WEBVTT 00:00:00.000 --> 00:00:07.600 Okay, now, before we go any further, the first thing that I noticed when you kind of cracked this thing open and tilted it up like this 00:00:07.600 --> 00:00:15.920 was this enormous cast aluminum base, and I wanted to ask, just because, you know, I have an expert handy, 00:00:15.920 --> 00:00:22.400 is that so large and so heavy because of the stability that's required in operation? 00:00:22.400 --> 00:00:27.600 Absolutely. All these pieces have been put together with an incredible deal of precision 00:00:27.600 --> 00:00:32.080 if anything has moved out of place, even just a small amount, the tape won't feed correctly, 00:00:32.080 --> 00:00:35.920 or there'll be some sort of gap opening up in a mechanism, or it won't load in and out. 00:00:35.920 --> 00:00:46.080 Right, because the thing to appreciate about this, guys, is 1986. 1986, right? This was digital 00:00:46.640 --> 00:00:50.560 in 1986 at these, I mean, what, 40 gigs of tape? 00:00:50.560 --> 00:00:55.760 Absolutely, and that's for a half hour tape. That's unfathomable in 1986 terms. 00:00:55.760 --> 00:01:00.400 Yeah, this was basically the moon landing in tape form. You know, somebody said, 00:01:00.400 --> 00:01:06.640 what if we made a tape that stored all digital video when computers of the day couldn't store all digital video? 00:01:06.640 --> 00:01:11.760 The biggest hard drive in the world in 1986 was what, maybe 150 megabytes? 00:01:11.760 --> 00:01:16.000 In 1986, no, not even. 150 megabytes would have come later. 00:01:16.000 --> 00:01:18.960 I'm talking like, if you let government bring, you know? Sure, yeah. 00:01:19.360 --> 00:01:25.440 And so to them say, no, we want to store 40 or even 80 gigabytes on a tape. Everyone, 00:01:25.440 --> 00:01:31.760 what's a Gigabyte? This was like Doc Brown 1.21 gigawatts, like no one had ever heard of a Gigabyte 00:01:31.760 --> 00:01:39.440 of data, and they proposed to put 40 of them on a tape. Wow. Yeah, the head spins at 10,000 RPM. 00:01:40.160 --> 00:01:44.000 There's over a kilometer of tape in an hour-long cartridge. 00:01:44.960 --> 00:01:48.080 So dumb. I love it. 00:01:48.080 --> 00:01:51.920 But that's what you had to do, and I have nothing but respect for the guys who figured out how to do it. 00:01:52.960 --> 00:01:57.120 Okay, let's see if we can do a break job on their masterpiece. 00:01:57.120 --> 00:02:00.000 Yeah, we're going to take the Mona Lisa and just touch up the eyebrows. 00:02:00.560 --> 00:02:03.600 It's just, it seems so sketchy. 00:02:03.600 --> 00:02:11.520 I get that feeling now, and I've been working with this for two years. To stretch it out like that, and go, and we'll just do, kind of like that, you know, 00:02:11.520 --> 00:02:15.760 instead of a, you know, like flat seems safer. 00:02:15.760 --> 00:02:21.840 It does, but they tried that in the early days, and the machines were the size of an industrial washing 00:02:21.840 --> 00:02:28.000 and drying machine, and they ran for, you know, multiple days at a time before breaking down, 00:02:28.000 --> 00:02:31.920 as opposed to multiple months and years at a time, so I can see why they changed. 00:02:31.920 --> 00:02:36.480 My God, look at it. It's pixel perfect. This looks as good as it does in your memories. 00:02:37.200 --> 00:02:41.280 You know how you go back to playing old game, and you think, wow, I don't remember having this many, 00:02:41.280 --> 00:02:47.120 like pixels. This is as clean as my memories as a seven-year-old tell me reboot used to look. 00:02:47.760 --> 00:02:48.720 That's incredible. 00:02:50.880 --> 00:02:57.600 And yeah, let's do no more than like three minutes, because we just used up 0.01% of the life of that 00:02:57.600 --> 00:03:02.720 head way to go. Yeah, that was like, that was like 50 cents of head time right there. 00:03:04.160 --> 00:03:09.440 You got to get a shot though of the front panel when you're ejecting. So you want to see, you want to get it in B-roll, but you got to watch Linus on this. 00:03:10.400 --> 00:03:13.360 Oh, okay. So you just tell me when. 00:03:15.600 --> 00:03:19.440 Just as an example of how far the engineers of this machine went, 00:03:19.440 --> 00:03:22.880 to make this the finest machine possible, watch what happens when I eject it. 00:03:24.000 --> 00:03:29.920 Oh, the animation. That is so cool. It mirrors what it's doing. 00:03:29.920 --> 00:03:30.400 Sure does. 00:03:34.000 --> 00:03:40.880 That is so cool. It's heavy. I hadn't noticed before because I was busy making a video. 