WEBVTT - My Test Image is the Centerfold 0:00:04.440 --> 0:00:12.319 Welcome to tech Stuff, a production from iHeartRadio. Hey there, 0:00:12.320 --> 0:00:15.720 and welcome to tech Stuff. I'm your host, Jonathan Strickland. 0:00:15.760 --> 0:00:19.439 I'm an executive producer with iHeart Podcasts and how the 0:00:19.480 --> 0:00:23.240 tech are you today? I want to tell the story 0:00:23.440 --> 0:00:27.880 of Lenna orsin her role in the history of image 0:00:27.880 --> 0:00:33.559 processing and technical papers, and why some publications and organizations 0:00:33.560 --> 0:00:39.199 are now banning papers that contain her photograph. Because I 0:00:39.240 --> 0:00:41.880 hadn't actually heard of any of this before I read 0:00:41.920 --> 0:00:46.560 an article in Ours Technica by BINGJ. Edwards titled Playboy 0:00:46.680 --> 0:00:50.000 image from nineteen seventy two gets banned from I Tripoli 0:00:50.280 --> 0:00:53.800 computer Journals. Yeah, we're going to be telling you about 0:00:53.840 --> 0:00:57.880 Playboy and nude modeling as well, because that all is 0:00:57.920 --> 0:01:01.480 involved in this story, all right. The first up, who 0:01:01.840 --> 0:01:05.880 is Lena for Senne? Well, she's a model from Sweden. 0:01:06.240 --> 0:01:09.800 She moved to America and in her early twenties she 0:01:10.000 --> 0:01:13.440 was a model and Playboy discovered her and she became 0:01:13.640 --> 0:01:18.520 Miss November nineteen seventy two. The centerfold photo of her 0:01:18.560 --> 0:01:24.200 pictorial would become entwined with tech history, and not just 0:01:24.240 --> 0:01:28.280 because some tech heads were subscribed to Playboy, although I 0:01:28.319 --> 0:01:31.200 suspected that played a large part in it, and her 0:01:31.240 --> 0:01:34.000 image is one that you've probably seen at least the 0:01:34.240 --> 0:01:38.080 portion of the image of her centerfold that has been 0:01:38.200 --> 0:01:41.399 used in the tech sector, because it's literally been used 0:01:41.440 --> 0:01:44.000 thousands of times. I know that when I saw the picture, 0:01:44.040 --> 0:01:47.600 I thought, oh, that's where that's from, because I had 0:01:47.600 --> 0:01:50.800 never thought to look any further into it. So a 0:01:50.840 --> 0:01:54.280 photographer named Dwight Hooker took the photo in question, and 0:01:54.360 --> 0:01:58.080 in the photo, Lena Foceen stands facing a mirror and 0:01:58.400 --> 0:02:01.200 her back is to the camerage. She's turned her head 0:02:01.560 --> 0:02:04.320 so that she's looking back over her shoulder to look 0:02:04.640 --> 0:02:09.000 at the camera lens. And the centerfold version of this 0:02:09.080 --> 0:02:12.600 photo is full body, you know, from head to toe, 0:02:12.680 --> 0:02:16.240 well technically hat to toe, because she is wearing a 0:02:16.280 --> 0:02:20.200 hat and a boa and really nothing else. The section 0:02:20.480 --> 0:02:23.040 of the image that matters the most to our story 0:02:23.480 --> 0:02:26.920 is from her shoulders up shoulders to the top of 0:02:26.960 --> 0:02:29.799 the feathered hat she is wearing, because that's the part 0:02:29.840 --> 0:02:31.560 of the image that we play a huge role in 0:02:31.600 --> 0:02:35.680 the development of image processing in general, and the fact 0:02:35.720 --> 0:02:38.280 that it was a cropped image from a centerfold photo 0:02:38.320 --> 0:02:41.600 in Playboy would become a source of debate and I 0:02:41.680 --> 0:02:44.680 hesitate to use the word controversy. I'm not sure that 0:02:44.760 --> 0:02:48.919 it was that controversial as much as it was concerning. 