How to collect typography inspiration you can actually use

Every designer's screenshots folder holds the same failure. A magazine spread, a shop sign, a competitor's pricing page, saved in the moment because the type looked exactly right, then reopened eight months later as a picture of some very good letters and nothing else. The shape survived. The name did not.

That gap is the actual cost of the habit most people have. Re-identifying a typeface from a bare screenshot can eat an afternoon: cropping the image, uploading it to a matcher, scrolling past twenty near misses, giving up and picking something close enough. Doing the identification at the moment you save the specimen takes seconds, because the type is right there, live, in its original context. The single habit that separates a type reference library you actually use from a folder of orphaned screenshots is exactly that: identify first, file second.

Why is a typeface screenshot worthless without a name?

A picture of type on its own tells you what it looked like, not what it was. It cannot tell you the weights that were actually available, the exact sizes used, or whether the version you are half remembering had tighter tracking than the one you are now looking at. All of that context lives in your head for about as long as the tab stays open, then it is gone along with the browser history that could have led you back.

A name changes all of that. Once you know the typeface, you can check whether it pairs the way you remember, whether the weight you want is even in the family, and whether you are allowed to use it at all. Skip the name and every one of those questions turns into guesswork later, at exactly the point in a project when you have the least time to spare for it.

How do you identify a typeface on a website?

For anything on the web, your browser's own inspector is the most reliable method there is, and it is free. It is not reading the shapes of the letters and guessing. It is reading the actual CSS the site is using to render them, which is as close to the source as you can get.

1. Right-click the text and choose Inspect in Chrome, Edge or Firefox. In Safari, turn on the Develop menu in Settings first, then Control-click the text and choose Inspect Element.
2. Open the Styles or Computed panel in the pane that appears alongside the page.
3. Read off font-family, font-size, font-weight and letter-spacing against the element you selected. The first name listed under font-family is usually the real typeface, and the names after it are fallbacks the browser only uses if the first one fails to load.
4. Check the Network tab for the actual font file if the CSS name looks generic or obscured. The file name and its URL, often pointing at a foundry's own delivery service, will usually confirm the real family name.

Chrome's own developer tools documentation covers this panel in more detail than you will ever need, but the four steps above are the whole method. Chrome DevTools' CSS documentation is a good reference if a page's styles are unusually layered.

How do you identify a typeface in a photo or screenshot?

Once the type is baked into pixels rather than living in CSS, there is no source of truth left to read, only a shape to match. Two established tools do this: MyFonts' WhatTheFont and WhatFontIs. Both let you upload or crop an image and return a shortlist of likely matches.

Treat what comes back as a shortlist, not an answer. These tools are strongest on clean, unmodified type and weakest on hand lettering, heavily customised display faces, or anything stretched, outlined or set on a curve. When a match looks plausible, follow it to the foundry's own specimen page and check the weights and details against what you actually saw before you commit to the name.

What to record alongside every specimen

The name is the headline fact, but it is not the whole record. Five things are worth writing down every time, while the type is still in front of you.

The typeface name and its foundry, once you have confirmed it. The weights that were actually present, not just the one you noticed first: a headline in a heavy weight often sits above body copy in a light one from the same family. The sizes used and how they relate to each other, since a headline running at roughly two and a half times the body size tells you more than either number alone. The letter spacing, because a display face set with tight negative tracking reads completely differently from the same face opened up for a line of small caps. And where you saw it: the URL, the shop, the book, so you can find your way back if a first-pass identification turns out to be wrong.

Capture the whole pairing, and file it by the job it is doing

A headline almost never gets used on its own. What actually gets reused in a real layout is a pairing: a display face for the headline next to a workhorse face for the body copy, at a size ratio and spacing that makes the two sit together properly. Crop down to one word of the headline and you throw away the half of the specimen that made it work.

Filing matters as much as capturing. Sorting a type library by serif or sans is the natural instinct and the wrong one, because nobody briefs a project by structural category. They brief it by the job the type needs to do: an editorial headline that needs weight and presence, a technical UI face that needs to stay legible at eleven pixels, something warm and humanist for a brand that should not feel corporate. Tag or file by that job and the library still makes sense the day you actually need it.

Licence it before you fall for it, and the free method that works

The typeface behind a beautiful piece of design is not always for sale, or not for the price you would expect. Some faces are cut exclusively for one brand and were never released. Others come from a foundry whose commercial licence costs a great deal more than a hobbyist typeface, or restricts the use you actually have in mind. Noting the foundry as soon as you identify a face means you find this out in thirty seconds rather than after you have already built a concept around it.

None of this requires buying anything to get started. A folder of specimen screenshots with the typeface name written straight into the filename, something like sohne-buch_landing-hero_2026-08.png, genuinely works. It costs nothing, it opens on any machine, and both Finder search and Spotlight can find a file by that name later without any extra software.

Muse does not identify a typeface for you. It will not read a font's name off an image on its own. What it keeps is the specimen image, the note where you wrote the name, weights, sizes and source, and the tag you filed it under, together as one searchable item in a visible library rather than a folder of orphaned screenshots. Type search finds it by whatever you recorded, and because Muse also reads any text that is actually printed inside the image, a specimen with a URL or brand name visible in the shot becomes searchable by that too. The identifying is still down to you. What changes is that, once you have done it, the answer stays attached to the picture instead of drifting away from it.