00:03:42.560 --> 00:03:46.880 Like the tape is heavy and it gets warm when it's in there. 00:03:46.880 --> 00:03:48.560 Like you pull it out and the tape is hot. 00:03:53.120 --> 00:03:57.200 Okay, we've gone on a long side quest here. 00:03:57.200 --> 00:04:00.320 Do we want to have a look at the break on that one? I'd like to look at the break on that one. 00:04:00.320 --> 00:04:09.280 All right, let's have a look. Okay, so cool. It's a memory card that you put in the slot to back up the settings of the deck. 00:04:10.960 --> 00:04:13.760 We found it in a box of cleaning supplies in mainframe's vault. 00:04:14.640 --> 00:04:18.000 You know, next to the wind decks. Memo. Basic setups. 00:04:19.200 --> 00:04:22.880 Before we do that, do you want to give us just a quick tour of some of the other 00:04:22.880 --> 00:04:26.240 key components of these systems? Absolutely. There's so many to see. 00:04:26.240 --> 00:04:31.040 First of all, we've got a power supply over here that puts out 5 volt at 50 amps. 00:04:31.040 --> 00:04:34.640 And it's made by the Brits, the guys who are famous for the quality of their electrics. 00:04:34.640 --> 00:04:40.640 So we have had to do a few repairs in there, including taking some silver polish to the connections and the relays 00:04:40.640 --> 00:04:45.840 to stop the machine from catching fire. If that's not enough, there's a second one of those standing up in the back 00:04:45.840 --> 00:04:52.000 and a third power distribution module over here. I don't even know how it works, but the fascinating one to me are the boards. 00:04:52.640 --> 00:04:58.560 This is a time before you could just go and buy off-the-shelf video conversion function chips. 00:04:58.560 --> 00:05:05.440 Everything in this machine had to really be designed from the ground up. And it shows when you start looking at the modules that are hiding in the bottom here. 00:05:05.440 --> 00:05:11.600 Because every one of these cards has custom BTS part number chips made at companies like 00:05:11.600 --> 00:05:20.320 Motorola just for this device. And then a sea of double ported SRAMs so that multiple components can talk to them simultaneously. 00:05:20.320 --> 00:05:23.760 I count five FPGAs. You like FPGAs, do you? 00:05:23.760 --> 00:05:26.720 I love FPGAs. Well, then this is your lucky day. 00:05:27.440 --> 00:05:32.160 I hope you're hungry. If this will open, which sometimes it does. 00:05:32.880 --> 00:05:35.600 Oh, baby. Look at those FPGAs. 00:05:36.160 --> 00:05:39.280 And this is 1993, this particular deck was built. 00:05:39.280 --> 00:05:46.880 FPGA was still like a new technology. This must have been like an entire week's output at the Xilinx factory back then. 00:05:47.440 --> 00:05:53.440 Then you add all the pals and gals had to be individually programmed, all the eProms, more of those crazy dual ported Rams. 00:05:53.440 --> 00:05:57.200 It's no wonder this thing cost the equivalent of a third of a million dollars. 00:05:57.200 --> 00:06:02.080 This machine was 10 times the price of a brand new cutting edge Digi Beta deck. 00:06:02.080 --> 00:06:05.440 And it shows. So what are we looking at here? 00:06:05.440 --> 00:06:09.600 Board one is processing control. Board two video record. 00:06:09.600 --> 00:06:16.240 The one you just had was video play. Then we've got digital audio and analog audio. 00:06:16.240 --> 00:06:20.960 Five boards each dedicated to just one function. 00:06:20.960 --> 00:06:26.320 And each two feet long and a foot wide, and covered in the most expensive ICs available at the time. 00:06:26.320 --> 00:06:31.680 So I haven't even shown you the cards in the back. Each one of these has its own special function on the tags. 00:06:31.680 --> 00:06:36.560 We've got, I guess, a servo pulse shaper. You need one of those. I'm an enthusiast for those. 00:06:36.560 --> 00:06:41.280 Absolutely. And this is the board in fact that I thought was giving us all those problems in those early days. 00:06:41.280 --> 00:06:45.440 Right. I spent hours going chip by chip on this, looking for a problem that didn't exist. 00:06:45.440 --> 00:06:49.280 Wow, what else we got in here? This is the motor control board 00:06:49.280 --> 00:06:52.800 that turned out to actually be responsible for our problems. 00:06:53.520 --> 00:06:57.760 Wasn't actually on the board either, but this is the thing that manages the clocks 00:06:57.760 --> 00:07:02.240 and also every single thing that turns or moves or spins or stops and starts. 