0:02:49.639 --> 0:02:51.880 But we'll talk all about that when we get toward 0:02:51.919 --> 0:02:55.000 the end of this episode. But let's go back to 0:02:55.080 --> 0:02:59.240 the early nineteen seventies, more specifically the summer of nineteen 0:02:59.400 --> 0:03:02.560 seventy three. Getting more specific than just the summer of 0:03:02.560 --> 0:03:04.760 seventy three is a bit tricky because the folks who 0:03:04.760 --> 0:03:08.880 have related the story as it unfolded about using this 0:03:08.960 --> 0:03:11.200 image in the first place, we're all working from memory, 0:03:11.520 --> 0:03:14.560 so they were just like, it's June or July nineteen 0:03:14.600 --> 0:03:16.760 seventy three. That's about as specific as we can get. 0:03:17.200 --> 0:03:20.440 The place, however, we can narrow down. The place was 0:03:20.480 --> 0:03:24.440 the University of Southern California, and specifically the Signal and 0:03:24.680 --> 0:03:29.680 Image Processing Institute or SIPI or SIPPY as I will 0:03:29.720 --> 0:03:34.120 call it. The institute at that point was pretty darn young, 0:03:34.400 --> 0:03:38.320 so an alumnus of USC named William Pratt led a 0:03:38.480 --> 0:03:43.920 group of alumni to establish this institute in nineteen seventy two, 0:03:44.680 --> 0:03:47.560 and they had the goal of tackling three big challenges 0:03:47.840 --> 0:03:51.640 in computing and digital images. So, according to the organization's 0:03:51.680 --> 0:03:55.640 fiftieth anniversary celebration page because they just recently celebrated that 0:03:55.680 --> 0:03:58.880 a couple of years ago. The challenges were to solve quote, 0:03:59.200 --> 0:04:06.400 image code, image restoration, and image data extractioned. So remember 0:04:06.760 --> 0:04:11.800 this is decades before we would get industry standards like JPEG, 0:04:12.160 --> 0:04:15.320 and even still more than a decade before the bitmap 0:04:15.360 --> 0:04:19.359 image file format. So the folks at USC SIPPY, which 0:04:19.440 --> 0:04:22.880 back then was just called IPPI, they were doing work 0:04:22.960 --> 0:04:26.359 that would inform these later standards. So this is like 0:04:26.400 --> 0:04:29.279 the early work where they're coming up with the methodologies 0:04:29.520 --> 0:04:33.760 that ultimately would find their way into various file formats 0:04:34.320 --> 0:04:36.560 much further down the line. So they were laying the 0:04:36.640 --> 0:04:40.640 groundwork now. In that summer of nineteen seventy three, an 0:04:40.640 --> 0:04:45.760 electrical engineering assistant professor named Alexander Sawchuk was working with 0:04:45.839 --> 0:04:48.640 grad students to find an image that they were going 0:04:48.680 --> 0:04:52.920 to scan. So Sawchuk had a colleague who was preparing 0:04:53.920 --> 0:04:58.720 a paper for a conference, and this paper was all 0:04:58.760 --> 0:05:02.440 about the process they were using at USC to scan 0:05:02.600 --> 0:05:06.560 images and to digitize them. And Sawchuk wanted to grab 0:05:06.600 --> 0:05:09.000 something that they hadn't already used a dozen times in 0:05:09.120 --> 0:05:12.000 various test scans, like they had some stock photos that 0:05:12.040 --> 0:05:15.480 were their go to, but everyone had seen those already, 0:05:15.520 --> 0:05:18.520 and moreover, the folks in the lab were just sick 0:05:18.560 --> 0:05:21.440 of them, so they wanted to get something new. Plus, 0:05:21.440 --> 0:05:24.560 they wanted something that would really show off their methodology 0:05:24.680 --> 0:05:27.440 to good effect. So they wanted something that was special. 0:05:27.440 --> 0:05:30.680 They wanted a glossy photo that had a lot of 0:05:30.760 --> 0:05:33.400 intricate detail inside of it and also a high level 0:05:33.400 --> 0:05:36.240 of contrast in it. They wanted something that had a 0:05:36.279 --> 0:05:40.960 lot of dynamic elements so that their methodology could be 0:05:41.000 --> 0:05:44.440 shown off in the best light, so to speak. So 0:05:44.680 --> 0:05:47.400 he also wanted the photo to be of a person's face, 0:05:47.480 --> 0:05:50.320 to really, you know, have something that people could see 0:05:50.400 --> 0:05:53.200 and recognize immediately as oh, that's a human being, and 0:05:53.240 --> 0:05:56.839 thus be able to tell how well the process worked. 