1
00:00:04,160 --> 00:00:07,520
Speaker 1: Get in text with technology with tech Stuff from stuff

2
00:00:07,520 --> 00:00:13,760
Speaker 1: works dot com and welcome to tech Stuff. I am

3
00:00:13,840 --> 00:00:17,360
Speaker 1: your host, Jonathan Strickland. I'm an executive producer at how

4
00:00:17,440 --> 00:00:22,200
Speaker 1: Stuff Works. I like you and I like technology, and

5
00:00:22,320 --> 00:00:24,200
Speaker 1: I don't really know that much about you. I just

6
00:00:24,239 --> 00:00:26,160
Speaker 1: get a good feeling, but I know a pretty good

7
00:00:26,160 --> 00:00:30,040
Speaker 1: amount about technology. Today, I want to talk about, you know,

8
00:00:30,760 --> 00:00:33,640
Speaker 1: tech and as I'm sure most of you guys know

9
00:00:33,680 --> 00:00:35,720
Speaker 1: by now, some of you know because you're doing it

10
00:00:35,800 --> 00:00:38,640
Speaker 1: at this moment. I live stream tech Stuff recordings on

11
00:00:38,680 --> 00:00:42,440
Speaker 1: Wednesdays and Fridays over at twitch dot tv slash tech Stuff.

12
00:00:42,840 --> 00:00:44,720
Speaker 1: There's a chat room over there, and I like to

13
00:00:44,760 --> 00:00:47,640
Speaker 1: spend some time chatting with viewers and listeners before the

14
00:00:47,680 --> 00:00:51,000
Speaker 1: show and during breaks. And during one of those breaks

15
00:00:51,000 --> 00:00:54,080
Speaker 1: in a recent show, one listener put forth a request

16
00:00:54,120 --> 00:00:56,160
Speaker 1: for a future episode of tech Stuff and the request

17
00:00:56,240 --> 00:01:00,320
Speaker 1: was for camera stabilization or image stabilization. So that's what

18
00:01:00,320 --> 00:01:04,920
Speaker 1: I'm gonna talk about the future is now. So I'm

19
00:01:04,920 --> 00:01:07,759
Speaker 1: gonna talk about image stabilization in general. And there are

20
00:01:07,760 --> 00:01:11,880
Speaker 1: two really big, really general ways to go about this.

21
00:01:12,720 --> 00:01:16,120
Speaker 1: One is to use mechanical elements to help compensate for

22
00:01:16,160 --> 00:01:19,600
Speaker 1: the little or sometimes not so little, jittery movements we

23
00:01:19,680 --> 00:01:22,440
Speaker 1: make as we try to capture video or still images

24
00:01:22,520 --> 00:01:27,120
Speaker 1: using handheld cameras. And the other is the software approach,

25
00:01:27,240 --> 00:01:31,240
Speaker 1: in which algorithms and programs attempt to reduce the jitters

26
00:01:31,240 --> 00:01:34,560
Speaker 1: sometimes reverse the jitters using some sort of trickery on

27
00:01:34,600 --> 00:01:36,759
Speaker 1: the back end. We're gonna look at both of those

28
00:01:36,800 --> 00:01:40,280
Speaker 1: different approaches in this episode. Now, before I do that,

29
00:01:40,600 --> 00:01:42,280
Speaker 1: I think it might be a good idea to talk

30
00:01:42,280 --> 00:01:45,240
Speaker 1: about how digital video cameras work in the first place,

31
00:01:45,319 --> 00:01:48,640
Speaker 1: so that you understand what is going on before we

32
00:01:48,720 --> 00:01:52,680
Speaker 1: even get into stabilization. That way, we can better understand

33
00:01:53,280 --> 00:01:56,560
Speaker 1: the actual mechanics of stabilization when we get there. So

34
00:01:56,920 --> 00:02:01,200
Speaker 1: let's crack open a digital camcorder. Let's say, you guys

35
00:02:01,240 --> 00:02:04,840
Speaker 1: remember cam quarters, right, I imagine there's still a thing.

36
00:02:05,000 --> 00:02:06,400
Speaker 1: I don't know. I haven't been in the market for

37
00:02:06,440 --> 00:02:09,280
Speaker 1: one for a really long time, But these days there

38
00:02:09,320 --> 00:02:11,280
Speaker 1: are lots of other devices out there that can capture

39
00:02:11,320 --> 00:02:15,040
Speaker 1: high resolution video, including several smartphones. So cam quarters a

40
00:02:15,160 --> 00:02:17,600
Speaker 1: term I don't really hear very much anymore. Anyway, that

41
00:02:17,639 --> 00:02:20,480
Speaker 1: doesn't matter, as the cam quarters inwards are pretty much

42
00:02:20,480 --> 00:02:22,320
Speaker 1: the same as you would find in any digital camera

43
00:02:22,360 --> 00:02:25,040
Speaker 1: device these days. Only now we've gotten really good at

44
00:02:25,080 --> 00:02:27,640
Speaker 1: maturizing those components, so you don't have to lug a

45
00:02:27,720 --> 00:02:30,360
Speaker 1: big thing on your shoulder, nor do you need something

46
00:02:30,360 --> 00:02:33,440
Speaker 1: that is part camera part VCR. And if you don't

47
00:02:33,440 --> 00:02:35,360
Speaker 1: know what a VCR is, you'll have to listen to

48
00:02:35,400 --> 00:02:37,120
Speaker 1: an older version of tech stuff. I'm not going to

49
00:02:37,200 --> 00:02:42,080
Speaker 1: go into that here. First, cameras record visual images. I

50
00:02:42,160 --> 00:02:45,720
Speaker 1: know that breaks it down to an incredibly simplistic and

51
00:02:45,800 --> 00:02:48,400
Speaker 1: almost silly statement, but it is something to keep in

52
00:02:48,440 --> 00:02:52,040
Speaker 1: mind that we are talking about visual information. Now there's

53
00:02:52,040 --> 00:02:56,440
Speaker 1: also audio information. Cameras mostly incorporate microphones these days. If

54
00:02:56,440 --> 00:02:58,400
Speaker 1: you have a video camera that doesn't have a microphone

55
00:02:58,720 --> 00:03:01,320
Speaker 1: and you aren't using external microphone, you might be an

56
00:03:01,360 --> 00:03:05,160
Speaker 1: avant garde performer, or you may just have a really

57
00:03:05,240 --> 00:03:09,919
Speaker 1: crappy camera. Breaking this down even further, cameras capture light.

58
00:03:10,040 --> 00:03:13,880
Speaker 1: Because everything we see is from light that's bouncing off

59
00:03:13,919 --> 00:03:18,440
Speaker 1: of surfaces or filtering through materials. Cameras attempt to capture

60
00:03:18,480 --> 00:03:22,120
Speaker 1: that combination of lighting effects in an effort to replicate

61
00:03:22,160 --> 00:03:25,720
Speaker 1: it in some way, typically to replicate in a way

62
00:03:25,760 --> 00:03:28,880
Speaker 1: that is as faithful as possible to the to what

63
00:03:28,919 --> 00:03:31,560
Speaker 1: we would see with our naked eyes, although obviously you

64
00:03:31,600 --> 00:03:35,000
Speaker 1: can set cameras so that you're capturing stuff and altering

65
00:03:35,000 --> 00:03:37,600
Speaker 1: it so it doesn't look the way it would naturally.

66
00:03:37,760 --> 00:03:40,320
Speaker 1: Thus you have all the different types of filters out there.

67
00:03:40,760 --> 00:03:44,240
Speaker 1: You Instagram fanatics out there know all about that, taking

68
00:03:44,520 --> 00:03:50,480
Speaker 1: photos with various filters, But generally speaking, cameras are capturing light.

69
00:03:51,560 --> 00:03:54,480
Speaker 1: Old cameras did this by focusing that light on photo

70
00:03:54,600 --> 00:03:57,880
Speaker 1: sensitive film. So you would take light, you would direct

71
00:03:57,920 --> 00:04:01,360
Speaker 1: it to this film, and that would alter the film chemically.

72
00:04:01,680 --> 00:04:04,080
Speaker 1: You would then develop the film, putting it through other

73
00:04:04,200 --> 00:04:08,040
Speaker 1: chemicals that would give you a negative, a reverse of

74
00:04:08,440 --> 00:04:10,880
Speaker 1: the image that you want, at least in color, and

75
00:04:10,920 --> 00:04:13,920
Speaker 1: then you would use that negative to produce prints, and

76
00:04:14,000 --> 00:04:17,240
Speaker 1: that would replicate what your eyes saw based upon what

77
00:04:17,360 --> 00:04:20,680
Speaker 1: the camera was capable of capturing. Assuming you had a

78
00:04:20,720 --> 00:04:23,720
Speaker 1: good camera, it would look pretty much the way you

79
00:04:23,760 --> 00:04:27,440
Speaker 1: saw it in person. Now, cameras typically use one or

80
00:04:27,480 --> 00:04:29,880
Speaker 1: more lenses to focus light in this way. In fact,

81
00:04:29,920 --> 00:04:32,760
Speaker 1: when we say a camera lens, you really mean a

82
00:04:32,800 --> 00:04:36,920
Speaker 1: series of lenses that are encased in some sort of

83
00:04:37,080 --> 00:04:40,080
Speaker 1: form factor. It's very easy to think of a camera

84
00:04:40,160 --> 00:04:43,560
Speaker 1: lens as being a single thing, because you might have

85
00:04:44,240 --> 00:04:47,000
Speaker 1: a camera body and then a selection of different lenses

86
00:04:47,000 --> 00:04:49,400
Speaker 1: where you can attach or detach them. But each of

87
00:04:49,440 --> 00:04:54,280
Speaker 1: those lenses contains several glass lenses inside of it. It's

88
00:04:54,279 --> 00:04:58,000
Speaker 1: not just a lens inside a casing. A digital cameras

89
00:04:58,000 --> 00:04:59,800
Speaker 1: by the way, you do this too. Uh. They do

90
00:05:00,360 --> 00:05:04,800
Speaker 1: use film to capture images. They use image sensors to

91
00:05:04,920 --> 00:05:07,919
Speaker 1: capture them, but they still use lenses to focus the

92
00:05:08,000 --> 00:05:12,479
Speaker 1: light onto the appropriate part of the camera. Uh. This

93
00:05:12,600 --> 00:05:16,880
Speaker 1: being the sensor as opposed to photosensitive film. The image