00:07:03.120 --> 00:07:07.520 Wow. You'll notice, by the way, there's an entire Intel 8186 CPU on here. 00:07:07.520 --> 00:07:12.960 A lot of cards in here have these little board computers. And they're all running their own x86 operating system, 00:07:12.960 --> 00:07:16.640 independent and all coordinating with them over a separate Ethernet bus. 00:07:17.200 --> 00:07:21.440 Wow. Now, we don't need all this I.O., do we? 00:07:21.440 --> 00:07:26.480 Thankfully not. But these things were meant to go into a studio that probably already had a half a dozen existing formats. 00:07:26.480 --> 00:07:31.600 So these machines were really intended to just slot right in and connect into the existing workflow as much as possible. 00:07:31.600 --> 00:07:37.440 If you're wondering what the difference between a DCR 500 and a 300 is, it's just that the 300 has fewer of these inputs and outputs. 00:07:37.440 --> 00:07:42.000 Otherwise, all the mechanics are exactly the same. Okay, I think that transitions us perfectly 00:07:42.000 --> 00:07:46.080 to talking a little bit about the mechanical side. Excellent. There's a lot to talk about. 00:07:46.080 --> 00:07:52.240 As a not-VCR technician, walk me through it. 00:07:52.240 --> 00:07:58.080 Well, tape goes in the top here. Thank you. The slot is built for three different sizes of tapes. 00:07:58.080 --> 00:08:01.840 And so there's actually a motor that moves the hubs in and out depending on the size tape. 00:08:01.840 --> 00:08:07.200 Yep, you actually see a little diagram on the screen as it figures out which tape you've put in and adjusts itself to fit. 00:08:07.200 --> 00:08:12.160 And then down it goes. And roughly 10 million components start moving in here all in unison. 00:08:12.160 --> 00:08:15.760 There's belts and gears and flex shafts and all kinds of things 00:08:15.760 --> 00:08:20.880 that put the tape exactly where it needs to be. So it wraps around this big head in the center. 00:08:22.240 --> 00:08:25.920 Wraps around it. Yep, it goes from one corner all the way around to the other 00:08:25.920 --> 00:08:29.440 so that it has as much room to go over that tape as it can get. 00:08:30.320 --> 00:08:33.600 This over here is the motor that actually pulls the tape through the machine. 00:08:33.600 --> 00:08:38.480 It's called the cap stand roller. Yeah. And then all of the little guides and things in here, 00:08:38.480 --> 00:08:42.080 they're designed to make sure that the tape is angled at just the right points. 00:08:42.080 --> 00:08:46.800 These ones are spring-loaded. They're the ones that tell how tight the tape is as it goes through. 00:08:46.800 --> 00:08:49.920 It's making constant little adjustments to make sure it's pulling it just right. 00:08:50.560 --> 00:08:56.160 And all the way through there are sensors, little lights here, like the kind of thing that makes the ding-dong when you go through a store 00:08:56.160 --> 00:09:04.160 that just tells it that the tape is in the right position. Okay, and then, I mean, forgive my ignorance, 00:09:04.160 --> 00:09:08.640 where is the part on the head that actually does the zeros and ones? 00:09:08.640 --> 00:09:13.840 I saw the spinny part earlier. And the spinny part is it. It's hiding in this little slot down here. 00:09:13.840 --> 00:09:18.080 There's little teeny tiny bits, basically one half of a transformer, 00:09:18.080 --> 00:09:21.280 if you think about it, that when it goes whipping past the tape, 00:09:21.280 --> 00:09:24.800 it's inducing current in there that it can pick up and turn back into a signal. 00:09:24.800 --> 00:09:29.440 And it has to spin at 10,000 RPM. So this thing sounds like a dentist's drill whenever it's running. 00:09:30.240 --> 00:09:34.640 Right, and then this is all going out into these boards. 00:09:34.640 --> 00:09:41.360 Absolutely. These are shielded RF cables. They're carrying very faint signals to these boards that amplify them 00:09:41.360 --> 00:09:45.920 and basically square the signal back up into nice ones and zeros so it can be decoded. 00:09:47.360 --> 00:09:52.160 That's insane. Absolutely. Okay, you got your money's worth. I do want to see it in action then. 00:09:52.160 --> 00:09:56.160 Do you have like a sacrificial tape we can throw in to just see it ripping around in there? 00:09:56.160 --> 00:09:59.760 Oh, I can show you a tape. We got a tape or two around here. Let's have a look at a tape. 00:10:00.560 --> 00:10:04.000 You guys ready? I'm ready. Everyone crowds around. This is the best part. 00:10:13.440 --> 00:10:17.200 Holy crap. And if any one of those belts breaks, 00:10:17.200 --> 00:10:19.760 I have no idea how I'd ever put it back together. 