0:05:57.160 --> 0:06:00.320 Now history has lost the name of the heat Row 0:06:00.360 --> 0:06:04.160 who walked by carrying an issue of the November nineteen 0:06:04.240 --> 0:06:07.719 seventy two Playboy magazine. Let me just say, the nineteen 0:06:07.760 --> 0:06:10.800 seventies were definitely a different time. But even as someone 0:06:10.880 --> 0:06:13.680 who was born in the nineteen seventies, as an I 0:06:13.800 --> 0:06:16.120 was born in the seventies, y'all, it is hard for 0:06:16.160 --> 0:06:19.840 me to imagine just casually bringing a Playboy magazine into 0:06:19.880 --> 0:06:22.480 an academic lab like that, or just carrying it around 0:06:22.480 --> 0:06:25.719 at a college campus. It's hard for me to imagine 0:06:25.720 --> 0:06:29.200 doing that. But then I'm also not an electrical engineer, 0:06:29.640 --> 0:06:33.120 so maybe I just I'm just built different. The scanner 0:06:33.360 --> 0:06:36.680 that the team was using was a mirror head or 0:06:36.800 --> 0:06:42.800 mir head muirhad, I'm butchering the pronunciation, but he was 0:06:42.880 --> 0:06:46.960 a wire photo machine and had a scanning resolution of 0:06:47.279 --> 0:06:50.920 one hundred lines per inch. Their plan was to scan 0:06:51.040 --> 0:06:55.000 a five twelve x five twelve image, So five hundred 0:06:55.000 --> 0:06:56.800 and twelve lines by five hundred and twelve lines I 0:06:56.839 --> 0:06:58.960 meant the photo that they scanned could only be five 0:06:59.040 --> 0:07:02.119 point one two inches to a side right one hundred 0:07:02.200 --> 0:07:04.880 lines to an inch. So that would mean that they 0:07:05.000 --> 0:07:08.120 could not do a full scan of the centerfold, and 0:07:08.440 --> 0:07:11.320 they shouldn't even have thought of it anyway, and I'm 0:07:11.320 --> 0:07:14.840 sure they didn't because it would have been incredibly inappropriate 0:07:15.000 --> 0:07:18.560 to present the centerfold as a scanned image in a 0:07:18.680 --> 0:07:22.720 conference paper, considering the nudity. That would just be inappropriate. 0:07:22.800 --> 0:07:26.240 So they just went the top five point one two 0:07:26.320 --> 0:07:29.760 inches of the image showing, which would crop the photo 0:07:30.160 --> 0:07:35.000 at Lenna's shoulders, so it would just be the shoulders up. Well, 0:07:35.040 --> 0:07:38.360 let's talk about the actual scanning technology for a minute. 0:07:38.480 --> 0:07:42.000 So we're in the early nineteen seventies, and you might 0:07:42.080 --> 0:07:45.960 wonder how old was wire photo technology. Believe it or not, 0:07:46.480 --> 0:07:49.880 wire photo services had been in operation for around fifty 0:07:50.320 --> 0:07:54.120 years by the time we're talking about making computer scans. Now, 0:07:54.120 --> 0:07:57.280 of course, in the early twentieth century, people were not 0:07:57.440 --> 0:08:01.920 scanning photographs into computers. That wasn't happening. But what they 0:08:01.960 --> 0:08:05.800 were doing was they were using technology to turn photographs 0:08:05.840 --> 0:08:09.920 into electrical signals that could then be transmitted over wire 0:08:10.200 --> 0:08:13.520 or over radio waves. Now, there were different particulars for 0:08:13.600 --> 0:08:17.560 the various methodologies. There wasn't just one device out there 0:08:17.600 --> 0:08:21.760 that did this. But generally speaking, a typical approach with 0:08:22.080 --> 0:08:26.360 a wire photo scanner would have a machine that consisted 0:08:26.360 --> 0:08:29.640 of a drum, and on this drum you would place 0:08:29.680 --> 0:08:32.560 the image because kind of similar to like a photocopier 0:08:32.640 --> 0:08:35.840 in that respect, and the drum would rotate and as 0:08:35.880 --> 0:08:38.480 the drum rotated, there would be this tiny dot of 0:08:38.559 --> 0:08:41.600 light that would hit the photograph and slowly make its 0:08:41.600 --> 0:08:45.720 way across the width of the photograph. So as the 0:08:45.800 --> 0:08:48.719 light was moving, the drum would rotate fast enough so 0:08:48.760 --> 0:08:52.760 that the light would scan the entire length of the photograph, 0:08:53.240 --> 0:08:56.480 and the photograph would reflect some of that light onto 0:08:56.600 --> 0:08:59.599 a photovoltaic cell, so kind of similar to what you 0:08:59.600 --> 0:09:02.840 would find in the solar cell right now. The intensity 0:09:03.080 --> 0:09:06.600 of the light that hit that photovoltaic cell would determine 0:09:06.640 --> 0:09:09.360 the amount of electrical charge that the cell could generate. 