94
00:05:16,920 --> 00:05:20,400
Speaker 1: sensors are solid state devices, and the two most common

95
00:05:20,440 --> 00:05:23,240
Speaker 1: types of image sensors are cc D also known as

96
00:05:23,360 --> 00:05:28,760
Speaker 1: charge couple devices and c MOSS or CMOS complementary metal

97
00:05:28,760 --> 00:05:32,680
Speaker 1: oxide semiconductors. But what the heck does that even mean? Well,

98
00:05:32,839 --> 00:05:36,240
Speaker 1: CCD sensors are the older of the two, and they

99
00:05:36,400 --> 00:05:38,599
Speaker 1: lead the way for many many years in terms of

100
00:05:38,640 --> 00:05:41,719
Speaker 1: image quality. CCD was considered to be the top of

101
00:05:41,720 --> 00:05:45,640
Speaker 1: the image quality. Was also more expensive and less energy efficient,

102
00:05:45,839 --> 00:05:50,080
Speaker 1: but it produced the best images. A CCD sensor was

103
00:05:50,160 --> 00:05:53,160
Speaker 1: able to produce much sharper, higher quality images, but s

104
00:05:53,320 --> 00:05:57,599
Speaker 1: MOSS sensors are more efficient the more power efficient they're Also,

105
00:05:57,720 --> 00:06:01,080
Speaker 1: there's also some more built in functionality with MOSS sensors

106
00:06:01,120 --> 00:06:04,040
Speaker 1: than there are with C c d s. Plus, eventually

107
00:06:04,080 --> 00:06:07,120
Speaker 1: the C MOSS quality pretty much caught up with and

108
00:06:07,279 --> 00:06:11,360
Speaker 1: in some ways surpass C c ds. So while there

109
00:06:11,400 --> 00:06:16,799
Speaker 1: are differences, and they're still different kinds of of cameras

110
00:06:16,800 --> 00:06:19,120
Speaker 1: out there that have different kinds of sensors, and different

111
00:06:19,120 --> 00:06:23,400
Speaker 1: photographers will favor one versus the other, the differences in

112
00:06:23,600 --> 00:06:28,200
Speaker 1: quality between the two have largely diminished. It's not as

113
00:06:28,360 --> 00:06:32,760
Speaker 1: dramatic as it once was. Uh, some budget cameras have

114
00:06:32,880 --> 00:06:34,599
Speaker 1: C c d s, A lot of the better cameras

115
00:06:34,640 --> 00:06:38,200
Speaker 1: have C mosses, but that's not that's not necessarily across

116
00:06:38,240 --> 00:06:41,920
Speaker 1: the board. Like photo reactive film, those image sensors are

117
00:06:42,000 --> 00:06:46,240
Speaker 1: photo sensitive. So when light in the form of photons

118
00:06:46,640 --> 00:06:50,000
Speaker 1: hits the photo sites, photo sites are the little photo

119
00:06:50,040 --> 00:06:54,040
Speaker 1: sensitive elements on these image sensors. Each photo site essentially

120
00:06:54,080 --> 00:06:57,680
Speaker 1: is a pixel, so you get one photo site per pixel.

121
00:06:58,360 --> 00:07:00,760
Speaker 1: When when a photon hits a photos ight, this produces

122
00:07:00,800 --> 00:07:04,520
Speaker 1: an electrical charge. Now a bright light will end up

123
00:07:04,560 --> 00:07:07,800
Speaker 1: generating a stronger charge than a dem light will. The

124
00:07:07,839 --> 00:07:12,240
Speaker 1: photo sites must register the relative brightness of the light

125
00:07:12,320 --> 00:07:15,080
Speaker 1: that's hitting them, as well as the relative brightness of

126
00:07:15,120 --> 00:07:18,320
Speaker 1: the color of light that is red, green, or blue light,

127
00:07:18,680 --> 00:07:21,440
Speaker 1: in order to reproduce color images. Otherwise you're just gonna

128
00:07:21,440 --> 00:07:24,560
Speaker 1: get black and white images. Some sensors do this by

129
00:07:24,560 --> 00:07:28,640
Speaker 1: having a color filter above each pixel photo site, uh,

130
00:07:28,680 --> 00:07:31,680
Speaker 1: and they will have a different color filter over the

131
00:07:31,720 --> 00:07:35,080
Speaker 1: different photo sites. This is called a Bear filter array

132
00:07:35,360 --> 00:07:38,680
Speaker 1: is named after Bryce Beyer, who invented the array. It's

133
00:07:38,680 --> 00:07:43,239
Speaker 1: a filter consisting of a mosaic or array of red, green,

134
00:07:43,360 --> 00:07:46,400
Speaker 1: and blue filters above the photo sites on the sensor chip.

135
00:07:47,040 --> 00:07:49,280
Speaker 1: So it looks kind of like a checkerboard, except you've

136
00:07:49,280 --> 00:07:53,800
Speaker 1: got three colors, not two, and uh. One fourth of

137
00:07:53,880 --> 00:07:58,280
Speaker 1: those filters are red, one fourth of those filters are blue,

138
00:07:58,680 --> 00:08:01,400
Speaker 1: and the other half are green. So you have twice

139
00:08:01,400 --> 00:08:04,040
Speaker 1: as many green as you have either blue or red.

140
00:08:04,400 --> 00:08:06,360
Speaker 1: So why is that Why would you have twice as

141
00:08:06,360 --> 00:08:09,720
Speaker 1: many green filters? Well, that's to mimic the color sensitivity

142
00:08:09,800 --> 00:08:13,240
Speaker 1: of the human eye, because our color sensitivity is not

143
00:08:13,320 --> 00:08:17,280
Speaker 1: even across the board from a biological standpoint. Some filters,

144
00:08:17,280 --> 00:08:20,280
Speaker 1: some Bayer filters, will include different shades of green, so

145
00:08:20,360 --> 00:08:22,880
Speaker 1: you won't just have one single shade of green filter

146
00:08:23,000 --> 00:08:24,720
Speaker 1: you might have to you might have one that's like

147
00:08:24,760 --> 00:08:26,920
Speaker 1: a dark forest green and one that's kind of more

148
00:08:26,920 --> 00:08:30,280
Speaker 1: of a lighter green or something along those lines. Uh.

149
00:08:30,480 --> 00:08:34,480
Speaker 1: Some filters will use clear filters in place of some

150
00:08:34,559 --> 00:08:38,479
Speaker 1: of the green, and that's in order to improve light sensitivity,

151
00:08:38,480 --> 00:08:42,360
Speaker 1: but in general, that's how most of those filters work.

152
00:08:42,840 --> 00:08:45,840
Speaker 1: When the camera records an image, initially, it's a picture

153
00:08:45,880 --> 00:08:48,400
Speaker 1: that's just red, green, and blue in the raw file.

154
00:08:48,559 --> 00:08:50,720
Speaker 1: So if you were to look at a raw photo

155
00:08:50,760 --> 00:08:54,000
Speaker 1: image before any processing has happened, it would look really weird.

156
00:08:54,080 --> 00:08:58,240
Speaker 1: It wouldn't look color realistic at all. It could look

157
00:08:58,280 --> 00:09:01,840
Speaker 1: pretty awful. It's what photogra is often called false color.

158
00:09:02,360 --> 00:09:05,800
Speaker 1: But then you have this image processing unit that's part

159
00:09:05,880 --> 00:09:09,360
Speaker 1: of digital cameras, and it can take this raw file

160
00:09:09,440 --> 00:09:12,959
Speaker 1: and convert it into more natural imaging using what is

161
00:09:12,960 --> 00:09:17,559
Speaker 1: called a demosaic stage, So it takes all that jumblee

162
00:09:17,640 --> 00:09:20,840
Speaker 1: information and makes some meaning out of it. One cool

163
00:09:20,920 --> 00:09:23,520
Speaker 1: thing that happens with this software is how it determines

164
00:09:23,559 --> 00:09:26,880
Speaker 1: the color of an individual pixels. So remember images are

165
00:09:26,920 --> 00:09:30,400
Speaker 1: made up of millions of these pixels. A pixels essentially

166
00:09:30,400 --> 00:09:32,680
Speaker 1: a point of light, or if you prefer a single

167
00:09:32,760 --> 00:09:36,680
Speaker 1: point of color in an image, and the more pixels

168
00:09:36,720 --> 00:09:39,360
Speaker 1: you have, the smaller they Let's say that you have

169
00:09:39,400 --> 00:09:42,600
Speaker 1: the same dimensions of a photo. Let's start with that.

170
00:09:43,120 --> 00:09:45,800
Speaker 1: So let's say you're using an eight by ten photo.

171
00:09:45,960 --> 00:09:49,319
Speaker 1: This is just for the purposes of explaining pixels and resolution.

172
00:09:50,480 --> 00:09:53,199
Speaker 1: If you increase the resolution of an eight by ten photos,

173
00:09:53,200 --> 00:09:55,960
Speaker 1: so you're not changing the size of the image, You're

174
00:09:56,000 --> 00:09:59,160
Speaker 1: just changing the resolution. If you increase the resolution, that

175
00:09:59,200 --> 00:10:02,800
Speaker 1: means you're packing or pixels into that same physical space.

176
00:10:02,880 --> 00:10:06,280
Speaker 1: That means the pixels themselves have to be smaller. If

177
00:10:06,360 --> 00:10:10,480
Speaker 1: you are reducing the resolution, you're decreasing the number of pixels.

178
00:10:10,480 --> 00:10:13,320
Speaker 1: That means the size of each individual pixel gets larger.

179
00:10:14,120 --> 00:10:16,839
Speaker 1: So if you've ever seen an image where someone has

180
00:10:16,880 --> 00:10:21,720
Speaker 1: taken a digital photo that was for a very small

181
00:10:22,720 --> 00:10:25,240
Speaker 1: set of dimensions and then they've blown it up, it

182
00:10:25,240 --> 00:10:29,920
Speaker 1: looks very blocky or pixelated, that's because it was a

183
00:10:29,920 --> 00:10:33,040
Speaker 1: certain resolution and a certain size, and now you've stretched it,

184
00:10:33,320 --> 00:10:36,199
Speaker 1: so all of those pixels have individually been stretched as well.

185
00:10:36,880 --> 00:10:39,080
Speaker 1: If you want to increase resolution, then you have to

186
00:10:39,120 --> 00:10:42,400
Speaker 1: reduce pixel size to add more pixels to that image.

