Playing around with Venetians

ivan louetteivan louette Posts: 327
edited November 2017 in Type Design Critiques
Hi ! After drawing Geranium, which was already inspired by Jenson and Venetians, I am trying to do something around Jenson itself. That was a very old dream, naïve and enthusiast. Here are my first attempts at small letters. No kerning at the moment, I only try to understand where Jenson placed the limits of every glyph. An I test also some variations like the small curve at the nose of the "e" on the second page in the PDF.



  • I haven't looked closely, but just at first glance, I really like it. It really captures the feeling of actual printed Jenson type from the 1470s.
  • Thanks a lot ! What I appreciate in Jenson is its modernity. For me its closer to modern design than Griffo. Every individual glyph in Jenson has the near abstract dynamics of Uccello's Horses (that's why I call this attempt "Uccello"). At the same time the low contrast in Jenson flatten it to the paper sheet like many Uccello's paintings are flatten to the canvas by their color and abstract shapes (and despite Uccello was a pionneer in perspective).
  • The font itself looks nice, but there seems to be some sort of problem with the PDF (on my machine, at least). Some characters (particularly the <m>) seem to be cut off on the right side.

  • Strange. Could you post some picture ? Is that visible at any size ?
  • I'm seeing clipping on the top of ascenders as well.
    You probably need to adjust your vertical metrics for that.
  • What André and I are talking about. (Viewed in latest Adobe Acrobat for macOS, in my case.)
  • That's weird. It looks fine in my two Linux PDF viewer. The ascender and descender are correctly inside the limits. I even open it without problem in an old Illustrator on Wine. Perhaps my working definition when creating the OTF is too high. I will try a lower one and post the result.
  • I’m also on macOS, so perhaps this is a platform-specific issue.
  • Here is another doc with the font at a lower definition. Please tell me if it's better.
  • Your new document seems to fix the problem.
  • ivan louetteivan louette Posts: 327
    edited November 2017
    Thanks a lot ! That's fine. That's the first time I was working at this definition. I expected the cause because I had also some problems when exporting PDF from LibreOffice with very high definition fonts this afternoon. I draw my glyphs in Inkscape and cut and paste them to FontForge (I made a template for that and it works fine). The second PDF is made with the font exported to OTF at 4000 EM but the first one was exported at 16000… which is far to much !
  • OK. I stand by my earlier comment — the font looks nice.
  • ivan louetteivan louette Posts: 327
    edited November 2017
    Thanks for your comment :) Here is a doc with the two versions of the font and a background for some more comfort. The "s" was somewhat changed. Its lower part was a little bit too flat. I had to change "s" from Jenson because the short "s" was only placed at the words end at this time and thus its drawing isn't well adapted for the middle of the words. The "z" was also changed because those ones I found in Italian prints from Jenson were much too big. And of course I did create "v" and "w" from the "y" and "j" from the "i" with a lower part showing a drop (not perfect at this stage) like Jenson's true "y".
  • Craig EliasonCraig Eliason Posts: 1,346
    edited November 2017
    Looks nice.
    I get a slightly leaning forward feel from the stems which I find pleasant--but the /p/ seems left out of that to my eye so its verticality (or even apparent backwards lean) seems misplaced.
    The /j/ descender feels a little sloppy; perhaps its curve is too gradual. 
    /e/ crossbar may be a hair too fine, but I'm not sure about that.
    I think the tittles' rightward drift could be reined in a little while still retaining that charming detail from Jenson's model.
  • John SavardJohn Savard Posts: 1,065
    Yes, viewing your original specimen on 32-bit Windows 7 with Adobe Acrobat Reader XI caused some letters to be cut off either on the right or the top, or both, but only in the Uccello samples on the left, not the Geranium samples on the right.
  • @Craig Eliason Thanks for your comments ! You are absolutely right about "p", "j" and rightward drift. The latter is an important element of Jenson's font dynamics but yes it's a little bit too strong here. For a while before beginning the drawings about three weeks ago I asked me if that was not an artefact in modern reproductions. But it was visible on left and right pages (and thus not dependent of the book position) and it didn't exist for example in Griffo's roman which gives a very different reading experience much closer to Garaldes. Its very visible in the stems and delicate to tune.

    @John Savard It's fixed now. That was because the first versions of Uccello were exported to OTF at a much too high definition (16000 EM) and Geranium was only generated at 2000 EM. Now I have adopted 4000 EM for Uccello ad it seems it works fine. Please don't hesitate to comment if you see problems again.

  • Here are the fonts regenerated from original drawings. Some work on "p", "j" and many other glyphs.
  • I'm surprised the original 16K em worked as well as it did!

