Yes, this website is quite simple. The technology is a couple years old but I just got started on the website. Please feel free to email with questions/comments.

Latest update: May 11, 2017

Thanks for visiting!

shawn@hifipix.com

The Progressive PNG Development Project

The goal of this project is to develop and commercialize a new way to send PNG images with multiple, high quality and progressively higher resolution images in a total file size less than that of the highest resolution image by itself, supporting both a form of responsive image delivery as well as a faster and higher quality presentation of images compared to the interlaced approach.

Where did this project come from? The basic algorithms for this project are based on Multi-Format Encoding (MFE) – a patented process originally developed to encode UltraWide moves onto Blu-ray discs with both standard and higher, anamorphic resolutions. It then seemed natural to cross develop the technology for internet images. The MFE process involves the use of extra data to add resolution to a given image. These extra data files are actually a type of image themselves that work best in a near-lossless file format so PNG was chosen as the platform.

Where is this project going? This technology will be commercialized as either an SaaS (Software as a Service) or licensed WordPress plugin or some combination thereof to make it easy to convert images to the progressive PNG format for distribution over websites (and wherever else it can be used). Whether this is something HifiPix launches itself or in partnership with someone else is TBD.

The Minimum Viable Product: High resolution images will be processed into a high quality “Base Image” of one sixty-fourth the total original resolution along with “extra data” image files that can be used with a simple JavaScript algorithm to convert each low resolution image into the next higher resolution for four total image resolutions. For simplicity, original images will be assumed to have dimensions evenly divisible by eight.

Status: Algorithms are being optimized in C++.  In June the decode algorithms will be converted to JavaScript so that downloadable demos of the algorithms from the website can be seen.

Yes, it works: Below are actual images using the progressive PNG process. The images are scaled down in the browser to fit on the page but can be clicked for full size. Even though the process offers images in multiple resolutions, the total file size of all images is less than just sending the highest resolution image by itself.

       
The above image is a “Base Image”. The Base Image is the only actual representative image stored on a server for the progressive PNG process. All higher resolution images will be reconstructed from the Base Image using extra data files and a JavaScript decoding algorithm to whatever higher resolution level is desired by the web developer. As an example PNG file this Base Image is 69 KB.
 
     
The above image has been reconstructed from the Base Image to twice the size by adding 186KB of extra resolution data.
 
     
The above image has been reconstructed from the previous image to twice the size by adding 738KB of extra resolution data.
 
         
The above, final image has been reconstructed from the previous image to twice the size by adding 2,829KB of extra resolution data. If sent by itself this image would take 4,129KB instead of the 3,822 KB total of the MFE process. In general, MFE progressive PNG images (even with all the smaller images) require less total file size than if the final, high resolution image is sent by itself because the extra data image files include a limited palette well suited for compression algorithms.