I started work on a 19th century French Didot style font

Dennis Hotson

Hello. I'm a newbie typography enthusiast and this is my first attempt at creating a font.

I'm creating it based roughly on scans from 19th century books printed by Didot. I've been eyeballing the printed font and using the pen tool in Illustrator before copy/pasting it into Glyphs app. I've also used Fontlab VI to tidy up some of the curves.

Here's my current progress: http://dn.ht/didhot/

I've only done uppercase so far and I'm generally pretty happy with the look—but I think I need some help from someone with a trained eye.

Any critique or advice would be greatly appreciated. Thank you!

Robin Mientjes

Your thin lines are all different weights. Some of your round outside shapes are relatively square-shouldered, but some are a lot rounded: compare C/G to D/O. Your P and R are strangely proportioned. The B is lacking the contrast you establish elsewhere. The S is too wide. The U is a bit off balanced – try to move the centre of the bottom to the left. The inner bottom right of the G is right inbetween a straight stem and a curvy one – I think you should choose one. The J is a very unresolved shape, and the terminal is perhaps too small. Let’s start there.

Dennis Hotson

Thank you very much!

What do you mean about the R and P being strangely proportioned? Do you mean the joining point is too high? Or more about the width:height ratio of the glyph?

Also, I spent some time trying to get the serifs looking more consistent. Do you have any tips on improving those?

Robin Mientjes

The P and R are very high-waisted, and I think need to be lowered a bit – I also think their bowls are too wide. Not much, but they feel bulky and like from a different family.

Which serifs are you asking about? C, G and S, or E, F, L, T and Z? I think either you need to bulk up C, G and S, or lighten E, F, L, T and Z.

Further, now that I look again, your M is too narrow. So is the D. Give us some words. Like sixty words. A whole lot. Go crazy. It might help narrow down where you should focus first.

Dennis Hotson

Thanks again Rob. I'm very grateful for your critique and advice!

I've made a bunch of adjustments based on your feedback and I think it's starting to look a lot more consistent overall.

Here's an updated version with a bunch of tester words:

http://dn.ht/didhot/test.html (I also attached a pdf version if that's easier)

- More consistent curves and stroke width on O/G/C
- Tuned waist height and bowl width on R and P
- Slightly tuned B and U
- Wider M and D
- Bigger terminal on J

Still todo:
- Fix serifs one way or another


I've also made a start on kerning the letters properly (using left/right kerning groups).

I found it difficult to kern the A properly since it's got so much empty space above it but the serifs stick out and almost touch the characters beside it in many cases.
I noticed it kind-of looks fine if I let it overlap sometimes (AX looks not bad for example). Is this an ok strategy in your opinion?

Robin Mientjes

I am fine with overlapping serifs in display type. They might cause trouble when people apply strokes, though – check for that in Illustrator and other such apps. It might not be what you’d want to see.

As for the rest, setting this as words has highlighted plenty of new issues.
Similar shapes are of wildly different weights. Seeing ‘ROCKETEER’ is the strongest case: R, O and C show different weights in their round shapes. Then the KET show how your straight shapes are also not harmonised in weight.

I’d suggest you set a few test phrases for yourself. Go back to HOH and OHO – Santa Claus is a type designer – and make sure they have the same optical weight. When you have that, you can start swapping in important straight and round letters for your straights and rounds. My spacing string is something like H/?HO/?OHH/?HHOO/?OO – where /? is replaced with the current selected glyph.

Keep doing this. After eight years, I think I’m finally starting to do this thoroughly enough. I’ll probably be embarrassed about saying that in a few years.

Austin Stahl

I wouldn't worry too much about kerning yet. If you want to fix the most glaring pairs just so they don't annoy you while you stare at words, that's cool. But don't worry about getting it perfect, because you will have to redo it later.

A few quick things I noticed: E, L, Z look slightly wide. Z is top-heavy (top serif is hanging out too far to the left). The diagonal stroke of your N should actually connect to the left stroke.

Georg Seifert

As @Austin Stahl said, in this stage you should not worry about kerning. Try to get you spacing and width of you letters right. There are only a fe pairs I would kern in your example (AT TA NA).

And compare the black surface of the vertical stem in the O and the M. The M might has twice as much black as the O.

Dennis Hotson

Thanks again for the advice everyone!

Sorry—just to clarify I've got a few followup newbie questions.

So far I've been making the thick vertical stems 130 units wide everywhere. The square shapes are the same width as the widest part of the thick round parts. Since there's more black area in the square shapes than the round ones, should I aim to make the round shapes a bit wider (or the square ones thinner) in order to make them look closer to each other in weight?
Also, how would you adjust for the smaller round sections like in R, P, B—should those be somewhere in between because they're shorter and "squarer"?

I've been reading a little bit about the optical illusions at play, especially around adjusting vertical/horizontal/diagonal lines: http://designwithfontforge.com/en-US/Trusting_Your_Eyes.html — I'm finding it difficult to judge after staring at this for so long though. I've heard it can help to make some physical printouts—any other tips? It just takes experience I guess?

Also, I've been making the round shapes (O, G, C) a bit thinner than usual at the top and bottom because I thought it looked cool. Does it just look wrong though?  

The K still needs a lot of work I think—I couldn't find a decent specimen in the scans I found online so I've been estimating it as best as I can. (I've heard the letter K is uncommon in French?)

Edit: Also, Austin I'm not quite sure I understand what you mean by "N should connect to the left stroke". Do you mean the diagonal should meet further down rather than at the top?
Edit2: Ok, i think I get it.. something like this? https://imgur.com/a/lzcQp

Austin Stahl

Yep, that's exactly what I meant. Bear in mind that they don't have to be identical — sometimes the M will connect a little higher than the N.