At the moment, only the primary ASCII characters have been created, and only the letters have been refined. I'm most interested in feedback on the letterforms before proceeding. There is no kerning at the moment.

In the italic, I've mainly been working on the lowercase letters. I'd be happy to hear suggestions for the uppercase letters, particularly the "diagonal" letters V, W, and Y (the lowercase v, w, and y have "flags" inspired by Scala Sans that I think work pretty well). It might be that the best solution for the uppercase letters is just to keep the oblique regular shapes.

Hi Stephen
Not too bad, but there are many things to take on. Many of the outline seem rather unrefined e.g. B and S. Still it doesn't look too bad in print. Let me give you a quick run down of what I notice firstly:

UC proportions seems out of balance. C, G and S seems too wide compared to O. A looks slightly too narrow. The tail in the Q is distracting.

LC proportions look much better, yet b, d, p and q seem slightly too narrow and t is way too narrow - you should extend it on it's right side. The loop in g feels distracting (yes, and thats comming from me ;-) r seems slightly too wide.

Generaly the metrics looks too tight, but this is always a matter of context so may I suggest trying out some variations.

Claus,

Thanks for your comments!

Regarding the proportions, Aurulent Sans is actually somewhere between a grotesque and a humanist sans serif. Hence, the UC letters are much more uniform in width. I'll try playing with the CGS, but their current widths were chosen to be on the wide side. The LC letters were also chosen as a compromise between grotesque and humanist proportions. The r being wide and the t narrow were chosen that way because they look better at medium sizes (16-22 pixels high) on screen. Looking at the t again, though, the narrowness really stands out. I'll certainly be adjusting that.

The open-looped g was chosen again to make the character more open at small pixel sizes (though at really small pixel sizes (less than 10 pixels) it doesn't work either). The tail going into the interior of the Q was chosen so that underlined O and underlined Q could be distinguished (underlined text frequently occurs on screen with links and hotkeys). I'd be interested if you had another suggestion how to do that.

Finally, I don't understand what you mean by "the outline seem unrefined" for B and S. Could you please be more specific? The S (and s) has given me a lot of trouble--it was by far the hardest character to design, and I'm still not completely happy with it. I'd appreciate any suggestions.

Hopefully I'll have an opportunity to work on this tonight and be able to post an updated version.

Hi Stephen
I have made an illustration of what I mean by 'unrefined'.

C and the green areas are where the connections from a curve to a straight line are too visible. They should be smoothed out so that they are invisible.

F and the magenta areas are where your curve is too flat. There are sections of too flat outline alternating with sections of more curved outline, which makes for an uneven impression.

B and the cyan areas are where the curve is much more curved than the sections meeting it, and thus that area looks as if it has a 'hump' at that location.

This is general issue with most of your characters. Some more than others. Another thing is the balance between similar shapes from character to character, and the need to differentiate the shapes from each other. In you face I miss some more harmonisation characters in-between.

Hope this is of inspiration...

Claus, thanks for your comments. I adjusted the widths of CGSA as you suggested--they look better.

I also improved the curves in the P. The new version at the top of the thread shows the new P. One problem I had was making the curve smooth from the straight line (the points C in your picture above). I noticed in Kulby that your points were not horizontally aligned, but the inside point was farther to the right than the outside point. In some othe fonts, I noticed the outside point was farther to the right. And I remember reading in some thread the comment (maybe by Paul Hunt? I can't find it again) that usually the points are aligned. Do you (or anyone else on Typophile!) have any suggestions about the best way to make these points smooth? Without additional points on the curve, it's impossible to make them exactly smooth--I wonder if there is a good way to add a few points to improve the smoothness.

Finally, could you explain more "harmonisation characters in-between"? Do you mean that my characters are not similar enough to each other? Thanks!

Hi Stephen
I don't know if you are using Fontlab? Anyway one way is to place curve-points on both sides of an original point, then delete the original point, then adjust the two new points, then set a new extreme point inbetween, and then deleteing the two placed points you set. Fontlab tries to keep the shape as it is when you do this, but due to the bezier-nature of the curves, the curve is naturally smoothed when you do this. Take a look here: http://briem.ismennt.is/2/2.3.6a/2.3.6.02.curves.htm

The harmonisation between characters is the natural relationship between n, m, h, r or between d, b, q, p, c, e, o for instance. In your face think about the shape of the very round c, o and e, and the not so round d, b, q, p.

Harmonisation means that the charactershapes should look as if they are from the same typeface - that they share an internal logic, a common recipe for their design. How you as the designer achieves this is entirely up to you. There are no rules, it just has to look 'right'.

There should be nodes at extremes, and as few as possible. Bezier handles should not reach over other points. Thats basically it.

Claus,

Thanks for your comments! After a long delay, I have finished another draft. The most notable improvement is the smoothing of curves in such letters as P, particularly where a straight segment goes into a curve.

I also fixed the width of the t, using the guideline that the right part of the crossbar should extend to the right of the stem as far as the stem extends above the crossbar. This seems to be a common proportion in many of the fonts that I looked at.

I'm still having trouble with the S. I'd appreciate any comments on how to improve its shape.

Claus, for the harmony: I think that my p,b,q are more related to my n,h,m,r than the rounder o,c,e. This is a deliberate choice. The shapes mirror those common in italic fonts; in some sense Aurulent Sans is an upright italic. I also find that this shape give more compact forms, which is desirable in a screen font.

I have gone through the UC and LC letters, and smoothed the curves in the round letters. I think they are getting closer to being "refined," as Claus was saying. The S and s still need work---I'd appreciate any suggestions on improving them.

I've begun filling in the Latin-1 character set since the letterforms are settling into their final shape. I hope to have those characters in my next post. I'm also still working on the italic shapes. The "eye" of the italic e seems to need help---I'd also appreciate any suggestions for it. Thanks!

In the Spiro thread, a discussion arose about how to parametrically define a font. Since Aurulent Sans is created in MetaType1, it is easy to change basic font parameters, such as x-height, and recompile the font. Hrant asked to see this in practice, so I generated a sample showing the word "skagy" with varying x-height. Many parameters of the font can be easily changed: stroke thickness, width, height, etc. However, to make things manageable when programming, some of the parameters are related in certain ways that might be unexpected (such as dependence of question mark on x-height).

The amount of time it took me was changing one line and hitting recompile. The tricky part was figuring out how to install several different versions of the same font.

The ability to change basic parameters globally over the entire font is quite nice, but it does require planning the scripts (somewhat) carefully. For instance, I can’t push the x-height much over 600, because then the script that generates the question mark barfs. Of course, as Hrant pointed out, it’s presumed that the parameters won’t vary too much from the value in mind when the scripts are written.

By the wya, the problematic straight-to-curve areas ('C' green areas in Claus' pretty diagram) are easier to manage with Spiro curves than Beziers, and this is one of their big selling points :-)

I've been following the discussions on Raph's spirals, and I've also tried them out to design some characters. I certainly can see the advantage of them.

However, since I'm creating this font using MetaType1, I don't think that I'll be able to use the spirals (at least until MetaSpiro comes out). However, using some macros I wrote for ensuring continuous curvature at the straight-to-curve joins, I have improved the letters in Aurulent Sans quite a bit. Do you still think they need improvement? Which letters specifically? Thanks!