Editorial method
How We Test
What we measured, on which setup and date, and what we have not tested yet. Last updated 20 November 2026.
Three kinds of claim, and how we label them
Measured means we ran a test, and the page gives the date, the setup and the result, including failures.
Reasoned means a conclusion drawn from what the characters are (which Unicode block, which fonts cover it) and not checked on a device. The platform ratings in our compatibility table are reasoned, and the table says so.
Sourced means a platform rule, such as which characters a username allows, quoted from that platform's own help page and linked.
What we have measured
| Test | Date | Setup | Result |
|---|---|---|---|
| Which styles draw in a browser | 31 Oct 2026 | Windows 11 Pro; Google Chrome 156, Microsoft Edge 156, Chromium 149 | Every letter of every style was drawn on a real page and compared against the "missing glyph" box: 972 glyphs drawn, 0 boxes. Three unassigned code points were added as a control, and all three showed as boxes, so the test could see a failure. |
| Do styled letters look like plain ones? | 1 Oct 2026 | Chromium 149, Windows 11 | None of 416 styled letters was drawn as the same shape as the plain letter it stands for. Of 39 Cyrillic, Greek and Latin-extended lookalikes, 36 were pixel-identical to a Latin letter. |
| Turning fancy text back into plain text | 13 Nov 2026 | One sentence written in all 514 styles this site generates, run back through the converter | 463 of 514 styles read back correctly; the 51 that did not are listed by name on the page. |
| Dot art and line breaks | 13 Oct 2026 | Each dot art example pasted into a single-line input and a multi-line one | Single-line fields strip line breaks, so only the five one-line pieces survive them; they are marked as safe on the page. |
| Dot art length | 6 Nov 2026 | Character count of every dot art example, blank Braille cells included | All 20 examples are 35 characters or fewer. |
How a test is run
Each test is a script we wrote for that question. It uses the same input every time and saves its output with the date, so it can be re-run and compared.
Where possible a test includes a case that must fail. The browser test is given three code points that are not assigned to any character, and it only counts if those come back as boxes. A test that cannot fail tells you nothing.
The result goes on the page it belongs to, with its limits next to it. If a test shows something we got wrong, the page is corrected and the date changes.
What we have not tested yet
No app has been tested on a phone yet: not Instagram, TikTok, Discord, WhatsApp, X or Facebook. What an app does to text after it is pasted (stripping it, refusing it, or turning it back into plain letters) is reasoned on this site, not measured.
Rendering on iOS, macOS and Android has not been measured either. The browser test used one Windows machine.
A phone test across those apps is planned. When it is published it will name the phone, the OS version, each app version and the date, and it will list the failures as well as what worked.
Who does this, and how to report a mistake
FontifyText is made by GFXSAGA. The tests are written and run by the FontifyText team.
If something on the site does not match what you see, email fontifygen@gmail.com with the page, the style, the app or browser, and your device. We re-run the test and correct the page.
