Lepton Image Compression Achieves 22% Lossless Compression of JPEG Images (on Average)

Dropbox stores billions of images on their servers, most of them JPEGs, so if they can reduce the size of pictures it can have a big impact on their storage requirements, so the company developed Lepton image compression, which – on average – achieved 22% lossless compression on the images stored in their cloud.

Lepton_CompressionCompression and decompression speed is also important, since the files are compressed when uploaded and uncompressed on the fly when downloaded so that the complete process is transparent to the end users, who only see JPEG photos, and the company claims 5MB/s compression, and 15MB/s compression, again on average.

The good news is that the company released Lepton implementation on Github, so in theory it could also be used to increase the capacity of NAS which may contain lots of pictures. So I’ve given it a try in a terminal window in Ubuntu 14.04, but it can be built on Windows too with Visual Studio:

If everything goes well for the last step, all tests should be successful:

I also installed it in my path with:

Now let’s go to some directory with photos I took with a DSLR camera:

44 pictures totaling 264 MB. I’ll compress them all, but first, let’s try with one to check the size difference and see if it is indeed lossless.

The Lepton file is definitely smaller (21.66% smaller):

Now let’s uncompress the file and see if there’s any difference:

The diff did not generate any output, so the compression is indeed lossless.

Time to compress all 44 photos:

I did so on a machine with an AMD FX8350 processor, and 8 cores were used during compression. The command took 5 minutes and 8 seconds, or about 7 seconds per picture. What about the size?:

That’s 207 MB down from 264 MB, or about 21.6% compression.

Via Phoronix

Share this:

Support CNX Software! Donate via cryptocurrencies or become a Patron on Patreon

ROCK Pi 4C Plus
Notify of
The comment form collects your name, email and content to allow us keep track of the comments placed on the website. Please read and accept our website Terms and Privacy Policy to post a comment.
6 years ago

The blog post explaining the algorithm is interesting since it works with predictions so I would assume it’s optimized for JPEGs spitten out of mobile phones and DCams (this will make up 99.99% of images stored on DropBox). Will test that later with JPEG images created with optimized encoders we use in various workflows (it’s always a trade-off between speed, image quality and size). BTW: It should read just ‘./configure’ (without .sh here) and compilation fails on ARM: In file included from src/vp8/decoder/boolreader.hh:33:0, from src/vp8/decoder/boolreader.cc:15: src/vp8/decoder/../model/numeric.hh:11:23: fatal error: smmintrin.h: No such file or directory compilation terminated. Makefile:1814: recipe for target… Read more »

6 years ago

That was very informative. I’ve been hoarding so many images, that I’m compelled to try it out as well! Many thanks.

5 years ago

Should now work on other platforms than x86 (SSE4 instructions used) when doing