187
00:10:42,840 --> 00:10:45,720
Speaker 1: This only works to make an image sharper up to

188
00:10:45,800 --> 00:10:49,800
Speaker 1: a point. Higher resolution does not automatically mean a better

189
00:10:49,880 --> 00:10:53,000
Speaker 1: quality of picture. There are other elements that are also

190
00:10:53,120 --> 00:10:57,920
Speaker 1: very important, things like color representation, so in contrast ratio,

191
00:10:57,960 --> 00:11:00,840
Speaker 1: So it's not just resolute shan that you need to

192
00:11:00,840 --> 00:11:03,559
Speaker 1: think about. This is why we've done episodes of tech

193
00:11:03,600 --> 00:11:06,960
Speaker 1: stuff in the past where we've talked about the megapixel myth,

194
00:11:07,440 --> 00:11:09,200
Speaker 1: the idea that if you go out and buy a

195
00:11:09,240 --> 00:11:13,000
Speaker 1: camera that has more megapixels that automatically takes better photos

196
00:11:13,040 --> 00:11:16,840
Speaker 1: than a camera that has a lower megapixel count. That's

197
00:11:16,920 --> 00:11:20,200
Speaker 1: just not necessarily the case. It might be true, but

198
00:11:20,320 --> 00:11:22,800
Speaker 1: it's not because of the megapixels. It's because of other

199
00:11:22,880 --> 00:11:26,320
Speaker 1: elements as well. The only time you really have to

200
00:11:26,320 --> 00:11:29,880
Speaker 1: worry about really high megapixels counts as if you wanted

201
00:11:29,920 --> 00:11:31,960
Speaker 1: to take a photo and then blow it up to

202
00:11:32,000 --> 00:11:34,760
Speaker 1: a really large size and you didn't want to have

203
00:11:34,800 --> 00:11:37,120
Speaker 1: too much distortion when you were doing that. You want

204
00:11:37,120 --> 00:11:39,839
Speaker 1: a really high megapixel count so that you can do

205
00:11:39,880 --> 00:11:43,920
Speaker 1: that without losing too much on the resolution side. Well,

206
00:11:43,960 --> 00:11:47,040
Speaker 1: the way this image processing software tends to work is

207
00:11:47,080 --> 00:11:49,920
Speaker 1: it looks at each individual pixel and then it looks

208
00:11:49,960 --> 00:11:53,199
Speaker 1: at all that pixels neighbors and it starts to say, well,

209
00:11:53,240 --> 00:11:56,640
Speaker 1: based upon all the neighbors of this pixel, what color

210
00:11:56,720 --> 00:12:01,880
Speaker 1: should this specific pixel be. And it starts to extrapolate

211
00:12:01,960 --> 00:12:05,040
Speaker 1: information based on this and it makes some decisions about

212
00:12:05,080 --> 00:12:08,520
Speaker 1: what color each pixel should be based upon the nature

213
00:12:08,640 --> 00:12:12,040
Speaker 1: of its neighbors. And as it turns out, this software

214
00:12:12,120 --> 00:12:16,959
Speaker 1: is pretty good. It can reproduce colors fairly faithfully. And

215
00:12:17,000 --> 00:12:19,040
Speaker 1: that seems pretty interesting to me when you think that.

216
00:12:19,080 --> 00:12:20,839
Speaker 1: You know, when you how do you start with that?

217
00:12:21,000 --> 00:12:25,080
Speaker 1: Right when you first start, you don't have any necessarily

218
00:12:25,160 --> 00:12:28,000
Speaker 1: any natural colors. It's all the red, green, blue. But

219
00:12:28,480 --> 00:12:32,840
Speaker 1: by using this kind of process of deduction, the image

220
00:12:32,840 --> 00:12:37,720
Speaker 1: processing software can create a natural looking image. And it's

221
00:12:37,720 --> 00:12:40,480
Speaker 1: all about again looking at all the neighbors of this

222
00:12:40,559 --> 00:12:42,920
Speaker 1: one pixel to determine what color it should be. This

223
00:12:43,000 --> 00:12:45,600
Speaker 1: happens super fast. By the way, it sounds like something

224
00:12:45,600 --> 00:12:47,720
Speaker 1: that would take a really long time, but the software

225
00:12:47,760 --> 00:12:50,440
Speaker 1: is incredibly fast. Now. Bear filters are found in many,

226
00:12:50,480 --> 00:12:54,440
Speaker 1: but not all cameras. Some use other types of filter systems. Uh.

227
00:12:54,480 --> 00:12:57,960
Speaker 1: The fob on sensor has red, green, and blue filters

228
00:12:58,040 --> 00:13:01,000
Speaker 1: over every single photo site. So instead of having either

229
00:13:01,120 --> 00:13:04,839
Speaker 1: a red, or green, or blue filter over each photo site.

230
00:13:05,240 --> 00:13:08,640
Speaker 1: Every single photo site has all three, and some photographers

231
00:13:08,640 --> 00:13:14,479
Speaker 1: say that this produces more natural looking images. Others say

232
00:13:14,520 --> 00:13:19,800
Speaker 1: they are completely bonkers crazy wah wah, who uh. I

233
00:13:19,840 --> 00:13:23,440
Speaker 1: guess it really depends on your perspective and what you

234
00:13:23,520 --> 00:13:27,080
Speaker 1: have experience with um, because as far as I can tell,

235
00:13:27,280 --> 00:13:30,840
Speaker 1: it's not that different. But then my visual acuity is

236
00:13:30,880 --> 00:13:34,520
Speaker 1: nowhere near the level of say a professional photographers. The

237
00:13:34,600 --> 00:13:37,360
Speaker 1: important thing to remember here is that the image sensor

238
00:13:37,520 --> 00:13:41,840
Speaker 1: generates an electrical charge based off these photons, whether it's

239
00:13:42,000 --> 00:13:44,440
Speaker 1: measuring the red, green, or blue, or just the amount

240
00:13:44,440 --> 00:13:47,720
Speaker 1: of light in general, and the camera has to measure

241
00:13:47,760 --> 00:13:52,160
Speaker 1: that that electrical charge and convert that value into a

242
00:13:52,240 --> 00:13:55,080
Speaker 1: digital value, or convert that electrical charge into a digital

243
00:13:55,160 --> 00:13:59,559
Speaker 1: value with an analog to digital converter, and then you're

244
00:13:59,600 --> 00:14:02,720
Speaker 1: left with digital data for all the processing. So the

245
00:14:02,840 --> 00:14:05,440
Speaker 1: software does all the work on the back end, so

246
00:14:05,480 --> 00:14:08,079
Speaker 1: you've got the hard work on the front end where

247
00:14:08,120 --> 00:14:10,760
Speaker 1: the image has to her image sensor has to take

248
00:14:10,800 --> 00:14:16,679
Speaker 1: all these this light information converted into electrical charges, measure

249
00:14:16,720 --> 00:14:19,560
Speaker 1: that and convert that into digital information, and then the

250
00:14:19,600 --> 00:14:22,960
Speaker 1: software takes that digital information and interprets it to create

251
00:14:23,000 --> 00:14:25,880
Speaker 1: the image that you're looking at. There are other things

252
00:14:25,880 --> 00:14:28,520
Speaker 1: that take into consideration here, such as the shutter speed

253
00:14:28,600 --> 00:14:31,000
Speaker 1: of the camera. Shutter is a device that cuts off

254
00:14:31,120 --> 00:14:34,600
Speaker 1: light exposure to the sensor in the camera. The shutter

255
00:14:34,600 --> 00:14:40,040
Speaker 1: speed determines the exposure time of a camera's UH the

256
00:14:40,080 --> 00:14:43,640
Speaker 1: image that you're taking, So a short exposure is good

257
00:14:43,680 --> 00:14:45,560
Speaker 1: if you're trying to take an image of something that

258
00:14:45,760 --> 00:14:48,360
Speaker 1: is moving very quickly, but you need a lot of

259
00:14:48,440 --> 00:14:51,720
Speaker 1: light to do that. You don't you're not leaving the

260
00:14:51,720 --> 00:14:54,720
Speaker 1: shutter open very long for light to get to the

261
00:14:54,760 --> 00:14:57,680
Speaker 1: image sensor. The same was true of film cameras. You

262
00:14:57,680 --> 00:14:59,640
Speaker 1: need a very fast shutter speed, but you need a

263
00:14:59,680 --> 00:15:01,960
Speaker 1: lot of light in order for the light to be

264
00:15:02,000 --> 00:15:06,680
Speaker 1: able to register against the the photo sensitive material, whether

265
00:15:06,720 --> 00:15:10,240
Speaker 1: it was film or an image sensor. If you're using

266
00:15:10,240 --> 00:15:13,200
Speaker 1: a very slow shutter speed where the shutter is open

267
00:15:13,240 --> 00:15:15,600
Speaker 1: for a longer amount of time, and this is all

268
00:15:15,640 --> 00:15:18,080
Speaker 1: relative by the way, we're talking about fractions of a second.

269
00:15:18,920 --> 00:15:20,920
Speaker 1: If you're leaving the shutter speed open for a longer

270
00:15:20,920 --> 00:15:25,160
Speaker 1: amount of time, that's great for low light UH activities

271
00:15:25,160 --> 00:15:26,640
Speaker 1: like if you want to take an image of something

272
00:15:26,680 --> 00:15:30,120
Speaker 1: that's in really low light. Using a longer shutter speed

273
00:15:30,160 --> 00:15:34,960
Speaker 1: where it's open longer is a good idea, But any

274
00:15:35,080 --> 00:15:38,280
Speaker 1: motion is going to insert a lot of blur into

275
00:15:38,320 --> 00:15:41,760
Speaker 1: that image, So you want the you wanted to be

276
00:15:41,880 --> 00:15:45,920
Speaker 1: very still. If you're going to use a slower shutter speed.

