Mozilla introduces more efficient jpeg encoder

Mozilla has introduced a more efficient encoder for jpeg images. The encoder is compatible with existing decoders and could make images on average five percent smaller. Facebook is already testing the new encoder.

Both images saved in the baseline and progressive jpeg formats are on average five percent smaller, Mozilla promises. Many images will be even smaller thanks to the Mozjpeg 2.0 encoder, according to the browser maker. The open source encoder is compatible with existing jpeg decoders, eliminating the need, for example, to update browsers to load the new, smaller jpegs. The improvement is mainly due to trellis quantization, an algorithm that is especially suitable for the compression of images and that is also used in Xvid and h.264, among others.

Facebook has already started the first tests of the new format. In addition, the social network has donated $60,000 to enable the development of Mozjpeg 3.0. Last year, Facebook requested Mozilla to support webp, a relatively new image format from Google. Mozilla has not yet responded to this, although the request has been considered. In 2011, Mozilla said it had no plans for support yet.

Research conducted by Mozilla itself has shown that jpeg is in some cases more efficient than webp when considering the number of bits required per pixel. The new standard high-efficiency video coding, which is used in the successor to h.264, is the most efficient in all cases, but support for still images with hevc compression is still minimal.