4 releases (breaking)
| 0.4.0 | Oct 16, 2024 |
|---|---|
| 0.3.0 | Apr 17, 2023 |
| 0.2.0 | Mar 24, 2023 |
| 0.1.2 | Jan 26, 2023 |
#58 in Images
440KB
9K
SLoC
Lepton JPEG Compression in Rust
This is a port of the C++ Lepton JPEG compression tool that was released by DropBox dropbox/lepton. We developed a port of the library to Rust, which has basically the same performance characteristics with the advantage of all the safety features that Rust has to offer, due to the work involved in performing an exhaustive security check on the C++ code and the fact that DropBox has deprecated the codebase.
With precise bit-by-bit recovery of the original JPEG, the Lepton compression library is designed for lossless compression of baseline and progressive JPEGs up to 22%. JPEG storage in a cloud storage system is the main application case. Even metadata headers and invalid content are kept in good condition.
How to Use This Library
Some operations of this library are vectorized such as the IDCT using the Wide crate, so you can get a significant boost if you enable +AVX2.
Building From Source
git clone https://github.com/microsoft/lepton_jpeg_rust
cd lepton_jpeg_rust
cargo build
cargo test
cargo build --release
Running
There is an lepton_jpeg_util.exe wrapper that is built as part of the project. It can be used to compress/decompress and also to verify the test end-to-end on a given JPEG. If the input file has a .jpg extension, it will encode. If the input file has a .lep extension, it will decode back to the original.jpg.
It supports the following options:
lepton_jpeg_util.exe [options] <inputfile> [<outputfile>]
| Option | Description |
|---|---|
-threads:n |
Runs with a maximum of n threads. For encoding, this limits the amount of parallelism that can be gotten out of the decoder. |
-dump |
Dumps the contents of a JPG or LEP file, with the -all option, it will also dump the cooefficient image blocks. |
-noprogressive |
Will cause an error if we encounter a progressive file rather than trying to encode it. |
-acceptdqtswithzeros |
Accept images with DQTs with zeros (may cause divide-by-zero). |
-iter:n |
Runs N iterations of the operation. Useful when we are running inside a profiler. |
-max-width:n |
Limit the maximum image width to n pixels, instead of the default 16386. Fails with an error if limit is exceeded. |
-max-height:n |
Limit the maximum image height to n pixels, instead of the default 16386. Fails with an error il limit is exceeded. |
Contributing
There are many ways in which you can participate in this project, for example:
- Submit bugs and feature requests, and help us verify as they are checked in
- Review source code changes or submit your own features as pull requests.
- The library uses only stable features, so if you want to take advantage of SIMD features such as AVX2, use the Wide crate (see the idct.rs as an example) rather than intrinsics.
Code of Conduct
This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.
License
Copyright (c) Microsoft Corporation. All rights reserved.
Licensed under the Apache 2.0 license.
Dependencies
~5–17MB
~181K SLoC