277
00:15:46,160 --> 00:15:48,800
Speaker 1: If you wanted to do something like high speed camera

278
00:15:48,880 --> 00:15:51,320
Speaker 1: work where you're going to show something back in super

279
00:15:51,400 --> 00:15:55,400
Speaker 1: slow motion, that shutter is moving incredibly fast. You're talking

280
00:15:55,400 --> 00:16:00,960
Speaker 1: about taking thousands of images every single second. In order

281
00:16:01,000 --> 00:16:03,120
Speaker 1: to do that, you have to have it very very

282
00:16:03,160 --> 00:16:07,400
Speaker 1: well lit. So if you've ever been on a set

283
00:16:07,480 --> 00:16:12,240
Speaker 1: where they're shooting super high speed UH footage in order

284
00:16:12,280 --> 00:16:15,360
Speaker 1: to show back at at slow speed, then you know

285
00:16:15,360 --> 00:16:18,120
Speaker 1: what I'm talking about. The lights in those kind of

286
00:16:18,160 --> 00:16:21,280
Speaker 1: situations tend to be out of control. They're really actually

287
00:16:21,360 --> 00:16:24,920
Speaker 1: very much in control. They're just extremely well lit. So

288
00:16:25,280 --> 00:16:27,400
Speaker 1: the whole reason I wanted to talk about this bit

289
00:16:27,640 --> 00:16:29,920
Speaker 1: is that if you move a camera around while you're

290
00:16:29,960 --> 00:16:33,400
Speaker 1: taking images, whether it's snapshot or video, you can end

291
00:16:33,440 --> 00:16:35,840
Speaker 1: up with a jittery mess and it's really difficult to

292
00:16:35,880 --> 00:16:40,480
Speaker 1: hold completely still. If you've ever tried to just hold still, uh,

293
00:16:40,520 --> 00:16:42,480
Speaker 1: you know even if you've got the hands of a surgeon,

294
00:16:42,520 --> 00:16:44,480
Speaker 1: you're gonna notice a little bit of a jitter in there.

295
00:16:44,960 --> 00:16:47,800
Speaker 1: Most of us have some of that whenever we operate

296
00:16:47,800 --> 00:16:51,040
Speaker 1: a camera. Some of it's more obvious than others, and

297
00:16:51,840 --> 00:16:54,600
Speaker 1: a little bit people can tend to look past, but

298
00:16:54,680 --> 00:16:56,600
Speaker 1: more than a little bit, it gets very distracting. So

299
00:16:56,640 --> 00:16:58,960
Speaker 1: if you ever wanted to do something like move around

300
00:16:59,080 --> 00:17:03,200
Speaker 1: while you're shoeing video, you might have so much jetd

301
00:17:03,360 --> 00:17:05,400
Speaker 1: that's distracting, and you want to figure out a way

302
00:17:05,440 --> 00:17:07,760
Speaker 1: to reduce that, and we found numerous ways to cut

303
00:17:07,800 --> 00:17:10,879
Speaker 1: that down. Some ways just involved locking down the camera,

304
00:17:11,160 --> 00:17:14,000
Speaker 1: so you might put it in a tripod, and then

305
00:17:14,240 --> 00:17:18,960
Speaker 1: the camera's motion is strictly limited and you can operate

306
00:17:19,000 --> 00:17:20,959
Speaker 1: it without it having too much jetter. But then you

307
00:17:21,040 --> 00:17:24,760
Speaker 1: have a pretty static shot. You could use cranes and dollies,

308
00:17:24,840 --> 00:17:29,240
Speaker 1: which tend to have very smooth movement along certain directions,

309
00:17:29,240 --> 00:17:31,480
Speaker 1: but you are limited in the ways you can move

310
00:17:31,680 --> 00:17:35,640
Speaker 1: the camera. In both of those, you can't go anywhere

311
00:17:35,640 --> 00:17:37,560
Speaker 1: an actor can go. For example, if you wanted to

312
00:17:37,560 --> 00:17:41,399
Speaker 1: shoot a film, you could put the camera on a

313
00:17:41,560 --> 00:17:45,960
Speaker 1: dolly that isn't on a track. So there's some track systems.

314
00:17:46,119 --> 00:17:48,560
Speaker 1: Those are great. If you have uneven ground or rough

315
00:17:48,680 --> 00:17:50,760
Speaker 1: terrain and you put tracks down, you can have a

316
00:17:50,840 --> 00:17:53,560
Speaker 1: nice smooth camera operation, but again you're limited to just

317
00:17:53,680 --> 00:17:56,160
Speaker 1: moving on the tracks, or you could do a wheeled

318
00:17:56,320 --> 00:17:59,360
Speaker 1: dolly that can move around uh an area, but then

319
00:17:59,400 --> 00:18:02,280
Speaker 1: you need some that's an area that's pretty smooth and flat.

320
00:18:02,920 --> 00:18:05,080
Speaker 1: That really still limits what you can do with a camera.

321
00:18:05,640 --> 00:18:09,679
Speaker 1: So one of the ways that we have come up

322
00:18:09,760 --> 00:18:13,320
Speaker 1: with two improve camera operation and to be able to

323
00:18:13,359 --> 00:18:15,880
Speaker 1: go in the same places where it say performers can

324
00:18:15,920 --> 00:18:20,879
Speaker 1: go is with steadicams. Now. The steadicam was invented in

325
00:18:20,920 --> 00:18:23,880
Speaker 1: the mid nineteen seventies as a way for camera operators

326
00:18:23,920 --> 00:18:26,680
Speaker 1: to move around while running a camera and get smooth footage.

327
00:18:26,960 --> 00:18:30,400
Speaker 1: Whether it's with a documentary or with a film where

328
00:18:30,400 --> 00:18:33,560
Speaker 1: you're following actual characters. The steadicam can remove the jitter

329
00:18:33,760 --> 00:18:36,800
Speaker 1: created by walking or running. And when we move around

330
00:18:36,840 --> 00:18:39,440
Speaker 1: in our own bodies, this is happening all the time.

331
00:18:39,480 --> 00:18:42,040
Speaker 1: There's always this jitter as we're walking or running, but

332
00:18:42,080 --> 00:18:44,280
Speaker 1: we don't really notice it because our brains are really

333
00:18:44,280 --> 00:18:46,199
Speaker 1: good at smoothing all of that out to create a

334
00:18:46,200 --> 00:18:49,640
Speaker 1: more steady experience in our consciousness. So it's only when

335
00:18:49,640 --> 00:18:53,480
Speaker 1: you're really concentrating on it that you become aware of it.

336
00:18:53,480 --> 00:18:55,199
Speaker 1: It's sort of similar to how the world does not

337
00:18:55,320 --> 00:18:58,560
Speaker 1: spontaneously go away and then come back every time you

338
00:18:58,600 --> 00:19:01,160
Speaker 1: blink your eyes. It's only when you really think about

339
00:19:01,200 --> 00:19:03,720
Speaker 1: blinking your eyes that you start to notice it. And

340
00:19:03,760 --> 00:19:08,120
Speaker 1: now you're noticing it, aren't you sorry about that? My bad?

341
00:19:08,640 --> 00:19:10,840
Speaker 1: Don't worry. You'll you'll stop thinking about blinking your eyes

342
00:19:10,840 --> 00:19:12,879
Speaker 1: in just a minute and then everything will be fine.

343
00:19:13,440 --> 00:19:16,080
Speaker 1: So how does a steady cam stabilize the image so

344
00:19:16,080 --> 00:19:18,640
Speaker 1: that you can get the really cool effects you see

345
00:19:18,680 --> 00:19:21,920
Speaker 1: in movies like the Amazing copa shot in Good Fellas.

346
00:19:22,280 --> 00:19:24,520
Speaker 1: So if you're not familiar with the sequence I'm talking about,

347
00:19:24,920 --> 00:19:27,840
Speaker 1: it's it's in the movie. It's well in the film.

348
00:19:27,920 --> 00:19:30,880
Speaker 1: There's a sequence that's a little bit more than two

349
00:19:30,880 --> 00:19:34,320
Speaker 1: and a half minutes long, and it's a an uncut shot.

350
00:19:34,640 --> 00:19:38,040
Speaker 1: It tracks two characters as they cross a New York street.

351
00:19:38,560 --> 00:19:41,879
Speaker 1: They walk down some stairs into the kitchen of a

352
00:19:41,920 --> 00:19:46,080
Speaker 1: busy restaurant. They walk through the kitchen, passing by dozens

353
00:19:46,080 --> 00:19:48,800
Speaker 1: of extras as they're moving around in the kitchen as

354
00:19:48,840 --> 00:19:53,520
Speaker 1: cooks or or waiters, they emerge onto a dining floor

355
00:19:54,359 --> 00:19:56,439
Speaker 1: and they're seated right up front at a stage. And

356
00:19:56,480 --> 00:19:59,480
Speaker 1: this is all in a single uncut shot. And it

357
00:19:59,560 --> 00:20:02,119
Speaker 1: was using a film camera, not a digital camera. This

358
00:20:02,240 --> 00:20:05,640
Speaker 1: was in the early nineties, and there were other challenges there.

359
00:20:05,640 --> 00:20:08,240
Speaker 1: For example, the way the film camera worked. It was

360
00:20:08,280 --> 00:20:11,920
Speaker 1: actually pulling film from one side of the camera and

361
00:20:12,000 --> 00:20:15,040
Speaker 1: it would go in through the camera where you would

362
00:20:15,040 --> 00:20:18,480
Speaker 1: expose the film to light coming through the camera. The

363
00:20:18,560 --> 00:20:21,800
Speaker 1: exposed film would then go into a canister on the

364
00:20:21,800 --> 00:20:24,080
Speaker 1: other side of the camera. Now, what that meant was

365
00:20:24,119 --> 00:20:26,840
Speaker 1: that as you shot film, the weight of the camera

366
00:20:26,920 --> 00:20:32,400
Speaker 1: started to shift because the side worthy the the exposed

367
00:20:32,440 --> 00:20:34,879
Speaker 1: film was coming out that started getting heavier and heavier,

368
00:20:35,240 --> 00:20:37,399
Speaker 1: so you had to compensate for that. It's actually pretty

369
00:20:37,400 --> 00:20:40,359
Speaker 1: remarkable if you watched that sequence that it doesn't just

370
00:20:40,400 --> 00:20:43,480
Speaker 1: slowly start tilting towards the right because the camera was

371
00:20:43,520 --> 00:20:47,200
Speaker 1: getting progressively more heavy on the right side. By the way,

372
00:20:47,240 --> 00:20:52,000
Speaker 1: if you're curious, there's some great behind the scenes documentaries

373
00:20:52,040 --> 00:20:54,840
Speaker 1: and interviews about the copa shot. It's one of those

374
00:20:54,880 --> 00:20:57,840
Speaker 1: things that's talked about in film school. It took them

375
00:20:57,920 --> 00:21:01,160
Speaker 1: eight takes and that was it, and they actually finished

