Maestro G Clef

[John Ruggero]

I have never liked the Maestro G clef because of its shape, but just looked at it enlarged. I see bumps, lumps and unevenness everywhere. In short, it is a complete mess. I am amazed. Totally.

[Knut]

It isn't merely the treble clef in maestro that has these problems, and Maestro is not the only music font with an abundance of imperfections or outright faulty vector paths. In fact, I would say that compared to the standards of commercial text fonts, mostl available music fonts are of a much lower quality.

Music fonts aren't generally done by professional font designers. The characters are less homogeneous, which makes for a more difficult design process. In addition, the details of music fonts seem to be given less attention because they are generally meant to be used in print, not on screen, and at relatively small sizes, where erroneous vector paths won't be particularly evident.

[John Ruggero]

I looked at several of the other commercial G clefs, and what you say is sadly true, Knut.

[Knut]

One might get the impression that bad vector design only has aesthetic consequences, and that the errors become more or less irrelevant when scaled to the relatively small point sizes mostly used in scoring applications. The truth is, however, that bad font design can result in a number of computational problems. Common faults like superfluous or inadequate nodes and wrongly angled control points can result in inaccurate screen display and printed output, bloated file size, and inefficient CPU usage.

Sometimes the errors are the results of workflow, such as conversion between file types (TrueType to PostScript), automatic transformations, or import of vectors from a drawing application like Illustrator (which uses fractional coordinates as opposed to the integer coordinates of font vectors), without subsequent corrections. Other times, it seems evident that errors are the result of inexperience.

The extensive usage of a font like Maestro with (to my knowledge) few complaints, indicates that it's errors only results in minor problems. It is important to note, however, that issues like file size and CPU usage becomes increasingly important with the advent of SMuFL and music fonts containing upwards of 1000 glyphs, than with the limited scope of music fonts designed prior to the implementation of Unicode.

[John Ruggero]

Knut, this is a very enlightening discussion. Thank you for this information which should interest everyone who cares about computer engraving. I had no idea that imperfections in the glyphs had far-reaching consequences. I think that I now need to check my home-made clef more carefully!

Are there other glyphs that are as prone to problems as the G clef? The C clef?

As a counter-example to the Maestro clef, here is one of my favorite clefs. It is from the Brahms Complete Works (piano works) Breitkopf and Härtel) I, at least, have the impression that the design and punch for this clef was of very high quality. The smoothness of the original seems to come through despite age, printing and scanning distortion and enlargement. I think that it is its high degree of coherence that gives us this impression, in whatever size it is printed.

Breitkopf Clef.jpg (25.58 KiB) Viewed 11403 times

[Knut]

Are there other glyphs that are as prone to problems as the G clef? The C clef?

The potential for errors generally increases with the numbers of curves in a shape. The C clef is certainly a candidate, but I would consider dynamics and other letters and numerals, segno and some rests to be just as error prone. In the case of Maestro there are multiple errors in almost all shapes containing curves. Some of these are minor, but very few of them would potentially be found in a professionally designed font.

As a counter-example to the Maestro clef, here is one of my favorite clefs. It is from the Brahms Complete Works (piano works) Breitkopf and Härtel) I, at least, have the impression that the design and punch for this clef was of very high quality. The smoothness of the original seems to come through despite age, printing and scanning distortion and enlargement. I think that it is its high degree of coherence that gives us this impression, in whatever size it is printed.

I would say that the condition of the plate is an even more likely cause of the quality of this shape. My impression is that while the quality of punch design may vary, the errors found in glyphs based on scans from the literature are mostly due to limited resolution and computational errors in the scanning or conversion from bitmap to vector formats, not the quality of the punches in themselves. The condition of the plates, and therefore the print, however, makes the source much less clear, and therefore much more prone to undesired variations and errors in the final vector image. Such errors should of course be corrected by the font designer, but often is not, unfortunately.

[John Ruggero]

Whatever computational scanning errors are involved here, I see the original clef and the original punch that made it in my mind's eye, and it is clean and beautiful. A little like listening to old recordings: we don't have to listen to the scratches and can supply in our imagination what is missing acoustically.

But it would be nice to clean up. Maybe another project... (-:

[Knut]

Whatever computational scanning errors are involved here, I see the original clef and the original punch that made it in my mind's eye, and it is clean and beautiful. A little like listening to old recordings: we don't have to listen to the scratches and can supply in our imagination what is missing acoustically.

But it would be nice to clean up. Maybe another project... (-:

Since the contours in your attachment are so clear, adapting this clef to a font is a bit tedious, but otherwise not that difficult, actually. Certainly much, much easier than creating the entire shape from scratch. An enormous amount of nodes, including the staff lines, will have to be deleted, but as long as you're tracing a mask layer of the original, and know which points the crucial nodes are supposed to define (i.e., where the extremium points in the shape are), it's pretty straight forward.

A perfect result will likely require you to be comfortable working with bezier curves, though. I would also reassemble the contours into one continuous shape (without the vertical projection intersecting the bottom of the eye and the spiral) once the staff lines are deleted. That way, you have much more control over the entire shape.

Here's a rough version of what you might end up with once the shape is relatively cleaned up. It's not perfect, and probably too bold for the larger staff sizes, but it's a good start.

Skjermbilde 2016-01-13 kl. 07.19.13.png (100.09 KiB) Viewed 11384 times

Skjermbilde 2016-01-13 kl. 07.19.51.png (65 KiB) Viewed 11384 times

[OCTO]

Amazingly beautifully done, Knut!
Maybe the middle line should be not straight but a bit bent?

What software do you use? Is it Fontographer?

[Knut]

Amazingly beautifully done, Knut!
Maybe the middle line should be not straight but a bit bent?

What software do you use? Is it Fontographer?

Thanks, OCTO!

Your probably right. It will need quite a few other small adjustments as well before it's finished. I just did this for the purpose of demonstration.

I did all the work in FontLab, except for the initial vectorization, for which I used Illustrator.