00:10:20.640 --> 00:10:24.000 Is it just me or is the tape wrapped around it at an angle right now? 00:10:24.000 --> 00:10:28.880 It is. So pretty much all video units use something called helical scanning 00:10:29.600 --> 00:10:34.800 so that the frame is basically drawn diagonally across the tape. 00:10:35.440 --> 00:10:41.040 And that lets you pack more information into a tape that moves slower through the machine. 00:10:41.040 --> 00:10:44.880 As fast as this is going, this is still so much better than it would have been 00:10:44.880 --> 00:10:48.240 if they just laid the bits out sequentially like on an audio tape. 00:10:48.240 --> 00:10:51.120 The tape would have to move at like 12 miles an hour to get through the thing otherwise. 00:10:51.760 --> 00:10:55.200 1986. I know. 40 gigabytes. Yeah. 00:10:55.200 --> 00:10:59.200 I had no idea this existed. The Walkman was the hot new technology. 00:10:59.200 --> 00:11:01.920 Yeah. And then they're like, bam, all digital tape. 00:11:04.160 --> 00:11:08.240 Damn, Sony. Sony wasn't content with being one of the best. 00:11:08.240 --> 00:11:12.320 They wanted to show they were the best. And I think they nailed it with this format. 00:11:12.320 --> 00:11:15.120 Should we talk at all about where BTS comes from? Sure. 00:11:15.760 --> 00:11:22.240 BTS was a partnership between Bosch and Philips. So speaking of, Bosch is better known for making a lot of automotive gear, 00:11:22.240 --> 00:11:25.600 including the engine computers in anything German you've ever seen. 00:11:25.600 --> 00:11:28.880 The MW Porsche companies like that all use Bosch electronics. 00:11:28.880 --> 00:11:31.280 Right. Philips had a background NVIDIA gear. 00:11:32.000 --> 00:11:37.280 Bosch, I'm not quite sure if they did previous to this project or not. But either way, they came together to form BTS. 00:11:37.280 --> 00:11:40.880 And they made some really, really high end. Really good songs. Oh, absolutely. 00:11:42.160 --> 00:11:47.040 So they made some brilliant video equipment that no one in the consumer world would ever have seen. 00:11:47.040 --> 00:11:50.480 It was always stuff for studios. They made audio gear. They made video gear. 00:11:50.480 --> 00:11:55.040 They made cameras. They made editing decks. And they made some Digi Beta and things like that. 00:11:55.040 --> 00:11:58.880 But D1 was really them going out and saying, no, no, no. We want to be part of the top end. 00:11:59.440 --> 00:12:01.760 Unfortunately, not long after these machines were built, 00:12:02.480 --> 00:12:06.400 Philips bought Bosch's half. That turned into Philips Thompson, 00:12:06.400 --> 00:12:09.440 which then turned into Thompson Grass Valley. I think from that point on, 00:12:09.440 --> 00:12:13.440 this division got spun off as digital film technology who's been recently bought by the Prasad Group. 00:12:13.440 --> 00:12:18.240 And that's a long way of saying nobody in any of those companies has any clue that they ever built such a machine, 00:12:18.240 --> 00:12:20.240 let alone where to find any spare parts or manuals. 00:12:21.440 --> 00:12:24.880 So that's why I'm here. 00:12:27.520 --> 00:12:30.880 It sounds a lot like the airplane industry, actually. 00:12:30.880 --> 00:12:35.760 The difference is the airplane industry has a lot of regulation and documentation. 00:12:35.760 --> 00:12:39.600 And if BTS had been forced to conform to some of that, my job would have been a lot easier. 00:12:39.600 --> 00:12:43.280 Well, I mean, on the spare parts side. So acquisitions get made. 00:12:43.280 --> 00:12:50.800 And then you end up with these kind of orphan aircraft that nobody really owns the responsibility. 00:12:51.360 --> 00:12:54.400 For producing any parts for them. Yes, that definitely sounds familiar. 00:12:55.120 --> 00:12:59.120 These machines, as far as I can tell, back in the day, there would have been guys 00:12:59.120 --> 00:13:04.240 practically in lab coats coming out to service them. The parts were not something that you sort of ordered and kept on sight. 00:13:04.240 --> 00:13:07.440 Radio shack, probably. No, not so much, no. 00:13:07.440 --> 00:13:13.520 So this is the kind of thing where they brought the part that you needed and they put it in for you and then they left and took their tools. 00:13:13.520 --> 00:13:19.680 Yep, and the manuals. And so there wasn't just a lot of surplus parts alongside the machines. 