0:09:09.559 --> 0:09:11.480 So that means you would end up with this variable 0:09:11.600 --> 0:09:15.080 electrical signal, and that signal would represent the amount of 0:09:15.160 --> 0:09:18.320 light that was reflected off the photograph. You can think 0:09:18.360 --> 0:09:21.200 of it as a really strong signal shows a bright 0:09:21.320 --> 0:09:23.760 part of the image, and a weak signal would show 0:09:23.800 --> 0:09:28.160 a dark part of the image. To oversimplify it. Now, 0:09:28.679 --> 0:09:32.679 you could then take this electrical signal and you could 0:09:32.679 --> 0:09:35.600 send it to a destination. You could do so directly 0:09:35.679 --> 0:09:39.400 over a wire, it's like a power line or phone 0:09:39.440 --> 0:09:43.640 line or something. Or you could further transform the information 0:09:43.720 --> 0:09:47.480 by converting the signal into radio waves and then broadcast 0:09:47.800 --> 0:09:51.040 the radio waves to a receiver which would then reverse 0:09:51.080 --> 0:09:54.360 the process to capture the radio waves, convert them back 0:09:54.360 --> 0:09:56.800 into an electrical signal, and then send them on. So 0:09:57.120 --> 0:09:59.480 either way this electrical signal would make its way to 0:09:59.520 --> 0:10:02.320 the other end, where you would have a similar device 0:10:02.400 --> 0:10:05.079 to the scanner. It would also have a drum on it, 0:10:05.280 --> 0:10:07.080 but instead of a photo on it, you would have 0:10:07.320 --> 0:10:12.440 photoreactive paper or film on it, and the drum would 0:10:12.559 --> 0:10:17.199 rotate in synchronization with the rotation of the original drum, 0:10:17.360 --> 0:10:20.480 synchronization of what it was going at when it was scanning, 0:10:20.480 --> 0:10:24.439 that is, and it would also project a light, and 0:10:24.440 --> 0:10:26.800 that light the projection of that light would depend upon 0:10:26.840 --> 0:10:30.480 that varying electrical signal, So in strong parts of the signal, 0:10:31.000 --> 0:10:34.560 the light would be more intense, and in weaker parts 0:10:34.600 --> 0:10:37.120 of the signal it'd be dimmer. And so this light 0:10:37.160 --> 0:10:40.880 would hit the photoreactive film on the drum. As the 0:10:40.960 --> 0:10:45.000 drum rotated, the light would also scan across the width 0:10:45.080 --> 0:10:49.560 of the film. And then you would take the film 0:10:49.720 --> 0:10:52.000 and you would develop it, and you would end up 0:10:52.120 --> 0:10:55.520 with a copy of the original image that was used 0:10:55.520 --> 0:11:00.199 on the first device. This was really clever. It's a 0:11:00.200 --> 0:11:04.720 clever way to take an image, transform it into a signal, 0:11:05.280 --> 0:11:08.439 and then take that signal and transform it back into 0:11:08.480 --> 0:11:11.679 a copy of the image. All Right, we're gonna take 0:11:11.679 --> 0:11:13.680 a quick break here. When we come back, i'll talk 0:11:13.800 --> 0:11:18.760 more about how this centerfold image found its way into 0:11:18.880 --> 0:11:21.800 tech history, but first let's take a break to thank 0:11:21.880 --> 0:11:34.520 our sponsors. Okay, we're back. So I talked before the 0:11:34.559 --> 0:11:39.520 break about the actual technical approach toward wire photo scanning, 0:11:39.920 --> 0:11:44.040 and it's a very clever approach. However, it's also an 0:11:44.200 --> 0:11:48.000 analog approach, so, in other words, it's not directly compatible 0:11:48.040 --> 0:11:51.680 with the world of computers, which is a digital world. 0:11:51.760 --> 0:11:55.280 It's built on top of the concept of binary right 0:11:55.480 --> 0:11:58.320 zeros and ones. You can think of the computer as 0:11:58.400 --> 0:12:01.440 viewing the world as stuff that's I off or it's on, 0:12:02.080 --> 0:12:06.000 whereas analog is more like a continuum. This, by the way, 0:12:06.720 --> 0:12:11.559 is one of the many bases of arguments that audio 0:12:11.679 --> 0:12:17.040 files make that analog audio is inherently better than digital audio, 0:12:17.440 --> 0:12:22.600 because analog is a representation of an unbroken signal, whereas 0:12:23.120 --> 0:12:25.559 digital is a bunch of zeros and ones that give 0:12:25.559 --> 0:12:27.760 you little, tiny steps. And yes, the steps can be 0:12:27.880 --> 0:12:31.600 very very tiny, so tiny that to our perception it's 0:12:31.679 --> 0:12:34.640 no different