive grid" i.e. a single, size/resolution/device/rendering combination, then why not draw an outline, or redrawn "the outline" to fit it?

Cheers!

cerulean. I know how frustrated you feel. I felt the same when I first started looking at hinting with VTT. But you really don't want to start pixel popping to define TT instructions. If you have a Light serif font that has a single pixel weight up to say 24ppm, the last thing you want to do is have to pop pixels for all those sizes in each glyph.

VTT is great at allowing you to play with instructions visually to find the best compromise for a range of sizes, just leaving you to fix a few problem sizes with deltas.

Yes - hinting can be both complicated but also repetitively boring. When I do it, I make sure I hit it without being disturbed and with my iPod turned up.

m

>with my iPod turned up

Then you will soon be deaf, see http://media.www.dailytoreador.com/media/storage/paper870/news/2008/10/2... for example

I can spend all my waking hours designing type without a problem but I would puke buckets of blue fuzz if I had to spend one minute doing hinting :-)

ChrisL

I think I see a middle path on this:

• Automated hinting generates initial hints
• this in turn generates bitmaps of a specified size
• Bitmaps would then be manually adjusted,which would modify the hinting instructions such that the adjusted bitmap is produced by the hints.

Conceptually, it makes sense to do the hinting interface in this way, but I have no idea what kind of monster it would be to program.