00:13:21.440 --> 00:13:25.840 And now if memory serves, I think I only need to remove a couple of these 00:13:25.840 --> 00:13:29.200 and the whole mechanism lifts out. Fortunately, there's a tool for that. 00:13:29.840 --> 00:13:32.400 Of course there is. Pardon me, gotta find it. 00:13:33.040 --> 00:13:37.360 That was the other tool I was supposed to have pulled out when I said I was going to make this look professional and coordinated. 00:13:37.360 --> 00:13:40.240 I feel like I'm helping dad to stay out of my life. 00:13:41.920 --> 00:13:47.200 There you go. Okay, so we use a pair of these fancy pliers. 00:13:47.520 --> 00:13:53.840 And we can take this little ring here and convince it to come off. 00:13:55.520 --> 00:13:58.560 If we're lucky, it doesn't go winging through the entire machine. 00:14:01.280 --> 00:14:04.480 Don't use that take. Oh, this is the floating one. 00:14:04.480 --> 00:14:07.760 I don't need to remove it. This is the one that needs to come off. 00:14:07.760 --> 00:14:11.600 That's right. That's how I did last time. Sorry, it's been a year since I did this. 00:14:11.600 --> 00:14:14.320 Okay, ignore that part where I dinged the machine. 00:14:15.280 --> 00:14:19.280 It's fine, it's not expensive. No, no, it wasn't my money, remember? 00:14:25.120 --> 00:14:26.000 Thanks, Kyokes, yeah. 00:14:29.360 --> 00:14:34.320 Okay, so there's a little ring down here and all you have to do is convince it to come off. 00:14:35.280 --> 00:14:39.040 And then someone has to catch it before it pings off throughout the store 00:14:39.040 --> 00:14:42.640 because we'll never find it again. Brilliant. So you're the catcher today. 00:14:42.640 --> 00:14:47.920 I'll put that there. That's my porn roll too. 00:14:47.920 --> 00:14:48.400 Oh no. 00:14:53.760 --> 00:14:56.480 I've seen the people who comment on your videos. Let's not encourage them. 00:14:58.160 --> 00:14:59.680 That's my job to encourage them. 00:15:02.880 --> 00:15:09.200 All right, so the problem with this mechanism is actually three-fold. 00:15:09.200 --> 00:15:13.600 There's three different things that go on with them. This is the break in the on position right now. 00:15:13.920 --> 00:15:18.880 The position it goes into when it loses power. And it should stop this from turning very easily. 00:15:18.880 --> 00:15:22.000 And it's still kind of... Yeah, there's a little bit of pressure on it. 00:15:22.000 --> 00:15:27.520 Yeah, but not as much as there should be. And part of the problem is just that the spring itself is worn out. 00:15:27.520 --> 00:15:30.800 It's gotten old. It's been wrapped around there so long it's lost some of its springiness. 00:15:30.800 --> 00:15:34.080 Well, how do you fix that? You take the spring off and tighten it. 00:15:34.080 --> 00:15:39.520 Right, right, right. But another problem is you see how these two bits are kind of rubbing against each other. 00:15:39.520 --> 00:15:46.160 There's this piece here and then there's this spring underneath. Sometimes they rub and get caught and it doesn't go all the way back. 00:15:46.880 --> 00:15:50.880 So we're going to bend this piece up so there's no way for it to touch it. 00:15:51.520 --> 00:15:55.600 And then lastly, we also want to make sure that the brakes turn off when they're not needed. 00:15:56.240 --> 00:16:01.840 This little rod is what's supposed to release the brakes. And you see how it only really touches it at the very end of its travel? 00:16:01.840 --> 00:16:05.520 Yeah. They put some adjustment screws here so you could move the whole thing back, 00:16:05.520 --> 00:16:10.960 but it's already about as far as it can go. So instead, once again, we're going to bend this very slightly inward 00:16:10.960 --> 00:16:14.800 so that it grabs it a little sooner and pulls it off a little farther. 00:16:14.800 --> 00:16:19.520 And that little bit of an adjustment is just what it needs to go both on all the way and off all the way. 00:16:19.520 --> 00:16:23.200 We hope. It's worked so far. I think we can do it again. 00:16:23.200 --> 00:16:23.520 Okay. 00:16:32.320 --> 00:16:36.720 This is hilarious. If you've been wondering why the project has taken so long. 00:16:37.760 --> 00:16:40.640 I haven't really. I just assumed that it was difficult. 00:16:40.960 --> 00:16:44.080 You were right. Your assumptions were correct. 00:16:44.080 --> 00:16:46.800 Yeah. I don't know what that was. My hunch says it was grease. 00:16:50.000 --> 00:16:56.400 Must have been. I mean, could it even be just like remnants of dust crap kicking around in here? 00:16:56.400 --> 00:17:01.200 Like it could almost be just about anything. It has a very greasy look though. 