Introducing Eugene — a new SMuFL font inspired by E. Fromont (work in progress)
Hello everyone,
I wanted to share a small project I’ve been working on: a new SMuFL-compliant OpenType music font called Eugene, inspired by the engraving style of the Parisian publisher E. Fromont, whose catalog includes some famous first editions of Debussy’s works like Prélude à l’Après-midi d’un faune (1895) and Suite Bergamasque (1905).
Fromont’s primary specialization was French Symbolist and Impressionist piano and vocal music.
Eugene grew out of an interest in the visual character of music printing from the Fromont era — a style that feels distinct from both earlier engraving and the German conventions that came to dominate later in the 20th century.
I’m a total beginner in font design and I’m learning by doing — so please be patient with me. The font is in early development (v0.5) and currently covers around 86 glyphs: some common noteheads, clefs, flags, time signatures, rests, articulations, dynamics, and a few others. Everything not yet redesigned falls back to Bravura, which is credited accordingly.
The project is open source under the SIL Open Font License 1.1 and available on GitHub.
I have a few specific questions for anyone willing to take a look:
Are there any obvious issues with how the SMuFL metadata JSON is structured?
Does the font load and display correctly in Dorico for you?
Any feedback on the visual style relative to the Fromont source material is welcome
I’m aware the glyph coverage is still limited and some metrics probably need work. Any suggestions, corrections, or comments from more experienced people in this community would be genuinely appreciated.
Welcome to the forum, @mikkopatama, and congratulations on your work so far on the font! I guess as a new poster here, you can’t easily share a link to your GitHub project, so here it is:
I’ve not had a chance to try the font out for myself today, but I will do as soon as I can.
The font renders fine in Dorico, but FontLab seems to suggest that some glyphs may still have outline issues.
Maybe it would be worth checking/cleaning the contours before the final export, especially:
duplicated or double contours;
overlapping contours that have not been removed or merged;
contour direction/winding issues, especially around counters and holes;
stray points or small unwanted outline fragments;
glyphs that may have been pasted twice or not fully cleaned after transformation.
Maybe running FontLab’s cleanup tools — such as Remove Overlap, Correct Contour Direction, and a general FontAudit/validation check — would help identify the affected glyphs before producing the final .otf.
Strangely, Dorico displays the font correctly, so it may not be a visible rendering problem there, but cleaning the outlines would make the font file safer and more reliable.
Your JSON looks fine at first glance, but I’ll check it more carefully later.
Thanks for sharing this nice work.
Thank you Marc, it really turned out to be all about path directions! I figured out how to fix it in FontForge (I’m currently working on just free open source tools: Inkscape and FontForge) and now the glyph table seems a lot cleaner!
@dspreadbury@NorFont.com thank you for checking out my font! I admire your work on software and music font design so this really means a lot to me.
Here’s another sneak peek to people who’d rather first see some screenshots of some actual music. It all started from trying to mimic the look of Suite Bergamasque, so here you have the most well known part:
I received some invaluable tips on how to work with vectors more efficiently and avoid most rookie mistakes, and already did a big overhaul on most of the complex shapes like clefs and time signature numbers. The currently available version 0.502 has all these corrections, all in all 95 glyphs (getting past 100 tomorrow…).
It turned out that starting to learn vector drawing by first drawing a G clef is like starting a beginner sight reading class with Hammerklavier sonata (I teach a weekly sight reading class for pianists). After todays efforts it’s maybe not absolutely perfect yet, but definitely a big improvement from v 0.5.
I have a couple of questions, one about .pdf rendering and another one about fallback fonts (@dspreadbury?):
I’ve been doing my best with the notehead anchor points in the .json file, and everything usually seems just fine when viewed in Dorico, but after exporting to pdf the anchor points seem to be a bit off, which is clearly visible in the demo files I’ve included on Github. What should I try next?
An active Musescore developer told me that Musescore automatically falls back to Bravura whenever glyphs are missing from the table, so there is no need for having bravura glyphs inside the .otf file. But isn’t it a bit different on Dorico? The user can substitute any missing glyphs on the Music symbols -editor, but by default, nothing will be rendered if a glyph is missing from the chosen font?
Just making sure whether it’s a good idea to ship those Bravura glyphs as fallback for now. Not that it really makes a huge difference, .otf files are not particularly big even with perfect SMuFL coverage.
This is exciting news, of course, and it seems like something similar could be done to reproduce many an old house style. And since you’re a beginner—not to worry; everyone starts somewhere—it’ll be interesting to see how this project unfolds. I have one question about how you’re handling an issue that no doubt appears when you’re using printed sources: how do you address uncommon glyphs that might not appear at all in a typical score, and which therefore would require scouring many other similar sources to reproduce properly?
By uncommon, I don’t necessarily mean unusual; two examples that came to mind were the time signature 5 and 7. Obviously, they both exist—even back then, without reviving anything—but you might be looking at a lot of scores of interest before you find them.
I understand that you’re filling in the gaps with Bravura for now. But, going forward, you’ll want to find those uncommon glyphs. Are you planning to study other Fromont scores, try to find the original metal type or full printed specimen, or try to design your own version of these glyphs in a sympathetic style?
