Download Average Font
Crista Wilbers
For some reason I decided one night I wanted to get a bunch of fonts. A lot of them. An hour later I had a bunch of scrapy scripts pulling down fonts and a few days later I had more than 50k fonts on my computer.
I then decided to convert it to bitmaps. It turns out this is a bit trickier than it might seem like. You need to crop in such a way that each character of a font is vertically aligned, and scale everything to fit the bitmap. I started with 512 * 512 bitmaps of all character. For every font you find the max y and min y of the bounding box, and the same thing for each individual letter. After some more number juggling I was able to scale all characters down to 64 * 64.
The result is a tensor of size 56443 * 62 * 64 * 64. Exercise for the reader: where does the number 62 come from? I stored it as a tiny little (13GB) HDF5 file that you can download here: fonts.hdf5..
To start with, let's recreate real font characters with characters generated from the network. Let's plot the real character together with the model outputs. For each pair below, the real character is on the left, the model output on the right.
These are all characters drawn from the test set, so the network hasn't seen any of them during training. All we're telling the network is (a) what font it is (b) what character it is. The model has seen other characters of the same font during training, so what it does is to infer from those training examples to the unseen test examples.
We can also interpolate between different fonts in continuous space. Since every font is a vector, we can create arbitrary font vectors and generate output from it. Let's sample four fonts and put them in the corners of a square, then interpolate between them!
We can also generate completely new fonts. If we model the distribution of font vectors as a multivariate normal, we can sample random vectors from it and look at the fonts they generate. I'm interpolating between a few of those vectors in the picture below:
I was going to type some stuff out on Google Docs instead of writing it out on lined paper but I'm not sure what size font to use. The rules are one page, single spaced so what would be the best font size and maybe line spacing?
This is driving me crazy. I have an iPad Mini as my real test device. There's an UITextView and it's insanely small. I can not tell why the UITextView is all normal when opening the app on an emulator. I can't use my desired font size on any UITextView on iPad Mini.
The application is incorrectly drawing text using font Courier New because each letter is being slighty cut off. The issue is when calling the method GetTextMetrics because it is filling the TEXTMETICS struct with differing tmAveCharWidth and tmMaxCharWidth values. The application then uses tmAveCharWidth to calculate character width with is wrong because that value it can be less than tmMaxCharWidth. That issue I will be fixing.
I courious why GetTextMetrics is returning differing tmAveCharWidth and tmMaxCharWidth values for Courier New? My understanding was that Courier New is a monospaced font and that tmAveCharWidth and tmMaxCharWidth should be the same. I tested with other monospaced fonts that that assumption is correct.
"There are a lot of possibilities," Resl said. "It would be interesting to try a version with only serif fonts and only non serifs. In addition, [typographer and designer] Erik Spiekermann suggested to me on Twitter to exclusively try regular cuts. The good thing is I can create a new version at any point in time, my collection of fonts is constantly growing with no end in sight."
In some applications, narrow fonts are required to squeeze the maximal amount of text into a given horizontal space. The table below summarizes average character widths for 3993 fonts, sorted by increasing letter width (column 3 of the table). Units are standard Type 1 font units: 1000 units == font design size in PostScript points (TeX big points) [72bp = 1in].
One of them is variable fonts. After being first introduced in 2016, I just realized they have really solid browser support all these years later. Google Fonts even started to roll out filtering by variable versions back in 2020. That's like a generation ago in web years, so I figured it was time to give them a shot.
Rather than splitting hairs over the weights an average site would use, I just decided to go with the full gamut. I don't think I use them all on my site, but between headings, subheadings, body copy, and the weird stuff I do in-between, it's close enough.
Again, the average site (probably) doesn't need all of these. But if you're working with anything over a moderately complex blog, it's really hard to determine which ones to exclude. Instead, it's easier on one's sanity to just load them all. There'll be more page weight, but at least you won't unexpectedly find some text lacking the correct font face.
Because of how flexible variable fonts can be, building them can mean those versions are significantly larger than you might expect, especially when they're not exclusively packaged for use on the web. When I loaded the entire variable version of Inter, it came to over 800kb of added weight. ???
