Originally Posted by coyoteaz
What do you mean by "can't mess with it while in the digital domain"? HD is normally completely digital from the ADC on the CCD (or other image sensor) all the way until the output of the decoder, and possibly all the way to the photons coming out of the display device. There's obviously quite a bit of manipulation that happens between the camera and the TV (or the DAC on your receiver), and it's all digital. It's even possible to manipulate video in the compressed domain as demonstrated by the splicers used by Fox stations, the Motorola (formerly Terayon) CherryPicker, and the Harris (formerly Leitch) DTP.
It's a fine distinction. "Mess with" was probably too broad a term, but I was trying not to be too technical for the benefit of those who's eyes glaze over when I do that.
But I guess that boat has sailed. What I mean is that once something is digitized and compressed, and if you assume that the transport that carries it continually provides enough bits to do the job, the typically "analog" aspects of video or audio such as the volume level, chroma level, resolution, etc., are now totally represented by binary coefficients. Unlike analog signals, the nature of which can change during transport due to any number of outside influences, the only thing that can change a stream of binary coefficients is if you were to perform a mathematical operation on those numbers to change them into different numbers. Otherwise, they have no choice but to remain exactly the same.
For instance, to change the level of a digital audio stream while still in the digital domain, you have to add the same number to each coefficient. Since there are 48,000 of them each second, that takes human technical intervention. Without that, there is no change, no loss of highs, no muddying of bass, essentially no change. If coefficients are unchanged during transport, then they are identical to what they were just after being compressed, all the way to the dedoder and final D-to-A.
That's why we have moved to digital transport in the first place, to preserve PQ and prevent it from being changed by the ravages of the hostile environment it must travel through. Identical TV receivers 1 mile from the transmitter and 50 miles from the transmitter will have the exact same PQ, even though the transport environment is very different, because digital signals are immune to being changed by natural phenomena.
So HD OTA signals from Cox look identical to those from OTA. Sure they "mess with" a lot of stuff, demodulating from 8VSB to baseband, demuxing the main program, converting to VOIP to move it to a remote headend, remodulating to QAM, etc., but none of that means actually manipulating the content itself by changing the ones and zeroes, the numbers that represent it, into different numbers.
DirecTV. on the other hand, converts MPEG-2 to MPEG-4. That actually does change the numbers, and adds rounding errors beyond those caused by encoding to MPEG-2, making them less representative of the original digitized content than they were. Is it visble, or perceptible? I don't think it really is.
The FOX signal changes from uplink to decode, granted. It arrives at the affiliate as a MPTS, is demuxed from the MPTS and remuxed to have local PSIP and Nielsen metadata added, squirted through an MPEG/AC3 splicer, and then reformatted to SMPTE310 and modulated to 8VSB (at a minimum). But again, the content itself stays in the digital domain, and nothing about any of those processes changes the numbers representing the pixels to new numbers, meaning that the PQ is locked until finally decoded in your home.
The packet structure might change, the wrappers change, what accompanies the content may change, FEC info might be added, etc., but all that is important to PQ is that the numbers that represent each pixel do not change, and if done properly, they don't. The decoder could care less what accompanied the content or how it got there. It simply decodes those unchanged binary coefficients into video that is virtually the same as it was just before being digitized way back at the beginning of the transport chain, ensuring that PQ remains untouched (other than the losses from MPEG encoding itself), that is, unless something or someone changes those numbers in midstream, which can only happen by design.
BTW, displays are analog. There is plenty of "100% digital processing", but most of that digital processing is still done to signals which have been converted back to the analog domain. The first thing that happens inside a Sony LCD just after the HDMI receiver chip is a complete conversion of that digital signal to analog YUV.
Our eyes are analog devices, and need to be fed analog signals. Every display is analog at some point even if it is just the last stage of the light engine, but typically signals are converted to analog straight away inside virtually all modern displays, which is a good thing. Think about my example of how to raise the volume of a digital audio stream above. Now apply that to changing the gamma of a video signal, and try to imagine a way to do that without adding severe rounding errors which would compromise the PQ, something that manipulating gamma in the analog domain would not do. Digital is not always the best choice, and typically not once you get the signal safely inside an HDTV.