00:17:02.160 --> 00:17:05.360 I'm guessing probably grease given that it had something covering it a whole life. 00:17:05.360 --> 00:17:10.000 But was it lubricating grease or was it just don't rust grease? 00:17:11.200 --> 00:17:15.120 Well, the good news is neither of these parts would rust. So. Oh, what is this? 00:17:15.760 --> 00:17:19.120 Actually, you know what? Now that you mentioned it, that might be spring steel given that it's a spring. 00:17:19.120 --> 00:17:22.320 So the man has a point. 00:17:24.800 --> 00:17:30.400 Multiple. So good news is I have a friend in the aircraft industry 00:17:30.400 --> 00:17:37.360 and I asked him what the best greases they had were and he gave me a little squirt of something that he's used before. 00:17:37.360 --> 00:17:44.560 I have no idea what this magic substance is. But I'm told that it's a really, really good grease. 00:17:44.560 --> 00:17:48.400 Monoject 412. Yeah, but is that the grease or is that the tube? 00:17:48.400 --> 00:17:50.800 Who knows? I'm certainly not one to ask questions. 00:17:52.160 --> 00:17:55.040 That is one of the thinnest washers I think I've ever seen in my life. 00:17:57.760 --> 00:18:01.280 It looks like a picture of a washer. Right, doesn't it? Like it was simply drawn on. 00:18:01.840 --> 00:18:08.240 Yeah, you can tell you're not meant to use a lot of this stuff by how it barely... 00:18:09.040 --> 00:18:16.640 Yeah. Comes out. So you'll see here that the two pieces pull it farther apart. 00:18:16.640 --> 00:18:22.480 They don't cross over. They sit like that. But now the mechanism wants to sit almost overlapping 00:18:22.480 --> 00:18:25.840 where before it had a gap between it. That's good. That means we've bent it the right way. 00:18:26.480 --> 00:18:32.240 But we also need to bend this ARM up and out so that it doesn't have... 00:18:32.240 --> 00:18:38.160 Or sorry, I guess that'd be in. So the angle of it a little bit. Just a little so that it doesn't rub against the mechanism. 00:18:38.320 --> 00:18:41.200 And so when this rod moves back, it pulls it just a little bit farther. 00:18:41.920 --> 00:18:47.840 So all we have to do is give this beautiful precision German-made spring 00:18:47.840 --> 00:18:52.880 a very ugly bodge with a pair of pliers. I love the generous use of the word we. 00:18:52.880 --> 00:18:55.360 Yeah, you are an integral part of this operation. 00:18:56.160 --> 00:19:00.000 You are there, so if I screw this up, I can say... But Linus told me to. 00:19:01.920 --> 00:19:06.960 So we just go like that. Man, this thing, it almost like looks Rube Goldberg-like. 00:19:06.960 --> 00:19:10.000 Oh, absolutely. And it's just dependencies, you know? 00:19:11.440 --> 00:19:16.880 Well, just as a brief example, this is a weight that stops one of the rollers here 00:19:16.880 --> 00:19:19.920 from vibrating around, and it locks in place. 00:19:20.560 --> 00:19:27.280 It gets screwed in by a flexible metal shaft, which is driven by a belt, 00:19:27.280 --> 00:19:33.760 like a timing belt out of a car. Now it's connected to this rod at an angle, so that belt is twisted. 00:19:33.760 --> 00:19:39.680 You don't want to run a belt when it's twisted, but it doesn't matter, because before that starts turning, this piece lifts up. 00:19:40.240 --> 00:19:45.040 And that's all triggered by a marble rolling down a chute onto a mouse trap, 00:19:45.040 --> 00:19:48.560 which launches the fireworks, which makes the monkey run to the cheese. 00:19:48.560 --> 00:19:51.760 Which wakes up your sister. Exactly. And then, you know, don't wake daddy. 00:19:54.560 --> 00:19:58.160 That's a 90s kid reference, if I've ever heard one. I'm a 90s kid. 00:19:58.160 --> 00:20:01.120 What do you expect? It'd be weird if I was making 70s references. 00:20:02.160 --> 00:20:08.480 Where's the beef? And the cool thing is, we can actually run the machine in this position, 00:20:08.480 --> 00:20:12.800 so we can watch these and see exactly what they're doing, and what kind of adjustments are needed. 00:20:12.800 --> 00:20:17.280 I think it's time to give this a go. 00:20:17.280 --> 00:20:21.440 Once it fires up, don't put your hands in it. You bet. That would be a very bad idea. 00:20:22.400 --> 00:20:25.760 And so I want you to go ahead and flip on the power supply at the very back. 00:20:26.640 --> 00:20:29.280 It's on the bottom there, the one that we plugged into. 00:20:30.880 --> 00:20:33.360 Oh yeah, you bet. Yeah. Okay, you're sure? Uh-huh. 00:20:34.480 --> 00:20:35.040 Okay, it's on. 00:20:37.680 --> 00:20:41.360 Okay. And you can see we're already getting diagnostic information coming up. Oh yeah, look at that. 00:20:41.360 --> 00:20:44.160 Yep. Let me get the everything set up here. 00:20:45.760 --> 00:20:48.320 We've got a little work to do before we can capture. 