But I wasn't out of luck, thanks to a great article from Trevor Harmon, which references another from Michael J. Harold. These two introduced me to the idea of subsetting fonts (removing unneeded glyphs) and reformatting the file to .woff2 for even better compression.
In the short-term, probably not for everything. Browser support certainly isn't an issue, but the font faces I'd like when building a site aren't always available in a variable version. If I'm really keen on using a particular font, I'll either need to scourge the internet for a variable version and possibly mess with more subsetting, or settle for a static version that's ready to go. The latter is way easier.
The part of variable fonts I'm more concerned with is my ability to use them effectively. There's a whole new world of CSS properties, technical concepts, and more out there on variable fonts that I'm not remotely familiar with. If I'm going to leverage them beyond the potential performance benefits, I'll need to spend some more time in the weeds as time goes on. And that's fine with me, because it'll probably produce fodder for more blog posts.
Tricky, as you should use the character frequency as well. If you've ever seen Finnish, with all its "i"s and "l"s, you will realize that even a pure ASCII font has no welldefined average character width.
Whether designers should use serif or sans-serif fonts for body copy is one of the most discussed and unresolved questions about typesetting on the Web. Some designers prefer to give their headlines serifs (which are short, decorative lines at the end of letter strokes) to give them more appeal. The main reason to choose a serif font for your headlines is that, at a large size, serif fonts are easy to read and look great. The contrast between a serif font for headlines and a sans-serif font for body copy can be interesting, too.
Some designers prefer serif fonts for body copy because they believe the lines at the end of letter strokes help guide readers from one letter to the next, making scanning and reading more comfortable.
Surprisingly, despite the growing popularity of font replacement techniques and growing availability of new pre-installed fonts (e.g. Windows Vista and Mac fonts), designs in our study mainly used the traditional, core Web fonts, the only exceptions being Lucida Grande (which comes installed only on Macs), Helvetica and Baskerville.
Verdana is used minimally for headlines. Only 10 websites use it for body copy to begin with, and only four use it for headlines. The main reason is that Verdana puts a lot of spacing between letters, which makes it not as tidy to read at a large size. If you are going to use it for headlines, you may want to take advantage of the CSS letter-spacing property. Georgia and Arial are most popular fonts for headings.
Of course, the choice of headline font size depends on the font used in the design. In any case, in our study by far the most popular font sizes ranged from 18 to 29 pixels, with 18 to 20 pixels and 24 to 26 pixels being the most popular choices.
An obvious outlier is Wilson Miner (screenshot below), who uses a massive font size of 48 pixels for his headlines. His website is a special case, though, because all of his posts have extremely short titles, only a few words.
Do you remember about seven years ago when Web designs had tiny, barely readable elements, and body copy was set to 8 pixels in Tahoma? Small font sizes are out, and more and more modern designers are turning to large font sizes. From our sample size, we saw a clear tendency towards sizes between 12 and 14 pixels. The most popular font size (38%) is 13 pixels, with 14 pixels slightly more popular than 12 pixels. Overall, the average font size for body copy is 13 pixels.
Between 73 and 90 characters per line is a popular choice among designers, yet we also found outliers: Monocle (47 characters per line) and Boxes and Arrows (125 characters per line). To get a more exact reading for each website, you would need to take an average character count from multiple lines.
Dear Kimberly Font is a hand-drawn elegant font that will add a unique touch to your work. This font offers authentic charm and is versatile for a variety of projects, including logos, book covers, wedding invitations, business materials, quotes, and more.
Thank you for looking this font. This is a chunky bold typeface, the geometric, tempered by softened edges and vibrant shapes. Perfect for any fun quirky design work! Introducing Hanson Bold Font! HANSON BOLD Is the latest font from Hanson Method. Works well for bold strong brand identities
Rachelasti Font is a balanced, smooth, elegant and stylish serif font manufactured by ibracreative. With its gracefully elongated letterforms and subtle, sophisticated serifs, Rachelasti evokes a sense of refined vintage aesthetics,
df19127ead