> > IMHO here lies a problem. Nearly all capture cards, where not designed to capture > the > > "whole" frame. Even in the Firefox thread, you will not see such a capture > (starting > > with page 1). Professional equipment will capture VBI data, but not the whole > frame. > > You may achieve this with hacked videodrivers, but this is not recommended and a > > ideal solution at all. I consider it very hard to do it the right way, especially > if > > you want to keep the interlace/fields intact, which would be crucial for further > > processing. To be honest, i could be wrong here as i never seen the original > > captures, but i think the captures from Aaron have errors regarding this topic. We > > only need to know, which lines of the VBI data are needed, but we dont need the > whole > > frame, this will cause only trouble. This is the reason why i asked if the software > > for proving captures, could be rewritten. Professional equipment doesnt work that > way > > (cheap stuff neither), like it is expected in the article of Aaron. So the approach > > of the LD tool in MAME is wrong, regarding the VBI data, as it is expecting the > > "whole" frame (525lines). > > > > Which brings me to the last point. Yeah, you can hack all kind of stuff, like the > IR > > remote for exact frame jumping and what not, capturing straight from the laser lens > > to your eye. I think we should avoid as much as possible of this hacks. The > > interlace/fields, analog sources/noise, VBI data etc. all of this, is very delicate > > and very picky and with a single shifted scanline, you will break all your > capturing > > effort into useless shit. Keep it simple, treat the source right, capture it > > uncompressed/lossless and thats it. > > Could you expand on this please? As I understand it, part of the issue with laser > disc support is the ongoing discussion of accuracy. > > Disclaimer: I'm not presuming to question MAME design decisions, I'm just wondering > about a different path. > > What caught my attention is this: What you're proposing sounds to me like you're > discussing the capture of the parts of the "frame" that are relevant to the game and > disregarding the parts of the "frame" that are present only because of the > requirements of the co-opted technology (Laserdisc). > > I'm wondering if that is an interesting discussion point as MAME doesn't attempt to > emulate all of the intricacies of the underlying technology (Not counting discrete > circuits). MAME isn't attempting to emulate any frequencies outside of the human > range of hearing for example, so I'm wondering if not grabbing unused frame > information might be similar?
There are actually two things that need to be accomplished here, which is a source of confusion.
1) MAME's primary goal is emulating the system. For this we can ignore information that has no effect on emulation requirements. In this sense all we theoretically need for debugging is VBI data and 'images' where images are parts of the video data consumers actually care about. AFAIK it doesn't even matter what the images are because the emulated system didn't actually depend on the content of these images.
2) MAME's secondary goal is having some method of verifying 'software' which is really just 'data' which is either digital or analog. Digital is very easy to preserve compared to analog. In order to properly preserve analog data we need to use a transparent process that has a guarantee in processing. That is, if I were to use a capture card, I would have to be able to determine what it is doing to the signal, stochastically speaking. Capture cards demodulate the signal which is problematic for a number of reasons. Therefore there is no capture card that can adequately represent the data as it was represented from the LD player.*
IMO 1 is much easier than 2 and so we should be putting our efforts into that, but we must also not conflate 1 and 2. 2 is something that should at least be recognized as needing to be accomplished.
*Even if we do preserve the video signal, we need to note the model of the LD player. That is, the output video signal is dependent on the LD player. I still consider this an adequate preservation because (a) a reasonably limited number of models were used for these games and (b) these are devices that were coupled with the media during their time.