best essay uk
Opportunities in ProAV for VARs
April 25, 2017
The Facts On Programs Of Crypto Boom
October 13, 2017

Transmitting high resolution content – Make sure you have the best solution!

Defining Color

Long ago researchers acknowledged that color was extremely subjective and personal. The attempt to attribute numbers to the brain’s reaction to visual stimuli was always very difficult. The color space method was created to aid the process of describing color either between people or between machines or programs.

What is the Color Space?

A color space is a method by which we can specify, create and visualize color. People define a color by its attributes of brightness, hue, and colorfulness. The computer describes colors by using the amounts of red, green and blue phosphor emission required to match a color.  A color is usually specified using three coordinates, or parameters. The parameters describe the position of the color within the color space being used.

The YCbCr Color Space

Television transmission color spaces are sometimes known as transmission primaries.
YIQ and YUV were analog spaces for NTSC and PAL systems respectively while YCbCr is a digital standard used mainly in computers and digital TV transmissions. These color spaces separate RGB into luminance and chrominance information and are useful in compression applications (both digital and analog). 

The challenge of transmitting and displaying digital images and video content

The ability to transmit digital images and video content and display them on a screen depends on the quality of the source content, the link bandwidth available to send it through, and the screen display resolution.

At any given time, the biggest challenge has always been the bandwidth available for the transmission. To overcome the lack of sufficient bandwidth, source content needed to be scaled down. Traditionally the process of scaling down the content was performed using color space conversion.

For example, RGB color space that contained the source content with all color elements existing in all the pixels (also known as YCbCr 4:4:4) is converted to YCbCr 4:2:2 or YCbCr  4:2:0 color spaces. In these color spaces, the Y (Grayscale) exists for all pixels, but the Cb/Cr color elements are diluted and do not exist for every pixel. This process created the required shrinking of the content to be able to transfer it over the bandwidth available.  Even though some color information was lost in this process, the visual result displayed on the receiving screen appeared “good enough” for the human eye. (In the early 1900s, researchers found that the human eye is mostly sensitive to black and white colors, somewhat less for mid range colors like yellows and greens and much less for colors of the end of the spectrum – reds and blues. This finding led to color conversions that skipped color information assuming the human eye would not notice the difference. This resulted in an image displaying on the screen that was converted as “good enough”).

Missing information is rebuilt based on the neighboring pixels, and hence the reconstructed picture is corrupted.

So why is Color Space Conversion not “good enough” today?

The common process of Color Space Conversion that takes out some color information  to be able to scale down the content that is transferred (YCbCr 4:2:2, YCbCr 4:2:0)  results in an image that when displayed on a high-resolution screen reveals the loss of color data (contrary to the belief that the process is “good enough” to the human eye).

Furthermore, consumer demand for higher resolution and high pixel density displays has skyrocketed in the last few years. Display manufacturers are pushing high-resolution displays with increased pixel counts which require increased bandwidth over the links driving these displays. Also, advances in the physical layer technology have not kept up with increases in pixel counts and some display drivers’ ICs store display frames. Thus, high-resolution displays require larger on-chip frame buffers. These factors have created the need for compression on the display links themselves.

The answer as defined by VESA - Video Electronics Standard Association

In January 2013, a call for proposals was made for a new standard to work with display interfaces such as MIPI, DSI, VESA, and DisplayPort.  The call for proposals included quite a few requirements to overcome the issues created by high-resolution displays, pixel heavy content, and limited bandwidth links.

The requirements for the new standard included:

In response to this request for a solution, a display stream compression (DSC) encoding algorithm was developed based on delta pulse code modulation with an indexed color history (ICH).  When comparing between YCbCr 4:2:2 / YCbCr 4:2:0 and DSC encoding algorithm results, it is simple to detect that DSC is much more professional and much better than “good enough”. Let’s see why.

Proving that Display Stream Compression (DSC) is superior to Color Space Conversion (CSC) was by testing it.

Two separate transmissions, of the same high-resolution source content were transmitted to a high-resolution display. One transmission was completed using display stream compression (DSC), the other transmission was completed using color space conversion (YCbCr conversion - CSC)

After both transmissions, the result was compared to the source.  An image showing the differences between the source and the displayed content was created so that the differences can be seen in the following table.

 

 

  1. Text Example Original Source vs DSC vs CSC (4K 60 4:4:4 / 8bit)

 

The scale of difference:

DSC difference from Source

DSC difference from Source

CSC difference from Source

CSC difference from Source

Random Pixel Comparison DSC compared to Source Visually Lossless

Random Pixel Comparison DSC compared to Source Visually Lossless

Same Pixel Comparison CSC compared to Source Not

Same Pixel Comparison CSC compared to Source Not "good enough"

Testing visualization was conducted with the comparison tool “Beyond Compare”.

The “image compare” view shows images side-by-side with their differences highlighted.

 

2. Movie/Video Example - Original Source vs DSC vs CSC (4K 60 4:4:4 / 8bit)

The scale of difference:

DSC difference from Source

DSC difference from Source

CSC difference from Source

CSC difference from Source

Random Pixel Comparison DSC compared to Source Visually Lossless

Random Pixel Comparison DSC compared to Source Visually Lossless

Same Pixel Comparison CSC compared to Source Not

Same Pixel Comparison CSC compared to Source Not "good enough"

Testing visualization was conducted with the comparison tool Beyond Compare”.

The “image compare” view shows images side-by-side with their differences highlighted. 

Regarding picture quality, side effects such as blurry text, chroma and luma distortions that appeared with CSC were not experienced with DSC. This proves that the DSC algorithm outperforms many proprietary algorithms. Developers of the DSC algorithm made it a point to pass strict testing. Every type of test content was included - white noise, zone plates, multi burst, high-density subpixel, rendered text, computer, phone, and tablet screen captures, photos and video.

How was the visually lossless performance validated?

Original images and uncompressed images were flipped back and forth in place on the same screen, and then users were asked to identify the difference if they could. The images selected were complex images. In one case study, a company analyzed thousands of images gathered from the internet and selected those that were most likely to have artifacts based on a mathematical analysis of the compressed images.

The studies showed DSC outperformed other proprietary algorithms on these picture quality tests, and that it was either visually lossless or very nearly so for all tested images at 8 bits/pixel.

The soon-to-be-updated HDBaseT standard for 2017 includes a new reference design that addresses the market need for long range transmission of full 4K 60Hz 4:4:4 and HDR-enabled content.  This reference design also addresses the requirements of the HDMI 2.0 specification, which will use visually lossless compression (DSC) to support the higher format resolutions. Other compression solutions will not be certified nor compatible with leading standard displays and projectors supporting 4K 60Hz 4:4:4 with HDR.

More here: [poynton], [vesa dsc], [valens], [beyond compare]