Advanced software for Mac OS X.
Developing technologies in order to save archive space on digital discs and also saving on bandwidth transmission (upload and download) is a must today.
Compression is with no doubt the philosopher’s stone of information technology. Companies pay billions for a good compression algorithm. To compress data means saving money in storage expansion and the web, it also means a faster data transmission.
INTRODUCTION:
Images and films are certainly the most cluttered digital objects but because of the extraordinary development of today’s communications network, we are able to access these files as well, by means of mail or web interactivity.
Unfortunately despite the development of more efficient compressors both Lossy (with information loss) and Lossless (no information loss), images and films still continue to weigh too much.
Ex. Image in TIFF or BMP format, 712 pixel (x) and 576 pixel (y) should weigh around 1.12 MB whereas a non-compressed movie in avi format, 720 pixel (x), 576 pixel (y) of 4/5 second should weigh 125MB.
THE APPROACH
When analysing the market it was possible to identify the most popular standard formats.
Photo and graphic images.
3 main formats:
- Original, not compressed
- Compressed with no loss – TIFF with LZW or PNG
- Compressed with loss of data – JPEG or JPEG 2000
- GIFF instead is an exception as it reproduces any kind of image reducing the entire colour space at only 256 triplets of Red, Green and Blue therefore positioning itself in Lossy formats.
Films
- Lossless formats – YUV soft and MSU being the most efficient
- Lossy formats – resulting from JPEG photographic basic approach such as Mpeg (frame by frame compression) and Mpeg2 (with estimation motion intra frames x dvd).
- Mpeg4 (internet) and specific formats such as H.264, x264 or DivX, Xvid.
SETTLING FOR THESE STANDARDS?
The work carried out by several development centres has been extraordinary and very efficient, but unfortunately especially in the case of web visualisation, or in the exchange, it was evident that in order to publish films, for example, you had to reduce the dimensions in a drastic way (320x240) but still with continuous blocking during the reproduction or as it happens on Google and Youtube the geometrical dimension is slightly bigger because of the enlargement of the video (zoom) with really poor results.
A PHILOSOPHICAL APPROACH
The primary idea of respecting as much as possible the visual quality with a higher compression has been badly supported by available literature furthermore it has found its self in the position to fight against the protectionism of the lobbies of big companies and already existing patents.
Finally with the release on the market of the iPod, the iPhone and its similar products there has been a total drastic change of needs, being in first instance, the primary necessity of greater compressions.
It was fundamental to find solutions that allowed the whole system to benefit from movie, photographic files exchanges. These being first of all concrete solutions, with low cost bandwidth, capable to archive more objects without loosing the very important quality aspect. A must for the personal pleasure of the final users!
HOW?
Two solutions where left: one being modifying the existent compression models or creating something absolutely new and original. At the same time an important question had to be answered: Why not utilise the important work achieved until now without bothering anyone especially the ones with existing patents?
Good movie compression utilise 4 important instruments
a. Reduction of space colour: for example from RGB (depth: 3 colours per pixel) to YUV where depth is 3 (Y+UV) one pixel and 1 colour (Y1) the successive pixel (reusing UV).
b. Estimation motion (intra frames) in order to eliminate the repetitions between contiguous frames.
c. Spatial correlation with elimination of the highest spatial frequencies that considered irrelevant to the human’s eye. (Forward Discret cosine transformation or other).
d. Final lossless compression (with no data loss) with entropy or mathematic coder (Huffman or other).
Notes:
1. The number of colours significantly weighs upon the memory activity. The first rule to follow is the one to save the img with the minimum of colours compatibly with the residual quality vision.
2. Every signal presents a certain grade of internal redundancy; therefore reducing such redundancy it is possible to achieve a more compact representation.
3. We have two different kinds of redundancy:
3.1 – Statistic redundancy: Usually the value of a pattern is very similar to other close patterns. In other words, a new pattern shows a certain dependency to the neighbouring ones, hence it is partially predictable.
3.2 - Psychophysics Redundancy: video signals normally contain more information that a normal person can actually appreciate. For example; the human’s being eye is less sensible to high spatial frequencies, especially when considering colours. This means that it is possible to eliminate such uncalled information maintaining the same subjective quality.
In the light of such claims we realised that the optimisations already achieved in points 2 and 3 indicated the way to proceed in order to find new effective solutions.