than an unbroken signal. But if you get 0:12:34.679 --> 0:12:38.280 down far enough, it is it's broken up. And apparently 0:12:38.320 --> 0:12:42.960 that's enough to make digital worse than analog. Not every 0:12:43.000 --> 0:12:44.840 audio file believes that. By the way, I don't want 0:12:44.840 --> 0:12:47.360 to paint everyone with the same brush, but that is 0:12:47.520 --> 0:12:49.960 one of the arguments that audio files make. I don't 0:12:49.960 --> 0:12:52.120 personally buy into it. I do think you can reach 0:12:52.160 --> 0:12:56.839 a level of fidelity that is so indistinguishable from any 0:12:56.880 --> 0:13:00.600 other format that it makes no difference, but that's my 0:13:00.679 --> 0:13:05.760 own personal opinion anyway. The wire photo scanning approach is 0:13:05.920 --> 0:13:11.439 an analog technology. You're getting this variable electrical signal that's 0:13:11.559 --> 0:13:14.600 a continuous thing, not a bunch of zeros and ones. 0:13:15.040 --> 0:13:17.640 So in order to be able to process this in 0:13:17.679 --> 0:13:20.480 a computer and to be able to make a computer 0:13:20.559 --> 0:13:25.080 scan of this, the lab actually had to use analog 0:13:25.160 --> 0:13:28.640 to digital converters to connect to the scanner in order 0:13:28.679 --> 0:13:32.680 to change those signals into binary data that a computer 0:13:32.760 --> 0:13:36.720 could make sense of. Actually, the lab had three analog 0:13:36.800 --> 0:13:39.960 to digital converters. They had one for red, one for green, 0:13:40.040 --> 0:13:43.240 and one for blue, and collectively these three converters would 0:13:43.240 --> 0:13:47.040 supply all the information needed to recreate the image on 0:13:47.080 --> 0:13:51.040 a computer with the proper colors, because obviously if you 0:13:51.160 --> 0:13:53.360 didn't have that then you would have to work with 0:13:53.400 --> 0:13:57.360 a monochromatic digital image, right you would just have information 0:13:57.440 --> 0:13:59.760 as to the brightness or darkness, but you wouldn't have 0:14:00.400 --> 0:14:03.559 that related to things like what color that was. By 0:14:03.640 --> 0:14:08.400 using filters and three different digital analog to digital converters, 0:14:08.440 --> 0:14:12.680 they could recreate that. Now. The computational side of their 0:14:12.679 --> 0:14:15.760 work was handled by a Hewlett Packard twenty one hundred 0:14:15.800 --> 0:14:19.000 mini computer. Now we don't tend to use the term 0:14:19.160 --> 0:14:22.840 mini computer anymore these days, and the name might give 0:14:22.840 --> 0:14:26.600 you the wrong impression if you aren't aren't ancient like 0:14:26.680 --> 0:14:31.200 I am. Many computers were not teeny tiny machines despite 0:14:31.240 --> 0:14:35.720 the name mini computer. Many computers were still honking big computers. 0:14:35.760 --> 0:14:39.120 They could weigh more than two hundred pounds easy. Now 0:14:39.160 --> 0:14:42.960 they were called many computers because despite their size, they 0:14:43.040 --> 0:14:47.440 were smaller and less powerful than the giant mainframe machines 0:14:47.440 --> 0:14:50.640 you might find in some places. Many computers came out 0:14:50.640 --> 0:14:54.920 before the era of personal computers, but there was quite 0:14:54.920 --> 0:14:58.040 a bit of overlap between the era of many computers 0:14:58.120 --> 0:15:01.560 and the era of PCs did come out. While many 0:15:01.600 --> 0:15:04.960 computers were still very much the regular type of machine 0:15:05.000 --> 0:15:09.280 you would encounter in say a scientific lab, a research lab, 0:15:09.440 --> 0:15:12.360 or maybe a financial institution. That kind of thing you 0:15:12.400 --> 0:15:15.000 were more likely to run across a mini computer than 0:15:15.040 --> 0:15:18.440 a personal computer because for lots of years, PCs just 0:15:18.600 --> 0:15:22.120 couldn't match the performance of many computers, and a lot 0:15:22.120 --> 0:15:26.680 of organizations had higher needs than a PC could provide, 0:15:27.040 --> 0:15:29.680 but they didn't have such high demands that they needed 0:15:29.680 --> 0:15:33.760 to invest millions of dollars into a mainframe computer, So 0:15:33.800 --> 