Cloudinary Blog

Which Image Compression Technique Looks Best to Human Eyes?

Cloudinary Uses Scale API to Focus on Image Compression

Here at Cloudinary, we provide a cloud-based tool that enables our users to compress images and video for their websites and apps. Our goal is to preserve the visual integrity of the content, but deliver the smallest file size to any device or browser to ultimately optimize website performance and end user satisfaction.

One of the hallmarks of the Cloudinary solution is the ability to automate many functions of image compression, so that developers don’t have to spend time tweaking each photo and making multiple copies of different sizes and resolutions to fit every possible scenario. Compression algorithms can be tricky because they’re trying to make changes that have the smallest visual impact, but different images can react differently to compression.

As we were developing the algorithm for our “q_auto” capabilities – which strikes a balance between visual quality and file size – we needed to test it to understand how the resulting images compared to the human eye. Enter Scale API.

Many image compression formats – like JPEG 2000 and JPEG XR – have been tweaked to score well on particular metrics, such as peak signal-to-noise ratio (PSNR). But these don’t always correlate with human perception on image quality.

We leveraged Scale API to compare pairs of images and give us perspective on which image was liked most by humans. With Scale API, we did a variety of tests, comparing several formats, including WebP, JPEG 2000, JPEG XR (lossy) Lepton (MozJPEG, recompressed with Lepton), FLIF, BPG, Daala, and PNG8 (pngquant+optipng). We also were able to get input on the difference between the uncompressed original image vs. a compressed version.

Scale API enabled us to create A/B comparisons that were viewed by human observers. We submitted over 4,000 image comparisons to Scale API, sending at least four independent Scale API requests for each pair of image. This resulted in at least eight actual human comparisons for each pair of images. The outcome of these comparisons were evaluated beside other perceptual metrics such as PSNR, Google’s Butteraugli, DSSIM (Structural (Dis)Similarity) and a new metric Cloudinary developed called SSIMULACRA (Structural SIMilarity Unveiling Local And Compression Related Artifacts).

The results showed that overall, PSNR is “correct” in only 67 percent of the cases. Butteraugli gets it right in 80 percent of the cases, and DSSIM in 82 percent of the cases. Our new metric, SSIMULACRA, agrees with human judgments in 87 percent of the cases. Looking just at the high-confidence human judgments, we found about 78 percent agreement for PSNR, 91 percent for both Butteraugli and DSSIM, and almost 98 percent agreement for SSIMULACRA. You can read more about SSIMULACRA and these results on the Cloudinary blog. Or if you want to give it a try: SSIMULACRA is free and open-source software!

The results of Scale API comparisons gave us useful data points to validate our metrics and provided more insights into the compression benchmarks we are running and the comparison of various image formats. And from these insights we were able to improve our visual perception metrics and fine-tune our “q_auto” functionality so we know how aggressively we can compress images.

Through this process we were impressed not only by the useful data points derived from the Scale API, but also the great support we got from the company and the product’s ease-of-use, all which came at a reasonable price.

This was originally posted on Scale API

Recent Blog Posts

Reimaging DAM--The Next-Gen DAM for Marketing & Dev

There are great digital asset management (DAM) products out there for uploading, storing, managing, organizing, and sharing digital assets. With Cloudinary's new end-to-end DAM solution, you can also upload and manage your assets efficiently, but the journey doesn't end there. It continues on to the development and delivery stages, so that your assets can be seamlessly manipulated, optimized, and delivered to create an engaging user experience that will in turn, increase conversion and loyalty.

Read more
Integrating Cloudinary with Forestry’s Media Library

At Forestry, we believe that there is a bright future for static HTML sites built with tools like Jekyll and Hugo. These tools can create sites that run well, and are easy to host and maintain, because they don’t require any server-side code.

Read more
Video Optimization With the HTML5 <video> Player

Lack of experience and compression knowhow can cause significant user-experience problems. For instance, on a major retail site, I recently ran into a 48 MB video-hero banner. Pulling out the video and encoding it as an H.264 MP4 reduces the size to 1.9 MB. So, despite the desire for more video content, developers have not yet caught up to best practices. How do we get the best of both worlds without creating a disaster like the one above?

Read more
Build a Facial Emotion Recognition Based Video Suggestion App

Developers are always looking for new and creative ways to deliver content that resonates with the way users feel. Often using the latest technical innovations the market has to offer such as Artificial Intelligence (AI) and Machine Learning (ML). What better way to demonstrate innovative uses of these technology in a consumer market than capturing expressions from your users and then serving content based on that expression!

Read more
Improve Customer Data Protection with GDPR Implementation

TL;DR

Yay! We've done it! Gold-Star for us! We've talked with all the people, made all the changes, paid all the lawyers and checked all the boxes. GDPR? ✅Done!

Not so fast. Of course, conforming to the GDPR regulations introduced in Europe is just the beginning. This is a process and a state of mind that must become part of our long-term cultural ethos.

Read more