act_thumbhash_flutter

Releases

1. ▸▾1.1.01 Aug 20262 sources agree

   Raises the minimum supported SDK of both packages: act_thumbhash and act_thumbhash_flutter now require Dart 3.8 (Flutter 3.32). There are no functiona

   Raises the minimum supported SDK of both packages: act_thumbhash and act_thumbhash_flutter now require Dart 3.8 (Flutter 3.32). There are no functional changes; 1.0.0 remains available for older toolchains.

   act_thumbhash 1.1.0

   • The minimum supported SDK is now Dart 3.8.

   act_thumbhash_flutter 1.1.0

   • The minimum supported SDK is now Dart 3.8 / Flutter 3.32.

   Open source →
2. ▸▾1.0.01 Aug 20262 sources agree

   First stable release of both packages: act_thumbhash and act_thumbhash_flutter .

   First stable release of both packages: act_thumbhash and act_thumbhash_flutter.

   act_thumbhash 1.0.0

   • Perf: the encoder and decoder now precompute their DCT basis tables and skip coefficients the format discards, making encoding roughly 9x and decoding roughly 4x faster. Output is byte-for-byte unchanged.
   • Fix: decodeSync now validates the hash length against what its header requires and throws the documented ArgumentError for truncated input, instead of surfacing an out-of-bounds RangeError partway through decoding.
   • Fix: decodeSync and decodeAsync now reject baseSize values below 1 with an ArgumentError, and small baseSize values can no longer round the decoded image's short side down to zero pixels.
   • Fix: encoding images with an aspect ratio steeper than 128:1 no longer throws; the internal downscale now keeps the short side at a minimum of one pixel.
   • Feat: Web support. dart:isolate is no longer imported unconditionally; on the web (both dart2js and dart2wasm) encodeAsync and decodeAsync run the computation on the event loop after yielding once, since isolates are unavailable there. Native behaviour is unchanged.
   • Fix: encodeSync emitted too few luminance coefficients for images with an aspect ratio steeper than roughly 2.8:1 (2:1 with alpha), in either orientation. The resulting hashes were shorter than any decoder expects, so decoding them threw a RangeError, and they were not readable by other ThumbHash implementations. The luminance channel is now always encoded with at least 3 coefficients per axis, matching the reference implementation.
   • Fix: the hash buffer was one byte too small for images with alpha whose coefficient count is odd (for example 100x75), making encodeSync throw a RangeError while writing its own output.

   Hashes for images that encoded correctly before are unchanged.

   act_thumbhash_flutter 1.0.0

   • Feat: ThumbHashImageProvider now exposes baseSize, controlling the length of the decoded placeholder's longer side (default 32, matching the previous behaviour). Available on the constructor, fromBase64, and the toImageProvider() extension, and part of provider equality.
   • Feat: Web support. Provider equality no longer relies on Uint64List, which dart2js does not support.
   • Fix: engine-level image decoding failures now surface as image stream errors instead of leaving the stream waiting forever.
   • Fix: comparing providers whose hash bytes are a view into a larger buffer at an offset not aligned to 8 bytes no longer throws.

   Both packages now support all Dart and Flutter platforms, including the web.

• act_thumbhashpub.devA pure Dart implementation of the ThumbHash image placeholder algorithm. Encode images into compact hashes and decode them into lossy previews.