0:15:36.800 the mini computer was kind of the best solution. Now, 0:15:36.840 --> 0:15:39.680 by the time you get to the late nineties early 0:15:39.720 --> 0:15:44.320 two thousands, personal computers were at a performance level where 0:15:44.600 --> 0:15:48.920 many computers weren't really relevant anymore. Many computers also were 0:15:49.120 --> 0:15:51.960 reducing in size. There was kind of this convergence happening 0:15:52.240 --> 0:15:56.080 where PCs were essentially being able to do the stuff 0:15:56.080 --> 0:15:58.520 that the mini computers of the past could do. But 0:15:58.880 --> 0:16:01.240 back in the nineteen seventy, if you were doing serious 0:16:01.280 --> 0:16:04.360 scientific research on the computer, chances are it was a 0:16:04.400 --> 0:16:07.920 mini computer unless the geeks and the college mainframe center 0:16:07.960 --> 0:16:10.160 really liked you and gave you time to use on 0:16:10.240 --> 0:16:13.560 their machines. But time on a mainframe was a really 0:16:13.640 --> 0:16:17.800 sought after commodity anyway. The HP twenty one hundred was 0:16:17.880 --> 0:16:21.080 a sixteen bit machine that means it could handle data 0:16:21.160 --> 0:16:24.320 units that were sixteen bits wide. That means it could 0:16:24.400 --> 0:16:27.600 store two to the sixteenth power of values, which is 0:16:27.640 --> 0:16:31.360 sixty five, five hundred and thirty six values. These days, 0:16:31.600 --> 0:16:34.120 you probably own a PC running. I mean, at least 0:16:34.160 --> 0:16:37.320 it's a thirty two bit mode, if not a sixty 0:16:37.360 --> 0:16:40.080 four bit mode. So we have come a very long 0:16:40.160 --> 0:16:43.640 way from those days in the early nineteen seventies. Now, 0:16:43.680 --> 0:16:46.520 the scanning process itself took quite a bit of time, 0:16:46.760 --> 0:16:50.360 and it was a little bit finicky. In fact, during 0:16:50.400 --> 0:16:53.560 that first scan there were a couple of errors in 0:16:53.800 --> 0:16:57.040 the process. For one thing, there was a software issue, 0:16:57.480 --> 0:17:00.240 which meant that at the end of the day there 0:17:00.320 --> 0:17:03.840 was a single line missing from the scanned photo. It 0:17:03.880 --> 0:17:06.879 only had five hundred eleven lines, not five hundred twelve. 0:17:07.160 --> 0:17:10.199 So the team decided that in order to fix that, 0:17:10.440 --> 0:17:12.840 they would just copy the very top line of the 0:17:12.880 --> 0:17:16.000 image and then place it at the very top of it, 0:17:16.160 --> 0:17:19.240 so the top line of the image was there twice, 0:17:19.840 --> 0:17:22.280 and that way they got the five hundred and twelve 0:17:22.359 --> 0:17:24.800 lines that they wanted. They also found out that their 0:17:24.800 --> 0:17:29.080 analog to digital converters were not properly synchronized with the drum, 0:17:29.480 --> 0:17:32.800 so the image they had was just a touch distorted, 0:17:33.240 --> 0:17:36.720 not like to a terrible level, but it was a 0:17:36.800 --> 0:17:40.800 tiny bit elongated. Now, from what I understand, you wouldn't 0:17:40.840 --> 0:17:44.280 really know it unless you were comparing the scanned photo 0:17:44.359 --> 0:17:46.720 to the original photograph. It wasn't like it was to 0:17:46.800 --> 0:17:49.960 a point where it was, you know, disturbing or something. 0:17:50.480 --> 0:17:52.879 But you know, if you were going to compare it 0:17:52.920 --> 0:17:54.760 to the original, it means that you just brought a 0:17:54.800 --> 0:17:58.040 copy of Playboy into the conference and that probably is 0:17:58.080 --> 0:18:00.840 a reflection on your own sense of propriety. But who 0:18:00.880 --> 0:18:04.080 am I to judge. I'm somewhat amused that the team 0:18:04.160 --> 0:18:07.600 felt that replicating a top line and accepting a little 0:18:07.680 --> 0:18:11.680 deformation in their copied image were both within acceptable limits 0:18:11.760 --> 0:18:15.800 for their colleague's conference paper. So my guess is they 0:18:15.880 --> 0:18:19.080 must have really been under the gun to meet a deadline. Otherwise, 0:18:19.119 --> 0:18:21.879 I can't imagine that they wouldn't just try the scan 0:18:22.200 --> 0:18:25.399 again to get a better