P.S. Another issue with font “revivals” generally is calibrating for the correct weight. Tracing directly from metal type results in glyphs that are too light, but when tracing from prints, take care not to make the glyphs too heavy, or worse, lumpy. Also, when you’re done, remember to make a companion “Text” font (cf. Bravura Text, Leland Text, etc.).
Thanks @mikkopatama ! I went to the GitHub site and tried to follow the installation instructions for the latest Mac OS and Mac mini M1 but the listed folder for the json file doesn’t seem to exist on my computer.
EDIT: I realized after the fact that the reference to “Library” was to the computer one rather than the user one. All is now sorted and I am able to use the Eugene font when desired. Thanks again to @mikkopatama !
Good points and questions!
I have already started expanding my research for good references beyond just Fromont, since other French publishers seem to have used the same engraving shops around that same turn of century era. For example, many Durand scores from around 1890–1910 have very similar look to Fromont, and high resolution scans can be found from Bibliothèque nationale de France (Gallica). I will state these references at some point in the Github page for clarity, but stick to the Eugene name since those scores were the ones that got me inspired to start this project.
About looking at a lot of scores: I find this “glyph hunting” part of this project quite fun and rewarding! There’s that nice little (quite nerdy) dopamine hit every time I get to put a new glyph in my “References” folder. There’s so much beautiful looking sheet music available to look at, and I’m starting to get quite familiar with the tiny details that make those French publications what they are visually. But it is a challenge of course, many times I will spot a glyph I’ve been missing but the print is so worn out that its too fuzzy to trace correctly, so I’ll carry on looking for a more solid looking specimen.
At some point I will naturally need to start designing my own glyphs with sympathetic style, since the style of the French publishing houses changed quite noticeably after that time period I’ve been focusing on and many of the more “modern” glyphs cannot be found from that era. We’ll see how long I’ll take this project on my own. Leland is a great example of a perfectly usable font with “just” 493 glyphs at version 0.8, and many current SMuFL fonts seem to settle at 1300–1500. I won’t try to predict the “final” glyph count of Eugene at this early stage, but let’s say I’ll try to get to 500 “for starters”, plenty of work there if I’m today at 100.
The correct weight is an interesting challenge, I’m working with just prints and the best working solution seems to be just checking how each symbols sits on a Dorico written project. If something seems to stick out a bit too much, I will work on it; for example, I’m considering thinning out the forte-glyphs at some point, propably closer to the current piano-glyph weight which feels more balanced with the notehead sizes.
Decided to replace the Glyph list.txt with a pdf, so you can easily take a peek at all the current glyphs without downloading the font files.
Today, among many other new glyphs and modifications, I added the old French style C Clef, as it’s found on some late 19th century Fromont editions. I personally wasn’t even familiar with this clef design before starting this project and it propably won’t find much use, but it is still there to be found if you like it.
@mikkopatama if you do not mind a question, what references did you use to learn to create your font? I am interested in the process. I have created a few very basic text fonts for specific purposes, and fingering for a wind instrument, but have not made the jump to SMuFL. Even basic font design references might be useful, need to learn more. Thanks
When you switch to a new music font via Library > Music Fonts, Dorico automatically sets up the substitution to Bravura for you. So you don’t need to include Bravura glyphs in the font yourself, though of course you’re welcome to.
If the anchor points are off, the thing to check is that the JSON metadata you’re using is really accurate for the precise glyph designs. I don’t know how anchors are handled in FontForge but perhaps others do (paging @Knut_Nergaard who I think might be using FontForge for some of his stuff; he’s also the author of SMUfoLib, which you might find useful if you decide to use the UFO font format).
I now double checked my .json metadata and the anchor points seem to be ok there. Maybe I’ll keep modifying and trying out slightly different anchors. The confusing part is that in the app the anchor points seem ok but Dorico’s pdf output looks like this (intentionally cropped closer so the difference is more obvious)
So things seem to go wrong only when printing a PDF file.
[Edited] there was a Musescore screenshot included but now with more rested eyes I realized that Musescore 4.6. was actually loading Bravura noteheads with this font, I’ll try to fix that to see how the anchors look there.
First, I just took a closer look at some Bravura and Leland glyphs to see how they were constructed: how many nodes were needed for more complex shapes, how different shapes were achieved etc. Then I jumped right in and tried it myself.
The first results were varying in quality, but then a friendly user at notat.io recommended me this article from Glyphs app website, which pretty much summarises the most essential points in drawing good paths.
When working on Eugene, I constatly have the SMuFL specification open in my browser.
It does, Thank you (feel free to add any others you might find to this thread )
I use Glyphs app, cannot remember if I read that article or not (I have been using paths for many years unrelated to font design) but should take a look at it as it relates to all this.
Best wishes with your font development it looks great already and I really appreciate the amount of time taken already! I have both those Debussy pieces in print, but not sure what publisher, presumably a more recent one, must try to find them…
Check the side bearings for the noteheads. I’ve still not had a chance to look at the font myself, but the first thing I’d look at is whether the noteheads have negative side bearings to the left and right. Make sure that the side bearings are zero on both the left and the right.