376
00:21:01,160 --> 00:21:03,640
Speaker 1: it in half a day of shooting, which if you've

377
00:21:03,640 --> 00:21:06,399
Speaker 1: ever been on a film shoot for a two and

378
00:21:06,440 --> 00:21:10,400
Speaker 1: a half minute sequence uncut, to get completed in eight

379
00:21:10,440 --> 00:21:13,680
Speaker 1: takes is pretty phenomenal. Also, it tells you the difference

380
00:21:13,720 --> 00:21:19,080
Speaker 1: between someone, say like Martin Scorsese and Stanley Kubrick, because

381
00:21:19,119 --> 00:21:21,960
Speaker 1: if Kubrick were shooting that, he would have still been

382
00:21:21,960 --> 00:21:27,439
Speaker 1: doing it years later. Uh. Anyway, the steadicam was a

383
00:21:27,480 --> 00:21:31,000
Speaker 1: big part of why this shot was even possible because

384
00:21:31,040 --> 00:21:34,640
Speaker 1: it meant following these actors through these different environments, including

385
00:21:34,720 --> 00:21:39,240
Speaker 1: downstairs and through a crowded kitchen, and it was something

386
00:21:39,280 --> 00:21:42,000
Speaker 1: that you just could not do on a track or

387
00:21:42,080 --> 00:21:45,600
Speaker 1: on a wheeled dolly. Uh. And it was all because

388
00:21:45,600 --> 00:21:47,679
Speaker 1: of the steadicam. Now, let's say he came itself was

389
00:21:47,720 --> 00:21:50,439
Speaker 1: invented again in the seventies, not in the nineties. It

390
00:21:50,520 --> 00:21:52,520
Speaker 1: was invented by a guy named Garrett Brown who was

391
00:21:52,560 --> 00:21:55,240
Speaker 1: a commercial director and a producer. He was not an engineer,

392
00:21:55,480 --> 00:21:57,520
Speaker 1: and he was just trying to come up with a

393
00:21:57,600 --> 00:22:01,320
Speaker 1: way to remove all this jetter that was coming around

394
00:22:01,400 --> 00:22:04,359
Speaker 1: whenever you wanted to do a handheld shot and you

395
00:22:04,400 --> 00:22:07,480
Speaker 1: wanted to follow an actor along a place where you

396
00:22:07,520 --> 00:22:11,880
Speaker 1: couldn't have these other traditional camera setups, and he got

397
00:22:11,880 --> 00:22:15,720
Speaker 1: a rough idea of it and called it the Brown's Stabilizer.

398
00:22:15,760 --> 00:22:19,359
Speaker 1: In nine three, fortunately, a more refined version would become

399
00:22:19,400 --> 00:22:23,280
Speaker 1: the first steadicam, and it consisted of three components. You

400
00:22:23,320 --> 00:22:28,639
Speaker 1: had an articulated iso elastic arm that would attach to

401
00:22:29,200 --> 00:22:33,640
Speaker 1: a sled. The sled was a kind of a it's

402
00:22:33,680 --> 00:22:36,399
Speaker 1: it's a housing for the camera and for all the

403
00:22:36,440 --> 00:22:39,640
Speaker 1: camera components. It looks like a big pole, and then

404
00:22:39,800 --> 00:22:42,040
Speaker 1: the various components attached to either the top or the

405
00:22:42,040 --> 00:22:43,560
Speaker 1: bottom of the pole, the camera at the top and

406
00:22:43,600 --> 00:22:46,720
Speaker 1: then everything else closer towards the bottom. And then a

407
00:22:47,200 --> 00:22:49,840
Speaker 1: vest that helped distribute the weight of the system and

408
00:22:49,880 --> 00:22:53,320
Speaker 1: provide more stability. So you had an iso elastic arm

409
00:22:53,400 --> 00:22:56,640
Speaker 1: that on one end would attach to the vest and

410
00:22:57,040 --> 00:23:00,359
Speaker 1: thus have that, you know, just weight distribution. On the

411
00:23:00,359 --> 00:23:04,280
Speaker 1: other end, it attached to the sled which held the

412
00:23:04,320 --> 00:23:07,680
Speaker 1: camera and its various components like a battery pack, a monitor,

413
00:23:08,240 --> 00:23:11,280
Speaker 1: that kind of thing. Now, the arm of the original

414
00:23:11,320 --> 00:23:13,720
Speaker 1: steadicam was a lot like a swing arm lamp with

415
00:23:13,760 --> 00:23:17,040
Speaker 1: a spring loaded arm. Uh if you ever look at

416
00:23:17,040 --> 00:23:18,760
Speaker 1: one of those. In fact, I have some right in

417
00:23:18,760 --> 00:23:20,920
Speaker 1: front of me right now, because the microphones I use

418
00:23:20,960 --> 00:23:24,920
Speaker 1: are on that this style of arm, you have two

419
00:23:25,119 --> 00:23:27,760
Speaker 1: bars that make up each segment of the arm, and

420
00:23:27,760 --> 00:23:32,760
Speaker 1: they're in parallel with one another. Uh. These two bars

421
00:23:32,920 --> 00:23:37,359
Speaker 1: would then end at metal blocks and they would also

422
00:23:37,520 --> 00:23:40,920
Speaker 1: join together with a pivot point, a pivot joint that

423
00:23:40,920 --> 00:23:43,520
Speaker 1: will allow them to kind of bend like a human

424
00:23:43,600 --> 00:23:46,920
Speaker 1: arm would, and the camera sled would fit onto the arm.

425
00:23:47,000 --> 00:23:49,879
Speaker 1: The sled would consist of that assembly that holds the camera,

426
00:23:50,040 --> 00:23:53,359
Speaker 1: the battery, the motor, counterweights, and some systems. All of

427
00:23:53,400 --> 00:23:55,840
Speaker 1: that's very important, and it's on was called the sled

428
00:23:55,920 --> 00:23:59,280
Speaker 1: pole central piece of the sled, and that fits into

429
00:23:59,320 --> 00:24:01,639
Speaker 1: the arm of the stead decam. Now, the way of

430
00:24:01,680 --> 00:24:06,520
Speaker 1: the sled and the camera is constantly pulling the arm

431
00:24:06,560 --> 00:24:10,080
Speaker 1: that's attached to the vest downward. Right. So imagine that

432
00:24:10,160 --> 00:24:13,879
Speaker 1: you're holding a weight at arms length. Let's say it's

433
00:24:13,920 --> 00:24:19,159
Speaker 1: a a fifteen pound weight. You're constantly feeling gravity pulling

434
00:24:19,240 --> 00:24:22,120
Speaker 1: down on that weight that's being held out arms length.

435
00:24:22,600 --> 00:24:24,680
Speaker 1: The steadicam arm is the same sort of thing. It's

436
00:24:24,720 --> 00:24:28,160
Speaker 1: holding up the sled and the camera. Well, the way

437
00:24:28,200 --> 00:24:29,960
Speaker 1: it holds it up and keeps it in the same

438
00:24:30,000 --> 00:24:33,720
Speaker 1: relative position, so that the camera doesn't just continuously sink

439
00:24:33,720 --> 00:24:38,080
Speaker 1: towards the floor. Is through this spring loaded system, it's

440
00:24:38,200 --> 00:24:41,160
Speaker 1: counteracting that downward force. The parallel metal bars in each

441
00:24:41,280 --> 00:24:46,919
Speaker 1: arm have these spring systems that are creating a force

442
00:24:47,000 --> 00:24:50,720
Speaker 1: that's directly opposite the downward force of the weight from

443
00:24:50,760 --> 00:24:55,640
Speaker 1: the sled and the camera. In fact, it's precisely calibrated

444
00:24:55,720 --> 00:24:58,840
Speaker 1: so it will maintain that that way. The only time

445
00:24:58,880 --> 00:25:00,760
Speaker 1: you change the elevation of the camera is if the

446
00:25:00,800 --> 00:25:04,439
Speaker 1: camera operator wants to. The camera operator can raise or

447
00:25:04,520 --> 00:25:07,639
Speaker 1: lower the camera, but it will then stay in that

448
00:25:07,720 --> 00:25:12,520
Speaker 1: relative position relative to the camera operator because the arms

449
00:25:12,880 --> 00:25:17,320
Speaker 1: spring loaded system is counteracting that downward pull from gravity.

450
00:25:17,560 --> 00:25:21,400
Speaker 1: It's really clever. It's also very difficult, obviously to explain

451
00:25:21,480 --> 00:25:25,639
Speaker 1: this in an audio format. Fortunately, there are numerous articles,

452
00:25:25,640 --> 00:25:28,480
Speaker 1: including one on How Stuff Works, about how steadicams work,

453
00:25:28,800 --> 00:25:32,560
Speaker 1: as well as videos on various platforms that show how

454
00:25:32,600 --> 00:25:36,199
Speaker 1: steadicam systems work to kind of help you get a

455
00:25:36,280 --> 00:25:39,960
Speaker 1: visual reference to what I'm talking about. So if you

456
00:25:40,359 --> 00:25:43,800
Speaker 1: still are having trouble imagining this, I highly recommend checking

457
00:25:43,800 --> 00:25:46,280
Speaker 1: out an article or watching a video to get a

458
00:25:46,359 --> 00:25:51,320
Speaker 1: little deeper understanding. To get that spring system UH into

459
00:25:51,320 --> 00:25:54,080
Speaker 1: further detail, it would require a very deep discussion, so

460
00:25:54,240 --> 00:25:56,679
Speaker 1: and again it's very difficult to explain it in an

461
00:25:56,680 --> 00:25:59,399
Speaker 1: audio format. So let's just say that there's a system

462
00:25:59,440 --> 00:26:03,520
Speaker 1: of pulleys and springs inside these arms that provide that

463
00:26:03,600 --> 00:26:05,919
Speaker 1: counter force, and will move on to talk about the

464
00:26:05,960 --> 00:26:10,160
Speaker 1: sled and the sled pole. So the sled includes an

465
00:26:10,200 --> 00:26:13,520
Speaker 1: element called a gimbal and a gimbals, a mechanism that

466
00:26:13,640 --> 00:26:17,000
Speaker 1: keeps something in a position of relative stability despite a

467
00:26:17,080 --> 00:26:21,840
Speaker 1: moving environment, typically horizontally, and in fact, gimbles have been

468
00:26:21,840 --> 00:26:24,520
Speaker 1: around for centuries and the good old days ships used

469
00:26:24,560 --> 00:26:28,000
Speaker 1: compasses mounted on gimbals, and that allowed the compass to

470
00:26:28,040 --> 00:26:30,919
Speaker 1: appear horizontal even as the ship was pitching or rolling

471
00:26:30,920 --> 00:26:33,919
Speaker 1: on rough seas. Typically gimbals consists of a couple of

472
00:26:34,000 --> 00:26:37,560
Speaker 1: rings pivoted at right angles to achieve this effect, and

473
00:26:37,640 --> 00:26:44,240
Speaker 1: so the UH, the the system around the compass moves,

474
00:26:44,280 --> 00:26:49,000
Speaker 1: but the compass itself appears to be steady relative to you.

