Scalable wavelet images for fast loading on web live demo

If you ever looked into digital images technologies you must have heard about wavelets and in particular about jpeg2000. The main advantage of images encoded using wavelet transform is their spacial scalability meaning one can take just a small chunk of data at the start of the image file and produce out of it an image that resembles the original one but with lower resolution. Then one can take (download over network) another piece of data and after decoding and placing it on top of the first result get a better resolution. And so on – the process similar to putting transparent Russian dolls painted with finer details from smaller to bigger onto each other.

The great thing about it is that you can use it on the web to enhance user experience by gradually showing improved image rather than making your visitors wait for whole thing to load. So that your users see the preview almost immediately and not the blank space, which is especially applicable to big images on slow connections.

Apart from being perceived as loading faster wavelet images for web can help a great deal in generating thumbnails or simply images of different sizes. That is you can get a thumbnail out of the image by generating a byte range request or transferring only the necessary part over the web sockets. And again at a later point you can decide to transfer some more data and show a bigger image.

Unfortunately scalable wavelet images and jpeg2000 never made it to the web... until now. Below is our new product – fast loading scalable wavelet images for web, which is at the current state demonstrate good quality and decoding time and can already be practically used. And we are working on a new universal web video based on this technology that will give us highly sought-after true scalability and allow us to use a single video file or a video stream for all devices.

If you are familiar with cosine or wavelet transforms you are probably acquainted with Lena - the reference image used in tests. Meet Olga.


Please wait...