The number of colours significantly weighs upon the memory activity. The first rule to follow is the one to save the img with the minimum of colours compatibly with the residual quality vision. To which we would like to add the colour range reduction in order to achieve more efficiency.
The first method of progressive “soft” reduction of space colour was born (R,G and B and/or YUV), both for quantities and for values. This clearly means that less colours = less necessary bit.
More complicated is to understand that less value (i.e. the max normal digital value being 255 is reduced to the new value of 190 meaning that the successive elimination system works more efficiently.
Even in the case of images it’s not anymore necessary to eliminate with a specific quantization table the higher spatial frequencies. The final result is a better compression, without modifying any productive process compression but simply essentially “feeding” the entire system.
Two technical ways of proceeding where to be considered in order to solve the problem:
1. Eliminating the light from the image, hence reconstructing the light in the decoding phase. (It’s like switching off the light in a room that is not utilized only to switch on when needed). Obviously the object with no light presents itself to every standard compressor as more compressible (less bits).
2. Using an inversion to the colour negative, with an internal reduction both of quantities and of values of the single colours. Also in this case the compressor receives less data that is better organised for the final compression. Note: this second scenario has the advantage of creating and defining an equation in the coding phase that is different (not directly inverse) to the one of decoding.
PRE-FILTER COMPRESSION CONCEPT
The very simple idea to create a universal pre-filter was born, being capable of preceding any kind of compression, without intervention or/and modification of third parties. Simply by leaving everything as it stands but allowing just a moment before the compression a pre-calculation in order to prepare the material to the compression.
THE APPLICATIONS
Photo and movie lossless (new format: CPD)
• Patent http://www.faqs.org/patents/app/20090067733
Photo and movie Lossy (new format: JPD)
• Patent http//www.faqs.org/patents/app/20090060324
Enhancement quality vision images:
- Patent http//www.faqs.org/patents/app/20090073182
Applications for photo and movies have been made possible with the creation of specific components for Apple QuickTime player and its pro version.
The component allows the exportation both of photo, movies (with the new formats or in the original format Apple PICT) and to the .mov container (QuickTime movie) with all the available pre-filters such as: lossless, Lossy frames by frames, but also with the pure pre-filter to the x.264 and H.264 compressions.
THE RESULTS
In the movies both with ways 1 and 2 the average advantage is around 30% in the case of re-compressions.
In the photos:
CPD vs. lossless PNG format: 30% reduction on size
JPD vs. Lossy JPG/JPEG2000 format: 25% reduction in size with the same PSNR
Apart from one last technology for graphic images, at the moment in its patent release phase, this new format GPD increases the performance both visual (completely lossless) and of weight of graphical, flat or design images (logos) even when compared to the GIF format.
CONCLUSION:
1. JPDM is a METHOD, therefore it is not a standard (as a matter of fact JPDM is a independent pre-filter for Lossy compression) and it’s not a format (the utilized container is actually the already in use .mov)
2. JPDM operates on the following rule: The number of colours significantly weighs upon on the used memory space. Consequently the first rule to follow is the one to save the image with the minimum quantity of colours without altering its vision quality. Furthermore we have added a new criterion, which is the one to of reducing the colour values.
Concluding JPDM establishes a different new reduction criterion of both number and values, therefore allowing an independent and preventive action (pre-filter) in relation to normal standardized compressions. In the decoding process, as a matter of fact, the post-filter operates after the decompression.
The theoretical construct, on which JPDM is based, is a universal concept capable of "feeding" the standard compressors with a more compressible TANGIBLE colour space therefore, once rebuilt, similar to the original image. After many long evaluation tests (achieved with different platforms and compressors) the result is as by scientific guidelines (protocols) stable, predictable and repeatable.
**Finally: this new method can with total simplicity respond to the primary need of saving storage space, bandwidth and energy consumption, these being very important factors especially today because of a worldwide industrial and financial crisis situation
JPDM (pre-filter for Video Lossy Compression) works in a corresponding manner reducing both quantities and values of the single colors (independently from the chosen space color).
The proportional reducing (quantity and values) of the space color allows two various approaches: dark method pre-filter and directed inverse and hybrid method pre-filter with inverse different.
The validity of these technologies has been released by Bill Plummer: www.wtpoptics.com
Technologies developed by RGBLight www.mediadefiner.com
 |
 |
Same-page cross-link controls:
Panel 1 | Panel 2 | Panel 3 | Panel 4 | Panel 5 | Panel 6 | Panel 7
