> Doing a video capture is not a dump. That's the point. A dump of analog data should > be representative of the original analog medium, which in this case is an electrical > video signal. A Rec. 601 translation is not a dump of the data. It is a conversion to > a modern format, and it is not reversible. Rec. 601 filters and samples the signal, > which is exactly what we will have to do, but it also demodulates the signal and > applies a comb filter, which, primarily for flexibility, is something we should do > digitally with shaders. > > It's similar to why I don't consider samples to be adequate in place of discrete > audio. We need to get as 'close to the source' as practical. Sampling the video > signal is certainly practical (admittedly expensive, though only one person needs to > do it). If the direct laser thing turns out to be practical (and reliable) we should > do that instead.
OK, let me pose this question in relation to the above: does this mean that MAME should also emulate the laserdisc players in their entirety for the games that used them? Digitised disc data could be fed to a virtual laser pickup, the VBI decoded within the LD player virtual machine, and the appropriate frame data fed from the virtual player's video output to MAME's framebuffer for display.
Granted, I am being slightly silly in asking that question, but not entirely. Given that we're talking about reproduction of an analogue medium in a digital format, we're always going to be stuck with the 'how far do we go' and 'where do we stop' questions.
I do actually agree with your points re: a dump vs. conversion and getting as close to the source as possible, but it is possible to agree on the storage method first, then worry about the finer points of reproduction later. The use of samples are a good example of this: it allowed placeholders (samples) to show what should happen until it was possible to make it happen more accurately.