475
00:26:49,920 --> 00:26:53,520
Speaker 1: The camera sled also changes a camera's center of gravity.

476
00:26:53,960 --> 00:26:56,680
Speaker 1: So the center of gravity, obviously that's the point at

477
00:26:56,680 --> 00:27:01,480
Speaker 1: which you can balance an object, and cameras typically there

478
00:27:01,560 --> 00:27:03,840
Speaker 1: their center of gravity, it doesn't make it very easy

479
00:27:03,920 --> 00:27:06,760
Speaker 1: to keep them nice and stable. But adding them to

480
00:27:06,800 --> 00:27:09,000
Speaker 1: the sled, if you add mass to something, you change

481
00:27:09,000 --> 00:27:12,679
Speaker 1: at center of gravity, and typically the the handle the

482
00:27:12,720 --> 00:27:17,080
Speaker 1: control system for a steadicam is located very close to

483
00:27:17,119 --> 00:27:21,119
Speaker 1: the actual center of gravity of the whole system. That

484
00:27:21,160 --> 00:27:23,920
Speaker 1: allows you to have much more control over the camera

485
00:27:24,240 --> 00:27:30,280
Speaker 1: and stabilize it. As such, the sled also changes the

486
00:27:30,280 --> 00:27:34,119
Speaker 1: the moment of inertia for the camera system. Uh. That

487
00:27:34,160 --> 00:27:38,280
Speaker 1: means that it becomes more resistant to rotation, and that's

488
00:27:38,320 --> 00:27:41,119
Speaker 1: really important for image stabilization. To the effect of a

489
00:27:41,119 --> 00:27:44,119
Speaker 1: steadicam is that you get this gliding shot that isn't

490
00:27:44,520 --> 00:27:47,879
Speaker 1: bothered by all the shakes of walking or running. And

491
00:27:47,920 --> 00:27:52,240
Speaker 1: there are tons of examples of steadicam shots in cinema.

492
00:27:52,880 --> 00:27:55,320
Speaker 1: Uh the Good Fellows shot being a really famous one.

493
00:27:55,560 --> 00:27:59,800
Speaker 1: Or there's also one in Kubricks The Shining there's a

494
00:27:59,800 --> 00:28:02,720
Speaker 1: part where Danny is running away from his dad in

495
00:28:02,800 --> 00:28:06,040
Speaker 1: a hedge maze in the snow, and we're following right

496
00:28:06,080 --> 00:28:09,280
Speaker 1: behind him as he's making turns left and right, and uh,

497
00:28:09,320 --> 00:28:12,960
Speaker 1: it's a steadicamp SHOT's pretty impressive. So that's another famous one.

498
00:28:12,960 --> 00:28:16,000
Speaker 1: But there are tons of them in movies now. As

499
00:28:16,000 --> 00:28:18,640
Speaker 1: we'll see in a bit, some cameras today incorporate equally

500
00:28:18,840 --> 00:28:22,560
Speaker 1: ingenious ways to reduce unwanted motion in footage that do

501
00:28:22,640 --> 00:28:25,320
Speaker 1: not require you to strap it to some sort of

502
00:28:25,400 --> 00:28:29,840
Speaker 1: complicated apparatus that you have to then wear so that

503
00:28:29,960 --> 00:28:32,240
Speaker 1: you can take some of the weight onto your own body.

504
00:28:33,040 --> 00:28:34,679
Speaker 1: And we'll talk about him in just a second, but

505
00:28:34,760 --> 00:28:37,520
Speaker 1: first let's take a quick break to thank our sponsor.

506
00:28:45,480 --> 00:28:47,800
Speaker 1: All right, So let's say you want to get smooth video,

507
00:28:48,040 --> 00:28:50,560
Speaker 1: or you plan on capturing still images, but you want

508
00:28:50,600 --> 00:28:53,040
Speaker 1: to use a lower shutter speed to capture in low light,

509
00:28:53,200 --> 00:28:55,880
Speaker 1: and you're gonna be holding the camera by hand, you're

510
00:28:55,880 --> 00:28:58,040
Speaker 1: gonna want to reduce that jitter, or else what you'll

511
00:28:58,040 --> 00:29:00,320
Speaker 1: get back won't look nearly as good as what you want,

512
00:29:00,520 --> 00:29:03,200
Speaker 1: what you saw in person, unless you're really going for

513
00:29:03,280 --> 00:29:07,080
Speaker 1: that shaky cam look, in which case image stabilization isn't

514
00:29:07,080 --> 00:29:10,120
Speaker 1: really your bag. Baby. For a camera to correct for

515
00:29:10,160 --> 00:29:12,880
Speaker 1: that kind of motion, you need a few elements. The first,

516
00:29:12,920 --> 00:29:16,240
Speaker 1: which you'll find in both of the major solutions to

517
00:29:16,360 --> 00:29:20,040
Speaker 1: this problem would be sensors that can detect camera motion.

518
00:29:20,280 --> 00:29:22,520
Speaker 1: For a camera to correct for a shaking motion. It

519
00:29:22,600 --> 00:29:24,840
Speaker 1: first has to detect that shaking is happening in the

520
00:29:24,880 --> 00:29:29,200
Speaker 1: first place, or else nothing can happen. So let's consider

521
00:29:29,240 --> 00:29:34,000
Speaker 1: the type of sensors found in these solutions that can

522
00:29:34,080 --> 00:29:37,000
Speaker 1: detect camera motions. So, for example, we'll talk about optical

523
00:29:37,080 --> 00:29:41,360
Speaker 1: image stabilization or o I S solutions. That's one of

524
00:29:41,400 --> 00:29:43,480
Speaker 1: the two that we're going to chat about in this section.

525
00:29:43,960 --> 00:29:46,920
Speaker 1: Optical image stabilization and in camera systems. Those are the

526
00:29:46,960 --> 00:29:50,600
Speaker 1: two big ones. Both of those rely on sensors to

527
00:29:50,680 --> 00:29:53,920
Speaker 1: detect when motion is happening. Otherwise there's no way to

528
00:29:54,080 --> 00:29:56,880
Speaker 1: compensate for it, right, You have to know something's happening

529
00:29:56,880 --> 00:30:02,600
Speaker 1: before you can fix it. Well, the optical image stabilization

530
00:30:02,680 --> 00:30:05,800
Speaker 1: uses the optical pathway, in other words, the lenses to

531
00:30:06,280 --> 00:30:09,480
Speaker 1: correct for camera shaking. And from a very basic standpoint,

532
00:30:09,480 --> 00:30:12,440
Speaker 1: the secret is that these lenses have a specific movable

533
00:30:12,600 --> 00:30:17,720
Speaker 1: lens of floating lens inside of them, so a an

534
00:30:17,800 --> 00:30:23,960
Speaker 1: actual lens singular inside this lens. Casing will be movable

535
00:30:24,120 --> 00:30:27,120
Speaker 1: with respect to the rest of the system inside of it.

536
00:30:27,120 --> 00:30:31,120
Speaker 1: It can shift left, right, up and down. Movable lenses

537
00:30:31,240 --> 00:30:33,760
Speaker 1: are pretty cool and I can't wait to talk more

538
00:30:33,760 --> 00:30:36,160
Speaker 1: about in a second, but again, in order to move

539
00:30:36,200 --> 00:30:37,959
Speaker 1: the lens to where it needs to be, you have

540
00:30:38,000 --> 00:30:40,760
Speaker 1: to detect the shaking in the first place. It's a

541
00:30:40,800 --> 00:30:44,360
Speaker 1: little complicated. So you've got these sensors, and with these solutions,

542
00:30:44,360 --> 00:30:49,720
Speaker 1: that's typically two piezo electric angular velocity sensors, which do

543
00:30:49,800 --> 00:30:52,840
Speaker 1: what gyroscopes do. And the reason you need two of

544
00:30:52,880 --> 00:30:56,440
Speaker 1: them is because each sensor really only detects motion along

545
00:30:56,520 --> 00:31:01,080
Speaker 1: one element of movement, like horizontal or vertical, So you

546
00:31:01,080 --> 00:31:04,000
Speaker 1: need two of them at sort of ninety degrees from

547
00:31:04,000 --> 00:31:07,960
Speaker 1: each other in order to detect vertical motion versus horizontal motion,

548
00:31:08,400 --> 00:31:10,840
Speaker 1: and together they can detect the motion you would find

549
00:31:10,840 --> 00:31:13,280
Speaker 1: in your typical camera shaking movements. They're not going to

550
00:31:13,320 --> 00:31:17,320
Speaker 1: be able to fix anything that's dramatic, so if you

551
00:31:17,840 --> 00:31:20,520
Speaker 1: whip the camera left or right or up or down,

552
00:31:20,880 --> 00:31:22,880
Speaker 1: it's not gonna be able to compensate for that. But

553
00:31:22,960 --> 00:31:27,320
Speaker 1: for just the small motions that our hands make while

554
00:31:27,320 --> 00:31:29,959
Speaker 1: we're holding a camera and trying to capture an image,

555
00:31:30,240 --> 00:31:35,680
Speaker 1: they can often correct for that. H So you gotta

556
00:31:35,720 --> 00:31:38,400
Speaker 1: keep that in mind. It's really the jitters that these

557
00:31:38,440 --> 00:31:43,520
Speaker 1: things fix, not the big stuff. Um. Also for the

558
00:31:43,520 --> 00:31:49,080
Speaker 1: optical image stabilizers, they cannot correct for rotation along the

559
00:31:49,160 --> 00:31:51,760
Speaker 1: optical axis. In other words, if you were to take

560
00:31:51,800 --> 00:31:54,120
Speaker 1: your camera. Let's say you're holding up a camera and

561
00:31:54,120 --> 00:31:58,000
Speaker 1: it's got a lock in on the way the image

562
00:31:58,040 --> 00:32:01,440
Speaker 1: is showing up like landscape, and you start rotating the

563
00:32:01,520 --> 00:32:05,040
Speaker 1: camera h so that you're doing either a clockwise or

564
00:32:05,040 --> 00:32:10,960
Speaker 1: counterclockwise motion, tilting the image. These sorts of sensors don't

565
00:32:11,600 --> 00:32:16,800
Speaker 1: detect that kind of momentum, that rotational momentum in that respect,

566
00:32:17,280 --> 00:32:20,880
Speaker 1: so you could get tilt, but no jitter in this case.