result. Then again, maybe the 0:18:25.440 --> 0:18:30.040 process really was so slow and cumbersome and unpredictable that 0:18:30.160 --> 0:18:32.879 no one really had the desire to give it another 0:18:33.040 --> 0:18:36.680 go without a guaranteed success. Right, So maybe the scan 0:18:36.840 --> 0:18:38.920 was actually better than what they normally got out of 0:18:38.920 --> 0:18:41.600 the process, and there, they felt they were lucky getting 0:18:41.600 --> 0:18:45.119 away with what they got. Whatever the reason, this imperfect 0:18:45.200 --> 0:18:49.080 scan was what made it into that conference presentation, and 0:18:49.200 --> 0:18:52.840 ultimately it's what would make Lenna something of a celebrity 0:18:52.960 --> 0:18:56.080 within the image processing world. Well, that and the fact 0:18:56.080 --> 0:19:00.199 that Lenna's photograph showcased her youth and beauty, and you know, 0:19:00.240 --> 0:19:03.080 the jaunt tilt of her hat with the long purple 0:19:03.119 --> 0:19:06.440 feather added some flare to it. And there's no denying 0:19:06.640 --> 0:19:09.560 that the photograph has just got great composition in lighting 0:19:09.920 --> 0:19:12.720 and that it captures Lenna's looks really well. I mean, 0:19:12.760 --> 0:19:15.800 it's a picture of a beautiful young woman, so it 0:19:15.960 --> 0:19:20.760 obviously had some captivating factors all by itself. Now, the 0:19:20.800 --> 0:19:24.080 story goes that attend days at this conference wanted to 0:19:24.080 --> 0:19:28.480 be able to test their own processes and methodologies against 0:19:28.480 --> 0:19:33.840 the Sippy Labs version. But to test like against like, 0:19:33.920 --> 0:19:36.280 you know, they would need to have two things. They 0:19:36.280 --> 0:19:40.120 would need a copy of Sippy's version of the scanned image, 0:19:40.359 --> 0:19:43.560 and they would need access to the original image so 0:19:43.600 --> 0:19:47.120 that they could perform their own scans and then compare 0:19:47.119 --> 0:19:50.080 them to Sippy's results. Well, access to the original was 0:19:50.080 --> 0:19:52.920 pretty easy because it was in a published magazine, which, 0:19:52.960 --> 0:19:56.000 from what I gather, was a pretty large circulation in general, 0:19:56.040 --> 0:20:00.720 but seemed particularly popular in computer engineering circles. So apparently 0:20:00.720 --> 0:20:06.359 there was no shortage of that original centerfold. The scanned 0:20:06.600 --> 0:20:09.320 version they would need to get from Sippy, but Sippy 0:20:09.640 --> 0:20:13.120 chose to share it liberally. They said, sure, yeah, well 0:20:13.160 --> 0:20:17.560 absolutely share this image of our scan. You can have 0:20:17.600 --> 0:20:19.719 a copy of it now. The result of all this 0:20:19.840 --> 0:20:23.040 is that Lenna's image became a de facto standard for 0:20:23.119 --> 0:20:27.560 testing scanning technologies and compression algorithms, and her photograph were 0:20:27.640 --> 0:20:32.360 rather more specifically, a scan of her photograph would find 0:20:32.400 --> 0:20:37.560 its way into countless journals and papers about the different 0:20:37.640 --> 0:20:42.480 methods for creating digital image files. This actually reminds me 0:20:42.560 --> 0:20:45.960 a little bit about the history of the MP three format. 0:20:46.119 --> 0:20:49.399 So you might remember that the engineers behind the MP 0:20:49.520 --> 0:20:52.000 three algorithm, you know, when they were trying to figure 0:20:52.000 --> 0:20:56.280 out the compression algorithm to use, they used a version 0:20:56.400 --> 0:21:01.879 of Suzanne Vegas Song Tom's Diner to calibrate their approach, 0:21:02.280 --> 0:21:05.640 so they would make tweaks to how the algorithm would 0:21:05.680 --> 0:21:08.760 choose which information to keep and which information could be 0:21:08.800 --> 0:21:13.000 tossed aside. That's how the MP three file format really works. 0:21:13.040 --> 0:21:16.600 The compression algorithm, and I'm talking about file size compression, 0:21:16.640 --> 0:21:21.760 not audio compression. The filesize compression algorithm looks at the 0:21:21.760 --> 0:21:25.600 elements of a sound file and says, what can we 0:21:25.640 --> 0:21:29.440 get rid of that isn't going to compromise the quality 0:21:29.440 --> 0:21:33.159 of the audio to such a degree that it's undesirable. 