Cloudinary image leader: developing the 'next' JPEG

This is a guest post for the Computer Weekly Developer Network written by Dr Jon Sneyers, senior image researcher at Cloudinary.

Cloudinary is a SaaS technology company headquartered in Santa Clara, California, with an office in Israel. The company provides a cloud-based image and video management solution to enables users to upload, store, manage, manipulate and deliver images and video for websites and apps.

Sneyers writes…

Last summer I had the opportunity to participate in my third JPEG meeting of the year, the 84th JPEG meeting in Brussels, Belgium, conveniently close to where I live.

These are very exciting times for those of us involved in the world of standards and image formats — and especially for those of us directly involved in the development of what will soon be a major update – the first in about 20 years – to the widely adopted JPEG.

A little history for those less familiar with image formats: JPEG stands for the Joint Photographic Experts Group, which created the standard in 1992. The main basis for JPEG is the Discrete Cosine Transform (DCT), a “lossy” image compression technique that was first proposed by Nasir Ahmed in 1972.

How did we get here?

You might be wondering how we got here and why is a new image format necessary?

Last spring, the JPEG Committee launched its Next-Generation Image Compression Call for Proposals, also referred to as JPEG XL.

The goal was to: “Develop a standard for image compression that offers substantially better compression efficiency than existing image formats (e.g. >60% over JPEG-1), along with features desirable for web distribution and efficient compression of high-quality images.”

My submission, which I dubbed FUIF or the Free Universal Image Format, was one of seven proposals selected to be presented at the 81st JPEG meeting, held in Vancouver, Canada.

That was the first JPEG meeting I participated in, presenting and defending my proposal.

Once all seven proposals were presented we spent the week discussing the pros and cons of each proposal, the subjective evaluation results, and how to get started on JPEG XL.

At the 82nd JPEG meeting in Lisbon, Portugal, it was decided to combine two proposals: my FUIF and Google’s PIK, and make that the starting point for JPEG XL.

Use cases & requirements

At that point, I was promoted to the role of co-chair of the JPEG XL Ad Hoc Group, together with Google’s Jan Wassenberg. The main discussions were on refining the use cases and requirements… and, on how to combine and unify the two proposals into a single image codec.

At the next meeting, held in Geneva, Switzerland, we started writing the formal standard – it was still a very rough ‘working draft’, but it was a good start.

Gradually we transitioned from a situation where the code was the ‘source of truth’ (and the spec draft was describing whatever the code did) to a situation where the spec became the source of truth and the code had to implement what the spec was saying.

The road to ISO

At the 84th JPEG meeting, we took JPEG XL to the next stage, which is ‘Committee Draft’. This is one of the first steps in the ISO standardization process. The next step will be DIS (“Draft International Standard”), and at the end – if all goes well – it will be an International Standard, with the nice number ISO/IEC 18181.

While most of the development work is done now, there is still a lot of tweaking, experimentation and evaluation to be done to make JPEG XL the best image (and animation) codec it can possibly be.

We hope it will become as successful as JPEG, and gain widespread adoption – eventually replacing JPEG, PNG and GIF. To help that transition, we made sure that existing images in one of those three formats can be converted to JPEG XL in a pixel-exact way with guaranteed compression gains.

New JPEG XL encoded images will need only one third of the bytes JPEG needs to reach the same perceptual quality.

We’ve come a long way since the first format and I think most of us can appreciate what these savings and quality of improvements means for the ‘visual web’. These are very exciting times indeed!

About the author

Dr Jon Sneyers, senior image researcher at Cloudinary invented the Free Lossless Image Format (FLIF). His image processing “Muscles from Brussels” help deliver super-optimised images, super-fast.

Cloudinary’s Sneyers: pumped up developing the ‘visual web’.

Data Center
Data Management