567
00:32:21,160 --> 00:32:23,400
Speaker 1: So how did those sensors work? Well, we gotta break

568
00:32:23,440 --> 00:32:27,080
Speaker 1: it down, and it's a little a little scary when

569
00:32:27,120 --> 00:32:30,800
Speaker 1: you see something saying piezo electric angular velocity sensors, What

570
00:32:30,840 --> 00:32:32,400
Speaker 1: the heck does that mean? Well, if you break it down,

571
00:32:32,440 --> 00:32:35,960
Speaker 1: it's not that tough. First, you've got piezo electric. You've

572
00:32:35,960 --> 00:32:38,280
Speaker 1: probably heard of the piezo electric effect. This refers to

573
00:32:38,320 --> 00:32:42,000
Speaker 1: the ability of certain materials to generate an alternating current voltage.

574
00:32:42,400 --> 00:32:46,320
Speaker 1: When those materials are subjected to mechanical stress or vibration.

575
00:32:47,280 --> 00:32:52,960
Speaker 1: They also will do the opposite. They will vibrate if

576
00:32:53,080 --> 00:32:57,600
Speaker 1: you subject them to an alternating current voltage, So it

577
00:32:57,640 --> 00:33:01,160
Speaker 1: goes either way. Quartz crystal do this, and that's why

578
00:33:01,200 --> 00:33:04,680
Speaker 1: they were used in watches and still are in some watches.

579
00:33:05,520 --> 00:33:10,840
Speaker 1: Uh So it's a very predictable behavior. If you know

580
00:33:11,080 --> 00:33:16,640
Speaker 1: the the the basics of that material, you can replicate

581
00:33:16,640 --> 00:33:18,160
Speaker 1: it over and over and over again. It's always going

582
00:33:18,200 --> 00:33:21,600
Speaker 1: to be the same. Next, we have the term angular velocity,

583
00:33:21,680 --> 00:33:24,280
Speaker 1: so that refers to the change in rotational angle along

584
00:33:24,280 --> 00:33:26,720
Speaker 1: an access per unit of time, and we express this

585
00:33:26,800 --> 00:33:30,920
Speaker 1: in degrees per second. That's how you measure angular velocity

586
00:33:30,960 --> 00:33:33,720
Speaker 1: in degrees per second. So a piece of electric angular

587
00:33:33,800 --> 00:33:37,160
Speaker 1: velocity sensor is one that detects changes in rotational angles

588
00:33:37,240 --> 00:33:41,240
Speaker 1: along an access by generating an alternating current voltage in

589
00:33:41,320 --> 00:33:45,000
Speaker 1: response to mechanical stress. And that mechanical stress is brought

590
00:33:45,040 --> 00:33:49,120
Speaker 1: to us courtesy of the Coriolis effect or the Coreolis force.

591
00:33:49,320 --> 00:33:52,000
Speaker 1: I really should say not the Coriolis effect, which is

592
00:33:52,040 --> 00:33:55,360
Speaker 1: a very specific thing that refers to an enormous system

593
00:33:55,400 --> 00:33:57,880
Speaker 1: called planet Earth. But the Coriolis force. This is an

594
00:33:57,920 --> 00:34:01,800
Speaker 1: inertial force that was first described by stove Gasparred Coriolis,

595
00:34:03,560 --> 00:34:08,719
Speaker 1: a French engineer of some renown, and as Encyclopedia Britannica

596
00:34:08,800 --> 00:34:13,000
Speaker 1: puts it, quote Coriolis showed that if the ordinary Newtonian

597
00:34:13,080 --> 00:34:15,400
Speaker 1: laws of motion of bodies are to be used in

598
00:34:15,440 --> 00:34:19,040
Speaker 1: a rotating frame of reference, and inertial force acting to

599
00:34:19,120 --> 00:34:22,480
Speaker 1: the right of the direction of body motion for counterclockwise

600
00:34:22,560 --> 00:34:25,840
Speaker 1: rotation of the reference frame or to the left for

601
00:34:25,920 --> 00:34:30,240
Speaker 1: a clockwise rotation, must be included in the equations of motion.

602
00:34:30,960 --> 00:34:36,960
Speaker 1: So essentially it's talking about detecting specific types of velocity

603
00:34:37,000 --> 00:34:40,040
Speaker 1: of changes in will not even changes, but just in

604
00:34:40,520 --> 00:34:44,719
Speaker 1: the direction and speed of motion. The important thing to

605
00:34:44,800 --> 00:34:47,360
Speaker 1: remember is that any motion would cause these piece of

606
00:34:47,440 --> 00:34:52,160
Speaker 1: electric sensors to deform slightly, so they can take different shapes.

607
00:34:52,200 --> 00:34:54,440
Speaker 1: And these are very tiny, tiny sensors, but they can

608
00:34:54,480 --> 00:34:57,759
Speaker 1: take different physical shapes like a tuning fork shape is

609
00:34:57,840 --> 00:35:04,280
Speaker 1: not uncommon. And as you move these things, it deforms them.

610
00:35:04,320 --> 00:35:07,919
Speaker 1: They change their shape slightly because of that inertia, and

611
00:35:08,000 --> 00:35:11,840
Speaker 1: that mechanical stress thus generates an alternating current voltage. Because

612
00:35:11,880 --> 00:35:16,120
Speaker 1: they are piezo electric materials, the sensor will detect that

613
00:35:16,440 --> 00:35:19,160
Speaker 1: change in voltage or that generation of voltage. And it

614
00:35:19,160 --> 00:35:22,600
Speaker 1: gets a little more technical from here, but it also

615
00:35:22,640 --> 00:35:25,359
Speaker 1: gets really complicated. So kind of like the steadicam, We're

616
00:35:25,360 --> 00:35:30,279
Speaker 1: gonna take a bird's eye view of this. Essentially, the

617
00:35:30,360 --> 00:35:34,880
Speaker 1: sensors consist of sensing arms and drive arms of this

618
00:35:34,920 --> 00:35:39,920
Speaker 1: piezo electric material, and the emotion sensed produces a potential difference,

619
00:35:39,960 --> 00:35:43,760
Speaker 1: an electrical potential difference. It's this potential difference that indicates

620
00:35:43,800 --> 00:35:47,080
Speaker 1: to the sensor the change in angular velocity, and the

621
00:35:47,120 --> 00:35:50,239
Speaker 1: sensor's output is an electrical signal which can then be

622
00:35:50,280 --> 00:35:54,000
Speaker 1: processed by a microcomputer. The other big element to this

623
00:35:54,080 --> 00:35:57,560
Speaker 1: optical image stabilization system is the movable lens, that floating

624
00:35:57,640 --> 00:36:01,960
Speaker 1: lens inside the overall camera lens. So this is completely

625
00:36:01,960 --> 00:36:05,320
Speaker 1: contained within a camera lens itself. If you've ever seen

626
00:36:05,840 --> 00:36:09,080
Speaker 1: a camera where you've got a body that can then

627
00:36:09,400 --> 00:36:12,000
Speaker 1: you can attach different types of lenses to it. You

628
00:36:12,800 --> 00:36:15,200
Speaker 1: attach the lens, and if you need a different lens,

629
00:36:15,200 --> 00:36:16,719
Speaker 1: you pop the first one off, you put a new

630
00:36:16,719 --> 00:36:21,120
Speaker 1: one on. This system is completely contained within those individual

631
00:36:21,680 --> 00:36:26,000
Speaker 1: lens cases. It's not in the actual camera body itself.

632
00:36:26,560 --> 00:36:30,440
Speaker 1: So we're talking about a piece of glass essentially that

633
00:36:30,560 --> 00:36:33,640
Speaker 1: shaped a very specific way, that's in a movable frame

634
00:36:34,200 --> 00:36:37,120
Speaker 1: inside this lens, and movable so that it can go up, down,

635
00:36:37,280 --> 00:36:42,400
Speaker 1: left or right, but it still remains aligned front and

636
00:36:42,480 --> 00:36:46,000
Speaker 1: back with the rest of the lens assembly. So it's

637
00:36:46,000 --> 00:36:48,320
Speaker 1: all meant to direct light back to the image sensor

638
00:36:48,400 --> 00:36:50,800
Speaker 1: in a proper way. And you typically have a lens

639
00:36:50,920 --> 00:36:54,120
Speaker 1: in the focusing group. That's the one that's closest to

640
00:36:54,120 --> 00:36:56,720
Speaker 1: the end, the part that you're you know, is facing

641
00:36:56,719 --> 00:36:59,200
Speaker 1: out to the outward world. That's the focusing group lens.

642
00:36:59,840 --> 00:37:02,600
Speaker 1: And then towards the back of the lens array you

643
00:37:02,640 --> 00:37:04,680
Speaker 1: have some other lenses that are meant to direct the

644
00:37:04,760 --> 00:37:08,560
Speaker 1: light properly. Between those two sets is where you put

645
00:37:08,560 --> 00:37:12,120
Speaker 1: the image stabilization lens. Really, you don't do it, the

646
00:37:12,200 --> 00:37:17,520
Speaker 1: lens manufacturer does. Don't open your lenses, that's crazy talk.

647
00:37:17,840 --> 00:37:22,040
Speaker 1: But the stabilization lens is in between these two other sets.