00:20:48.960 --> 00:20:52.080 We've got our video capture tool, and we have our very custom made 00:20:52.800 --> 00:20:59.520 error logging tool that Brian here developed. So these machines have like a hash function inside. 00:20:59.520 --> 00:21:03.680 They can tell when a bit has been read correctly or not, 00:21:03.680 --> 00:21:08.480 and it's got some rudimentary error correction to try and patch those bad bits. 00:21:08.480 --> 00:21:11.680 Got it. Basically, it like grabs from the previous frame and fills in the gaps, 00:21:11.680 --> 00:21:18.320 and hopes that things kind of haven't moved too much. We could smart. Yeah, but we found a diagnostic interface inside the machine 00:21:18.320 --> 00:21:24.720 that lets us tap into that information in real time. So we can get a log of when uncorrectable errors occurred, 00:21:24.720 --> 00:21:28.480 time coded to match the video. So we can see, oh, there was a big spike right here. 00:21:28.480 --> 00:21:33.200 Let's go to that section of video and really scrutinize it, and see if we can see something that's wrong, 00:21:33.200 --> 00:21:37.120 that maybe we need to take that same piece of footage and replace it with one from an alternate copy. 00:21:37.680 --> 00:21:42.560 Got it. Right, and because you guys have in many cases multiple edits, 00:21:42.560 --> 00:21:47.920 like one that had less censoring for YTV, one that had more censoring for the American network, 00:21:48.640 --> 00:21:51.920 you can kind of... Absolutely, absolutely. I'm still in the gap. 00:21:51.920 --> 00:21:56.720 So, so cool. Yeah, so Jacob here has been scrutinizing every episode, 00:21:56.720 --> 00:21:59.840 just absolutely obsessively looking for those, and keeps finding... 00:21:59.840 --> 00:22:05.360 I've spent the last month watching basically every episode in real time looking for pixels that are off. 00:22:05.360 --> 00:22:09.120 Does it take any of the magic out of it, or does it put more magic into it? 00:22:09.120 --> 00:22:12.320 I don't know, maybe I'm crazy, but it's... He's 1200 kilometers away from me, 00:22:12.320 --> 00:22:15.920 so I can't strangle him. So instead, I just curse away and do it again. 00:22:15.920 --> 00:22:19.040 But we've got tons of basically Excel spreadsheets. 00:22:19.040 --> 00:22:24.160 Every tape has a custom fix list. And so every time we render it, if there's an issue, 00:22:24.160 --> 00:22:29.120 we're going to go back and readjust that command. And I think this has to be put out there. 00:22:29.120 --> 00:22:32.480 There is no AI involved, no LLMs involved. 00:22:32.480 --> 00:22:35.680 This is old school, if we want to talk about AVI, Synth, 00:22:35.680 --> 00:22:42.560 and all the different plugins out there. This is truly algorithmic, and nothing for interpretive logic. 00:22:43.120 --> 00:22:47.920 Because everyone says you can AI upscale this show, and we will eventually show you why you cannot. 00:22:49.440 --> 00:22:53.520 Well, that's the other scary thing, is that when this thing was young and full of life, 00:22:54.400 --> 00:22:59.520 by default, it would move the tape through the machine at a 40 times fast forward and rewind. 00:23:00.480 --> 00:23:05.040 40? Yeah, and you can hear and see that this thing ain't no slouch on a normal day, 00:23:05.040 --> 00:23:09.440 but 40 was what it would do by default. So I have to just manually do it. 00:23:09.440 --> 00:23:12.640 I only go to six. Any more than that, and I feel like 00:23:12.640 --> 00:23:16.080 we're just asking for the tape to explode. Yeah, this was undocumented, 00:23:16.080 --> 00:23:21.360 but there is an RS422 control port that somehow these guys were so bleeding edge, 00:23:21.360 --> 00:23:25.200 they're using the same control protocols as what's being used in digital capture today. 