0:21:33.720 --> 0:21:38.320 So they would make changes to their algorithm. Then they 0:21:38.320 --> 0:21:42.040 would put Tom's Diner through the algorithm. Then they would 0:21:42.080 --> 0:21:45.760 listen to the compressed song and find out if the 0:21:45.840 --> 0:21:49.239 changes they made were manifest, if they were obvious when 0:21:49.280 --> 0:21:52.040 the song is played, they were perceptible. So the goal 0:21:52.119 --> 0:21:55.480 obviously was to maintain the song's quality as much as possible, 0:21:55.720 --> 0:21:58.239 even as they would reduce the file size. Well in 0:21:58.280 --> 0:22:02.840 some ways, this play always centerfold photograph of Lenna served 0:22:03.040 --> 0:22:06.840 much the same purpose, but in image research labs, you know, 0:22:07.160 --> 0:22:10.280 the labs that we're working on the future of digital imagery. 0:22:10.640 --> 0:22:14.000 And what's interesting to me about this part of it 0:22:14.040 --> 0:22:17.840 is that Playboy found out about this. The magazine found 0:22:17.840 --> 0:22:20.000 out about this, and in a move that is a 0:22:20.040 --> 0:22:24.080 little bit surprising considering how we typically see companies really 0:22:24.119 --> 0:22:28.719 move swiftly to protect their intellectual property. Playboy chose not 0:22:28.880 --> 0:22:32.119 to do anything about it. For one thing, the distribution 0:22:32.200 --> 0:22:36.240 of Lena's scanned image was pretty good publicity for the magazine. Plus, 0:22:36.280 --> 0:22:38.919 the various labs that wanted to test their own processes 0:22:38.960 --> 0:22:42.280 would need copies of the nineteen seventy two November issue 0:22:42.359 --> 0:22:46.200 to measure their results against what Sippy produced, So you 0:22:46.320 --> 0:22:49.520 had some guaranteed magazine sales out there. So Playboy's like, 0:22:49.560 --> 0:22:52.160 you know what, this is not bad for us. We're 0:22:52.280 --> 0:22:55.480 not gonna pursue the fact that this image that we 0:22:55.600 --> 0:23:00.119 owned the copyright too, is being used in various academic papers, 0:23:00.440 --> 0:23:04.440 in thousands of journals. As for Lena, she got word 0:23:04.440 --> 0:23:06.879 that her picture was being used by computer labs to 0:23:06.920 --> 0:23:10.199 advance digital imagery technology, and it sounds to me like 0:23:10.240 --> 0:23:13.120 she was pretty tickled by the whole thing. She certainly 0:23:13.680 --> 0:23:16.560 was happy to appear at a nineteen ninety seven conference 0:23:16.560 --> 0:23:20.359 in Boston. Playboy actually helped track her down so that 0:23:20.520 --> 0:23:25.000 she would be invited to this engineering conference. So there 0:23:25.040 --> 0:23:30.080 she goes, a former Playboy model appearing at an electrical 0:23:30.119 --> 0:23:34.720 engineering conference in Boston, Massachusetts, where she signed autographs for 0:23:34.840 --> 0:23:38.000 people who had been using her photograph while working on 0:23:38.080 --> 0:23:43.760 their own imagery projects like she was a celebrity. There. Okay, 0:23:44.040 --> 0:23:47.520 we've got more to say about this iconic image and 0:23:47.600 --> 0:23:49.919 its place in the tech sector and how that place 0:23:50.200 --> 0:23:53.200 is changing. But before we get to all that, let's 0:23:53.240 --> 0:24:05.840 take another quick break to thank our sponsors. All Right, 0:24:05.880 --> 0:24:09.840 we're back now. According to numerous sources that I came 0:24:09.840 --> 0:24:13.399 across when I was researching this, the image of Lenna 0:24:13.520 --> 0:24:18.320 found its way into literally thousands of papers and articles 0:24:18.680 --> 0:24:22.080 over the following decades, in the nineteen seventies, the eighties, 0:24:22.119 --> 0:24:25.560 the nineties, the two thousands. I mean, it's it's still 0:24:26.440 --> 0:24:30.280 found in circulating articles and papers to this day. And 0:24:30.960 --> 0:24:36.080 not only was it essentially ubiquitous in the tech literature 0:24:36.560 --> 0:24:40.760 and often used by people who had no idea of 0:24:41.000 --> 0:24:45.880