Go to file
2022-01-03 21:57:41 +03:00
.github/workflows Update CI config (reference enabled) 2022-01-03 17:15:18 +03:00
assets Add assets for tests and reference checks 2021-12-29 20:56:07 +03:00
bench Big batch of renames: strip qoi prefix everywhere 2022-01-03 21:40:24 +03:00
doc Add qoi format specification pdf to the repo 2022-01-03 14:25:40 +03:00
ext (Update the upstream qoi submodule) 2021-12-29 20:56:30 +03:00
fuzz Big batch of renames: strip qoi prefix everywhere 2022-01-03 21:40:24 +03:00
src Big batch of renames: strip qoi prefix everywhere 2022-01-03 21:40:24 +03:00
tests Big batch of renames: strip qoi prefix everywhere 2022-01-03 21:40:24 +03:00
.gitignore (Add fuzz coverage to .gitignore) 2022-01-03 21:57:41 +03:00
.gitmodules Add reference implementation as a submodule 2021-11-29 04:21:09 +00:00
Cargo.toml Big batch of renames: strip qoi prefix everywhere 2022-01-03 21:40:24 +03:00
LICENSE-APACHE Add license files 2022-01-03 14:25:40 +03:00
LICENSE-MIT Add license files 2022-01-03 14:25:40 +03:00
README.md Update readme + update bench timings 2022-01-03 20:57:25 +03:00
rustfmt.toml Initial working version (but lots of temp stuff) 2021-11-28 16:20:24 +00:00

qoi-fast

Build Latest Version Documentation Apache 2.0 MIT unsafe forbidden

Fast encoder/decoder for QOI image format, implemented in pure and safe Rust.

Quick summary:

  • One of the fastest QOI encoders/decoders out there.
  • Compliant with the latest QOI format specification.
  • Zero unsafe code.
  • Supports decoding from / encoding to std::io streams directly.
  • no_std support.
  • Roundtrip-tested vs the reference C implementation; fuzz-tested.

Examples

todo!();

Benchmarks

Comparison to the reference C implementation (as of 00e34217), benchmarks timings collected on Apple M1 (1782 images, 1187 MB total):

codec          decode:ms    encode:ms  decode:mp/s  encode:mp/s

qoi-c            4406.63      5515.80        282.4        225.6
qoi-fast         3071.49      4545.08        405.2        273.8

License

This project is dual-licensed under MIT and Apache 2.0.