00:23:26.000 --> 00:23:29.440 So we, just out of luck, our Blackmagic software, 00:23:29.440 --> 00:23:34.080 we were able to connect right into that, and it's auto-capture everything from the auto rewind 00:23:34.080 --> 00:23:38.800 to the play, even the time codes. We're actually getting the time codes synced up 00:23:38.800 --> 00:23:43.520 so that every time we receive an episode on my side, we're at 59 minutes on the mark. 00:23:43.520 --> 00:23:45.440 The episode starts at the one hour mark, 00:23:47.440 --> 00:23:51.360 one minute or so for the intro, and so we have consistency tape over tape. 00:23:51.360 --> 00:23:55.360 Previously, we'd have to figure out the offset of every single recording and figure out 00:23:55.360 --> 00:23:59.440 if we're grabbing A from B, we have to figure out what that offset was. 00:23:59.440 --> 00:24:04.160 So this has been tons of development. So this time code's actually embedded in that. 00:24:04.160 --> 00:24:08.640 That's the raw serial data, and then just convert to hex. And even that's not the end-all, be-all. 00:24:08.640 --> 00:24:12.160 We had one tape where it fixes itself, 00:24:12.160 --> 00:24:15.200 but not at regular playback rate. So poor Mark here. 00:24:15.200 --> 00:24:19.520 I had him here one night playing back at 0.2 times, 00:24:19.520 --> 00:24:24.160 and then I have a capture file that's two hours long and manually cut basically every fifth frame 00:24:25.360 --> 00:24:29.120 until we had good clean frames for it. Because some of these are just one tape. 00:24:29.120 --> 00:24:30.160 That's all you got. 00:24:49.840 --> 00:24:51.840 Hey, that's you. 00:25:13.840 --> 00:25:18.560 This is all digital, right? If you're familiar with the broadcast technologies, 00:25:18.560 --> 00:25:21.600 this has the earliest version of SDI that we are aware of. 00:25:21.600 --> 00:25:26.240 So the SDI digital video goes digitally to our capture card one-to-one. 00:25:26.240 --> 00:25:30.800 The audio also is digital audio in the same format one-to-one on our system. 00:25:30.800 --> 00:25:34.880 That's basically CD audio. And because it's the exact same bit rate, 00:25:34.880 --> 00:25:41.040 no noise, same levels, we can use it to cancel out and hear theoretical, 00:25:41.040 --> 00:25:45.760 basically studio recordings of certain areas. It's just a matter of slightly adjusting the time 00:25:45.760 --> 00:25:50.560 throughout the tape, but we have done some tests that are quite eerie. It sounds like you're sitting in a studio recording. 00:25:51.120 --> 00:25:56.640 Oh, that's cool. So yeah, at the end here, we're seeing this titleist credits where it is the intro. 00:25:56.640 --> 00:26:03.120 Sometimes it does have music, but most of the time it's silent, but there's none of the produced by, written by here. 00:26:03.120 --> 00:26:07.520 And so when it would be shipped safe Quebec for the French dub, 00:26:07.520 --> 00:26:11.520 they would put their own overlay in there, so it would be localized. Some of these tapes actually cut off slightly early, 00:26:11.520 --> 00:26:16.000 like there's just no anything on them. So I've had to truncate this one by about a minute. 00:26:16.080 --> 00:26:21.360 Hopefully I won't cut anything off. So if we just let it run for another four minutes, 00:26:21.360 --> 00:26:27.040 it will, we might get a version of the intro with no audio and no text on it for, again, for localization, 00:26:27.040 --> 00:26:31.520 and then it'll stop itself. Oh, interesting. Okay. Oh yeah, there was some Doctor Who episodes 00:26:31.520 --> 00:26:34.240 that were lost or something, wasn't there? Yeah. There's just several. 00:26:35.360 --> 00:26:38.800 But what they're finding is, there's these BBC retransmitters in, you know, 00:26:38.800 --> 00:26:42.640 Africa that someone's about to sell the land, they go inside and they're finding tapes 00:26:42.640 --> 00:26:47.360 that are in far worse shape than this and have a lot more money to spend to try and recover them 00:26:47.360 --> 00:26:53.040 because they are considered lost media. And that's, Reboot's just one example of lost media right now 00:26:53.040 --> 00:26:56.240 that we are trying to help, especially on the Canadian front, to restore. 00:26:56.240 --> 00:26:59.680 And there's several other mainframe shows that have been discussed, 00:26:59.680 --> 00:27:03.440 and frankly, licensing is beyond a nightmare. 00:27:03.440 --> 00:27:06.800 So mainframe owns the Reboot property. 00:27:06.800 --> 00:27:10.320 We have a direct contact there. So that's relatively simple. 00:27:10.320 --> 00:27:14.720 But let's say, for example, a toy company were to own the intellectual property. 00:27:15.360 --> 00:27:18.640 Yeah. Then it gets more complicated. We're hoping, you know, for... 00:27:18.640 --> 00:27:23.120 Unless the aforementioned toy company was going through some bad PR right now 00:27:23.120 --> 00:27:26.160 and they were looking for a win. Can I Hasbro?