Amazon has begun converting the images inside Kindle ebooks to AVIF, the format derived from the AV1 video codec. The rollout has drawn some attention from publishing and developer communities because it represents one of the largest single deployments of AVIF as a container for static images — happening not in a browser or a photo-sharing app, but inside downloadable ebook files. It’s worth pausing on what that choice actually means at the file level, because AVIF is doing things that JPEG simply cannot.
What AVIF Is, and Why It Compresses Differently
AVIF is a still-image format built from AV1 intra-frames — single frames encoded using the same inter-prediction and transform tools the video codec uses for independent keyframes, just applied to a standalone image rather than a sequence. The practical consequence is that AVIF’s encoder can describe smooth gradients and continuous-tone regions (skin, sky, covers with photographic backgrounds) with far fewer bits than JPEG’s discrete cosine transform can manage at equivalent quality settings.
JPEG works by slicing an image into 8×8 pixel blocks, applying DCT to each block, and then quantizing the resulting frequency coefficients. At high compression, those block boundaries become visible as the familiar JPEG blocking artifact — the slightly stepped or buzzy edges you see in heavily compressed thumbnails. AVIF’s intra-frame coding operates over larger, variable-size prediction units, which spreads gradient information more efficiently and produces ringing artifacts (when they appear) that are usually less structurally obvious than JPEG blocking.
For ebooks, that distinction matters in a specific way. Book cover thumbnails, author photos, and interior illustrations are often displayed at small sizes on e-ink and LCD screens. A small image that was aggressively JPEG-compressed to keep file size down shows blocking plainly when the reader zooms in or the device renders it on a high-DPI panel. A comparably sized AVIF encode can hold together better at those small dimensions without the same block structure becoming visible.
Color Depth and HDR Handling
JPEG is limited to 8 bits per channel in its standard form. AVIF natively supports 10-bit and 12-bit color depth, and it handles the HDR transfer functions — PQ (Perceptual Quantizer) and HLG (Hybrid Log-Gamma) — that are used in wide-gamut display pipelines. For most Kindle content, that’s presently less relevant: e-ink displays are not HDR-capable, and even color Kindle panels are not wide-gamut in the sense that a modern phone screen is. But publishing workflow files that pass through conversion pipelines benefit from a format that doesn’t clip or round-trip color information unnecessarily, even if the end display doesn’t exploit the full range.
AVIF also handles transparency via an alpha channel encoded in the same bit depth as the image data itself. JPEG has no native transparency. For publishers compositing images against variable backgrounds — or for text callouts with soft drop shadows — this is a meaningful capability difference, though how Amazon’s conversion pipeline uses it for Kindle content specifically isn’t fully documented.
What Changes for Readers and Publishers
For most Kindle readers, AVIF images are invisible as a format. The Kindle app or device handles rendering; readers do not interact with raw files. The change only becomes noticeable if file sizes shift detectably (smaller books download faster on metered connections), or if older reading software fails to render the images — which is a real concern. AVIF support in software is not universal, and check current support before relying on any specific compatibility claim: the ecosystem has been moving, but older Kindle firmware versions and third-party ebook readers may not have AVIF decode capability.
Publishers and ebook developers working in the EPUB or KF8/KFX space are in a different position. If Amazon is re-encoding images at the point of ingestion or during format conversion on their side, publishers don’t necessarily need to supply AVIF themselves — but understanding what the pipeline does to source images matters. A high-quality PNG or JPEG supplied by the publisher will be re-encoded. If the AVIF encoder Amazon uses applies lossy compression (and it almost certainly does for photographs — lossless AVIF exists but produces larger files than lossless PNG for many image types), there is a generation loss: information present in the original that was not in the AVIF-encoded quantized output is simply gone.
That’s not a criticism specific to AVIF. It’s the nature of any lossy encode. Re-encoding from a high-quality JPEG source does introduce an additional quantization step, compounding prior loss. Publishers who care about image fidelity in the final Kindle product should supply the highest-quality source files their workflow allows, since every lossy encoding step discards data that cannot be recovered.
For context on how format choices interact with compression decisions more broadly, our File Formats coverage addresses the underlying mechanics across a range of formats.
The Broader Signal: AVIF Arriving in Unexpected Places
Browsers have been the primary deployment surface for AVIF in public conversation, with support rolling out gradually across Chrome, Firefox, and Safari over recent years (check current compatibility for any specific version before relying on a given browser rendering AVIF reliably). The Kindle move suggests the format is now mature enough for platform operators to use it in large-scale, offline, download-and-render contexts where they control both the encoder and enough of the decoder ecosystem to be confident images will actually display.
That’s a different kind of deployment than a website that can serve fallback formats via <picture> elements to browsers that don’t support AVIF. Amazon is essentially committing that the Kindle platform will render these images correctly — a statement of confidence in the format’s stability and the coverage of their own decode support. The fact that they made this choice quietly, without a public specification change or compatibility announcement aimed at readers, implies the engineering team considers render support sufficient to proceed without warning users.
The AVIF format is derived from an open, royalty-free codec (AV1 itself is developed by the Alliance for Open Media), which removes patent licensing concerns that historically complicated JPEG 2000 adoption. That mattered for large-scale platform deployment decisions, and it’s part of why AV1 and AVIF gained traction where earlier technically superior formats stalled.
What to Watch For
If you work on ebook production or digital publishing, a few things are worth tracking as this develops:
- Source file quality: Supply the best-quality originals you can. If Amazon’s pipeline applies lossy AVIF encoding, higher-quality inputs mean the output holds more of the original detail.
- Older device compatibility: Readers on older Kindle hardware or third-party EPUB readers that haven’t updated AVIF decode libraries may see broken images. Monitor reader feedback, and check Amazon’s current device compatibility documentation for any guidance they publish.
- Lossless vs. lossy encoding behavior: Amazon has not publicly detailed the exact quality parameters they apply. If image fidelity in specific titles is critical (art books, photography collections, illustrated children’s books), it’s worth reviewing how converted files actually look on device rather than assuming the encoder will apply conservative settings.
- Metadata handling: AVIF supports Exif, XMP, and IPTC metadata. Whether Amazon’s conversion pipeline preserves or strips that metadata from ingested images is not clearly documented — relevant if you’re embedding accessibility or rights information in image metadata.
The practical action for anyone working in ebook image pipelines right now: pull a converted Kindle file, extract the image assets, and inspect what the AVIF encoder actually produced. Format and quality decisions are checkable at the file level, and that’s where the real answer to “how much did this change?” lives.