    As best as I recall, the limit in the OpenType CFF format is that all coordinates must be within +/- 4096 font units. I'd expect you to go far outside that when working with a 16K em square.
  • As I was not aware about this limit (and 16000 EM work fine inside my system except when I export a Libre Office page to PDF) I used this high definition to reduce the differences when I paste vectors from FontForge to Inkscape. But in fact copy/paste from FF to Inkscape stopped to work some weeks ago for unknown reason after an update while from Inkscape to FF it still works fine. And this is a good thing because of course copy/paste in the two directions several times alterates progressively the design because of rounding up the coordinates on the way…
  • I don't have any criticism to add, very nice design. Just wanted to note how much 'cleaner' Uccello looks compared to your earlier Geranium.
  • Thanks a lot ! I get kind and constructive criticism on this forum. Probably I did some progresses from the time I began to draw Geranium two years ago (I did my very first drawings on November 15/2015 and I kept all the steps). But Geranium was already drawn with a dream of Jenson. Simply I wasn't ready at this time.
  • Some more changes in "u" (less slanted) and "e" (wider counterpunch)

    On the right I also tried a version with slightly wider thin strokes on "a" and "e" only.
  • Nick ShinnNick Shinn Posts: 2,059
    I encourage you to make it more authentic, by adding bounce and pseudo-random features.

    You have copied one artefact of the Incunabula printing process, namely the way in which the image of letters on type swells when it is printed on dampened rag paper, losing definition, becoming soft and fuzzy.

    But, your font is subject to the artefacts of digital type, namely cookie-cutter regularity of letter-form and rigid adherence to the baseline.

    What this does is draw attention to the irregularities of the glyph shapes, taking them out of context, making them look quaint and slap-dash.

    In fact, all the artefacts of Incunabula typography are of a comparable and related, holistic quality. This is why Jenson’s pages are magnificent.

    I’m surprised nobody has yet, to my knowledge, produced a pseudo-random letterpress font revival. It’s been on my to-do list for a decade, but it’s unlikely I will ever get around to it.

    The bounce effect is easy to do, you don’t even need to create new glyphs. It’s similar to the <cpsp> feature.

    Pseudo-random is a little more work, but in practice you only need one alternate set of glyphs to make it quite convincing.

    After all, it’s the “organic” irregularities of letterpress printing that make it so appealing to typophiles. 

  • ivan louetteivan louette Posts: 327
    edited November 2017
    Thanks for your very interesting and judicious comment. Of course when working around old printed types many ways are possible and every one could be interesting to envisage. But what I try to do is not a pure imitation at the moment. I am fascinated by Jenson type and i try to capture something in it which can be translated in digital type without recreating the original random effects. These irregularities give a soft and emotional touch to the original (and I am very concerned by this kind of emotion too), but about ink artefacts for example I am sure some postprocessing filter could be more efficient than using pseudo random versions of the same character. I think only bounce couldn't be obtained by svg filters for example. At the moment I try to understand how work the rythmics and the very particular ribbon like effect of this type. The latter is fairly unique and seems do depend from many details I try to find, like for example a good balance between the strength of the x height area and the serifs. I don't say that a more "authentic" version couldn't be tested using the techniques you suggested and that it couldn't give amazing results. But at the opposite I tried also to understand the exact shape of the letterpunch, but I found this kind of shape unpracticable to create an harmonious digital type.
    And there is my next attempt in the PDF. Amongst lots of tiny changes I reduced slightly the slanting and reworked the stems strenghts.
  • Interesting that @Nick Shinn sees it as too regular, and my urge is to tell you to regularize it further (make the tails of /a/d/ not dip below the baseline so much, make the end strokes of /v/w/x/y/ not rise above the mean line so much). 
    So maybe it's just right as it is!  :)
  • I don't know if it's an Adobe Acrobat bug or a bug in Uccelo, but /m looks great when I zoom in on it, and terribly crooked when I zoom out (the rightmost legs dips lower and lower below the baseline).
  • ivan louetteivan louette Posts: 327
    edited November 2017
    @Craig Eliason One of the problems is that Jenson original type is far from being absolutely regular and the x height of glyphs (and even inside glyphs  for /m) is very variable. But Once you have regularized that Jenson typeface is dead. The end strokes of /x and /y are higher too in the original and that's one of the tricks which restore the whole regularity and avoid some weird gaps in the main line. I experienced it was useful to do the same for /v and /w (I discovered that while designing Geranium because I already intensively looked at Jenson type at this time).
    @Samuil Simonov Did you see that in the last PDFs ?
  • @ivan louette Yes, in the Nov 20 specimen
  • @Samuil Simonov Did you try to print it at several sizes for examples. If that occurs only on screen and that is caused by the fact that I generate my font at 4000 em that could be considered as a bug of Acrobat because it seems that a 4000 em definition doesn't outpass the specifications. No display problems here in my two different PDF viewers on Linux… But as noted above I didn't have display problems even at 16.000 em :/
  • Yes, over-regularizing can be dangerous, and I think you’ve found a middle path that works here. I love the /u/! 
    If you like the hand-carved look of most of this, you may want to “roughen up” the circular tittles and the straight lines of /z/. 
Sign In or Register to comment.