Hello all. I'm very new to type design - I attended the Granshan 2015 conference to give a presentation about the SILE typesetting software and met a load of lovely and interesting people (some of whom I recognise on this site!). Since then I've been devouring everything I can find about type design.
As a first exercise, I took the SILE logo contributed by Raphael Bastide -

- and tried my hand at filling in the rest of the upper case letters. I know I haven't done a great job but I feel like I am now at a stage where it might be beneficial to get comments from others to help me improve. The idea is a sans display font for titles, logos, etc.
Any comments would be gratefully received!
Comments
I recommend that you print paragraph samples and make notes. Make changes, and test again. If you keep testing, making notes and repeating, you'll do fine.
For example, see the thickness of the O, how does that compare to the S, G, C, U? How about the thicknesses of the lines on the V and W? Do some parts feel too thick or thin? Make notes, make changes, and test again. Also check out this article about overshoots.
I’d only add: combine your characters and consider their width. Compare, for example, how E, F, H, M and N relate in width, or B, D, P and R. In my first few fonts, my capital widths were consistently unbalanced. Heck, I still don’t know what I’m doing.
I wouldn’t suggest working blind. I think it makes more sense to lay this down right next to Alternate Gothic, find the spots where Sile Sans differs, and focus on bringing those to life throughout the typeface.
It looks like you've made the horizontal part of the vertices of /M/N/W/ the same width as the stems. They can be longer, and that can help sort out the angles and counters of those letters more easily (as you've done with /A/ and /V/).
The D and O have 'reversed contrast' - they are thinner in the sides than on the ends
The middle of Y is a bit too high for me, also the bar of the G
The tail of the Q is too skinny
The N needs 'inktraps' to lighten up its corner joins
Craig: Aha, useful. I have stretched out /M/W accordingly. Looks better now. I haven't yet gone through and thought about balancing all the widths yet; that's next to do.
Dave: Thank you! I have made those alterations.
Ray: Thanks - I've thrown away the /S so that I can harmonize more with the curves of /P/D/C/G. And also to get over my fear of /S. (I'm still a bit scared of /S)
Here's another sample. The /S sucks, but it's a work in progress.
EDIT: Is this what you are going for?
Check out Frank Blokland's www.lettermodel.org to learn more about proportions, and spacing.
Look for other methods on spacing; Miguel Sousa wrote a good one, its floating around online
Quick comparison, one year. Keep in mind that I have better tech now.
The join of the arm to the stem in /r is too high
Where the 3 parts of the /y meet is too dark
The /z seems like the top is wider than the bottom
One more thing: for a bunch of professionals looking at the first-time work of a rank amateur, every single one of you has been courteous, helpful and encouraging. This is literally the first time I've seen this on any Internet forum, on any topic. Thank you.
One quick observation: once the /S gets this flattened at top and bottom it needs to have less overshoot.
I have started looking at a lighter weight, but learning the lesson of getting early feedback before charging ahead - does this look like I am doing the right sort of thing? Red is the original; lightest weight is the new master; middle is an interpolation.
I feel like the light /e is now too narrow, but it's proportionally similar to the other letters - ~85% width. And the light /d looks more "blocky" than the original. Or is it just me?
I don't like the proportions of the light /R or /K, but I'm not sure how to improve them. I can't work out which way to shift the leg of the /R to improve its balance; nothing seems to help. I feel like the thin /C is a bit overcompressed, but if I widen it, it would be wider than the semibold, which seems wrong.
I guess more generally I wonder about how my attempts to translate the design to a new master have gone.
If you're having trouble with specific letters, look at similar typefaces for ideas, or simply draw as many versions of those letters as you can until you find something that feels right.
BTW for reference below are the relevant passages from Rault's Excoffon book (first in the original French, followed by my translation).
Chambord/Touraine:
Marcel Olive lui demande de dessiner un caractère typographique pour concurrencer un alphabet de la fonderie Deberny & Peignot, le Peignot, dessiné par Adolphe Mouron Cassandre : À l’époque, il faut bien avouer que je ne connaissais rigoureusement rien à la typographie. J’étais jeune ; j’ai voulu faire quelque chose de différent, et j’ai fait avec ce caractère, le Chambord, du « sous-Peignot ».
Marcel olive asks him to draw a typeface to compete with one of Deberny & Peignot’s alphabets, Peignot, drawn by Adolphe Mouron Cassandre: It must certainly be admitted that at the time I knew resolutely nothing about typography. I was young; I wanted to do something different, and I achieved it with the Chambord alphabet, a “sub-Peignot”.
Banco:
Nous apprenons que la fonderie Deberny & Peignot est en train de préparer un nouveau caractère de publicité dessiné par Jacno. Un article était paru dans une revue professionnelle avec la photo de Jacno en train de dessiner son caractère. Je vais être honnête : j’ai fait de l’espionnage industriel, j’ai pris ma loupe et j’ai scruté cette photo avec le plus de précision possible.
We learned that the Deberny & Peignot foundry was preparing a new advertising typeface by Jacno. An article with a photo showing Jacno drawing his alphabet had appeared in a trade magazine. I’m going to be honest: I carried out industrial spying – I took my magnifying glass and examined that photo as carefully as possible.
Trying to make this all of practical usefulness to Simon, I would say that to me it's a matter of intent and healthy skepticism: if you intend to make a real cultural contribution, and you don't assume that precedents are necessarily well-conceived, freely look at everything.