The @w3c Web Fonts Working Group has updated the Incremental Font Transfer (IFT) specification.
All the things that an encoder or a client (decoder) must do are explicitly called out, to make testing easier.
#webfonts #fonts #IncrementalFontTransfer
https://www.w3.org/TR/2 025/CRD-IFT-20251118/
The @w3c Web Fonts Working Group has updated the Incremental Font Transfer (IFT) specification.
All the things that an encoder or a client (decoder) must do are explicitly called out, to make testing easier.
#webfonts #fonts #IncrementalFontTransfer
https://www.w3.org/TR/2 025/CRD-IFT-20251118/
We rolled out a new nav for https://typographer.com, with fresh new search functionality from @bram I’m thrilled with how it turned out. The details of this were really fun to get right, like the subtle glass effect under the search bar and deeply integrated search on every page. You can search by family name or style, and it’s so seamless and fast.
We rolled out a new nav for https://typographer.com, with fresh new search functionality from @bram I’m thrilled with how it turned out. The details of this were really fun to get right, like the subtle glass effect under the search bar and deeply integrated search on every page. You can search by family name or style, and it’s so seamless and fast.
Incremental Font Transfer is now an @W3C Candidate Recommendation.
IFT defines a method to incrementally transfer a font from server to client. The client loads only the portion(s) of the font that they actually need, significantly reducing data transfer. Multiple incremental additions to the same font are possible, such as a user agent updating a font as a user browses multiple pages.
Incremental transfer improves on unicode-range by avoiding damage to layout (kerning, ligatures, etc) rules, meaning it can efficiently support fine grained increments to Latin and to complex, shaping scripts like Indic or Arabic.
We also have an explainer, which is shorter and easier to read than the actual specification:
Incremental Font Transfer is now an @W3C Candidate Recommendation.
IFT defines a method to incrementally transfer a font from server to client. The client loads only the portion(s) of the font that they actually need, significantly reducing data transfer. Multiple incremental additions to the same font are possible, such as a user agent updating a font as a user browses multiple pages.
The @w3c WebFonts working group now has a fully functional Demo of Incremental Font Transfer which compares IFT to normal font loading with Unicode-range static subsetting. Just click on "next text sample" to see the font being upgraded in real time to support more writing systems and more font variation axes. There is a running total of bytes transferred, you can see that IFT loads much less data.
Both CFF2 and glyf fonts are used in the demo.
The @w3c WebFonts working group now has a fully functional Demo of Incremental Font Transfer which compares IFT to normal font loading with Unicode-range static subsetting. Just click on "next text sample" to see the font being upgraded in real time to support more writing systems and more font variation axes. There is a running total of bytes transferred, you can see that IFT loads much less data.
Both CFF2 and glyf fonts are used in the demo.
For the enthusiastic, who want to go beyond a canned demo: it is all on GitHub
https://github.com/garretrieger/ift-demo
There is a client (in rust, transpiled to wasm) with brotli, harfbuz and woff2 support; the sample texts are in a json file so you can edit them or add your own.
Initial font subset, and the various patches, use a mixed (glyph and table keyed) encoding as described here
https://github.com/w3c/ift-encoder/blob/main/docs/experimental/closure_glyph_segmentation.md
and the implementation is also on GitHub
The @w3c WebFonts working group now has a fully functional Demo of Incremental Font Transfer which compares IFT to normal font loading with Unicode-range static subsetting. Just click on "next text sample" to see the font being upgraded in real time to support more writing systems and more font variation axes. There is a running total of bytes transferred, you can see that IFT loads much less data.
Both CFF2 and glyf fonts are used in the demo.