648
00:37:22,200 --> 00:37:24,680
Speaker 1: So you got the focusing lens at the at one

649
00:37:24,800 --> 00:37:28,120
Speaker 1: end of the the lens array. You've got the other

650
00:37:28,200 --> 00:37:30,319
Speaker 1: groups in the very back that are directing it towards

651
00:37:30,320 --> 00:37:32,680
Speaker 1: the image sensor. In between those, you have this floating

652
00:37:33,080 --> 00:37:38,040
Speaker 1: image stabilizer lens, and it's not fixed in relation to

653
00:37:38,080 --> 00:37:41,279
Speaker 1: the rest of the lens array the way the other

654
00:37:42,040 --> 00:37:46,680
Speaker 1: lenses are. So it's in this frame that uses electro

655
00:37:46,760 --> 00:37:50,200
Speaker 1: magnets to move the frame with respect to the rest

656
00:37:50,200 --> 00:37:56,000
Speaker 1: of the lens array, and it's this frame that reacts

657
00:37:56,040 --> 00:37:58,760
Speaker 1: to the information from the motion sensors. So the motion

658
00:37:58,800 --> 00:38:02,759
Speaker 1: sensors are picking up jet and based upon those electrical

659
00:38:03,480 --> 00:38:07,360
Speaker 1: potential differences, it's able to identify how much jeer is

660
00:38:07,400 --> 00:38:10,400
Speaker 1: coming up and down versus left and right. It sends

661
00:38:10,440 --> 00:38:14,160
Speaker 1: that information to the micro computer or micro controller rather

662
00:38:14,680 --> 00:38:18,680
Speaker 1: that controls the movement of the frame holding this image

663
00:38:18,719 --> 00:38:23,200
Speaker 1: stabilizer lens. The image stabilizer lens then is moved into

664
00:38:23,200 --> 00:38:26,400
Speaker 1: place so that the light coming in through the focusing

665
00:38:26,520 --> 00:38:29,760
Speaker 1: lens can be redirected towards the lenses in the back

666
00:38:30,000 --> 00:38:33,120
Speaker 1: of this lens array to thus go to the image

667
00:38:33,160 --> 00:38:36,839
Speaker 1: sensor in the very back of the camera. And it's

668
00:38:37,200 --> 00:38:40,560
Speaker 1: the end effect is supposed to be as if there

669
00:38:40,640 --> 00:38:43,319
Speaker 1: was not any jitter at all, as if there was

670
00:38:43,320 --> 00:38:46,400
Speaker 1: no broken line between the light coming in through the

671
00:38:46,400 --> 00:38:50,440
Speaker 1: focusing lens to the image sensor in the back. So,

672
00:38:51,239 --> 00:38:54,080
Speaker 1: if you want to think about in another way, imagine

673
00:38:54,080 --> 00:38:57,760
Speaker 1: that you've got a character, let's call him a Z's

674
00:38:58,360 --> 00:39:01,480
Speaker 1: and the ceas has a mirror, and the seas is

675
00:39:01,560 --> 00:39:04,480
Speaker 1: standing at the entrance of a cave and light is

676
00:39:04,520 --> 00:39:07,680
Speaker 1: coming in, and you need to have some light directed

677
00:39:07,800 --> 00:39:10,319
Speaker 1: from the outside of the cave into the cave. So

678
00:39:10,360 --> 00:39:13,200
Speaker 1: you can read some really funky hieroglyphics that someone wrote

679
00:39:13,280 --> 00:39:16,319
Speaker 1: years ago that suggested maybe aliens came down ages ago

680
00:39:16,760 --> 00:39:18,960
Speaker 1: and you yell a seas light and the seas has

681
00:39:19,000 --> 00:39:22,399
Speaker 1: to manipulate the mirror in such a way so that

682
00:39:22,520 --> 00:39:25,600
Speaker 1: light coming from outside is then reflected to go further

683
00:39:25,680 --> 00:39:29,080
Speaker 1: into the cave and illuminate the cave wall. That's what

684
00:39:29,160 --> 00:39:31,880
Speaker 1: this image stabilizer lens is doing, except instead of reflecting

685
00:39:31,960 --> 00:39:35,960
Speaker 1: light obviously, it's it's directing the light by making it

686
00:39:36,080 --> 00:39:39,879
Speaker 1: go through a specific part of the lens. And yes,

687
00:39:39,920 --> 00:39:42,640
Speaker 1: that was a fifth element reference for those of you

688
00:39:42,680 --> 00:39:44,640
Speaker 1: who are paying attention, And if you don't know what

689
00:39:44,760 --> 00:39:47,880
Speaker 1: that is, you should watch The Fifth Element because it's amazing.

690
00:39:48,120 --> 00:39:50,279
Speaker 1: It doesn't have a lot of steadicam shots in it,

691
00:39:50,480 --> 00:39:54,760
Speaker 1: but it's a great movie anyway. The cool thing about

692
00:39:54,880 --> 00:39:58,640
Speaker 1: this particular approach, this optical image stabilization approach, is that

693
00:39:58,680 --> 00:40:02,480
Speaker 1: it can turn any compatible camera into an image stabilized

694
00:40:02,520 --> 00:40:06,280
Speaker 1: camera because all the technology is inside the lens itself.

695
00:40:07,040 --> 00:40:09,840
Speaker 1: So if you have a lens that's got this image

696
00:40:09,840 --> 00:40:13,239
Speaker 1: stabilization system in it, then you can attach that to

697
00:40:13,280 --> 00:40:17,400
Speaker 1: any compatible camera and you get that image stabilization. The

698
00:40:17,480 --> 00:40:21,799
Speaker 1: downside is it really makes those lenses more expensive, and

699
00:40:21,880 --> 00:40:25,080
Speaker 1: lenses are not cheap to start off with, so if

700
00:40:25,120 --> 00:40:27,799
Speaker 1: you're building up a selection of lenses and you want

701
00:40:27,840 --> 00:40:32,040
Speaker 1: them all to have image stabilization capabilities, you start really

702
00:40:32,160 --> 00:40:35,720
Speaker 1: racking up the cost pretty quickly. But there is another

703
00:40:35,800 --> 00:40:39,399
Speaker 1: route to go, and that is called the in camera stabilizer.

704
00:40:39,440 --> 00:40:41,839
Speaker 1: This takes a different approach to stabilization, but it uses

705
00:40:41,880 --> 00:40:45,880
Speaker 1: a very similar philosophy. So instead of having a floating

706
00:40:46,000 --> 00:40:49,120
Speaker 1: lens internally in that array that can move around and

707
00:40:49,160 --> 00:40:54,520
Speaker 1: redirect light, UH, it actually has the image sensor itself

708
00:40:54,719 --> 00:40:58,239
Speaker 1: on a movable frame. So this would be as if

709
00:40:58,280 --> 00:41:01,080
Speaker 1: you could move that cave wall so that it was

710
00:41:01,120 --> 00:41:03,400
Speaker 1: in line with the light. As opposed to moving the

711
00:41:03,440 --> 00:41:06,520
Speaker 1: mirror to redirect the light to the cave wall, you're

712
00:41:06,520 --> 00:41:10,680
Speaker 1: actually moving the sensor itself. Otherwise it's behaving pretty much

713
00:41:10,760 --> 00:41:13,279
Speaker 1: the same way as the optical image stabilizer. These are

714
00:41:13,320 --> 00:41:18,040
Speaker 1: also sometimes called mechanical image stabilizers UH to differentiate the two,

715
00:41:18,160 --> 00:41:22,040
Speaker 1: or in camera image stabilizers. Because it's inside the camera,

716
00:41:22,120 --> 00:41:24,320
Speaker 1: the body of the camera itself, not inside the lenses.

717
00:41:24,560 --> 00:41:26,960
Speaker 1: So the big pro here is that if you've got

718
00:41:26,960 --> 00:41:28,920
Speaker 1: that in the camera, it doesn't matter what kind of

719
00:41:29,000 --> 00:41:31,799
Speaker 1: lens you use, because it's already got the image stabilization

720
00:41:31,840 --> 00:41:35,400
Speaker 1: in there. The downside is, if you're using lenses for

721
00:41:36,280 --> 00:41:39,080
Speaker 1: something that's further away, you're using like zoom lenses that

722
00:41:39,160 --> 00:41:42,560
Speaker 1: kind of thing, the reduction in jender is less effective

723
00:41:42,600 --> 00:41:44,560
Speaker 1: than it would be if you were using the optical

724
00:41:44,600 --> 00:41:49,439
Speaker 1: image stabilizer. So depending upon your use, one may be

725
00:41:49,440 --> 00:41:52,400
Speaker 1: better than the other. Depending on your budget, the in

726
00:41:52,520 --> 00:41:54,719
Speaker 1: camera version may be better. And in fact, you can

727
00:41:54,760 --> 00:41:58,640
Speaker 1: find this specific type of image stabilization, and lots of

728
00:41:58,680 --> 00:42:02,440
Speaker 1: smartphones out there actually have this capability of moving the

729
00:42:02,480 --> 00:42:07,319
Speaker 1: image sensor tiny tiny amounts to reduce jitter. Not all

730
00:42:07,360 --> 00:42:08,759
Speaker 1: of them do this, by the way. Some of them

731
00:42:08,840 --> 00:42:11,520
Speaker 1: use post process image stabilization, which we'll talk about in

732
00:42:11,520 --> 00:42:16,920
Speaker 1: a minute. So it's kind of a neat an elegant solution. Um,

733
00:42:16,960 --> 00:42:19,359
Speaker 1: if you have stabilized your camera, let's say that you've

734
00:42:19,400 --> 00:42:21,919
Speaker 1: got one of these two systems in place in your

735
00:42:22,080 --> 00:42:28,239
Speaker 1: camera system. Uh, and let's say that you've decided to

736
00:42:28,239 --> 00:42:30,279
Speaker 1: to really lock down your camera. Let's say that you

737
00:42:30,280 --> 00:42:32,840
Speaker 1: put it on a tripod, you actually probably want to

738
00:42:32,840 --> 00:42:37,120
Speaker 1: turn off image stabilization at that point, and if you

739
00:42:37,200 --> 00:42:40,040
Speaker 1: don't turn it off, the motions of the stabilizer itself

740
00:42:40,080 --> 00:42:42,279
Speaker 1: can end up being picked up by the system, which

741
00:42:42,320 --> 00:42:45,799
Speaker 1: then tries to compensate for that movement. Which puts me

742
00:42:45,840 --> 00:42:47,480
Speaker 1: in mind of the time I was in college and

743
00:42:47,520 --> 00:42:49,879
Speaker 1: tried tight rope walking for the first and only time

744
00:42:49,920 --> 00:42:52,160
Speaker 1: in my life. So let me explain, because this analogy

745
00:42:52,200 --> 00:42:55,320
Speaker 1: I think is very apt. When I tried tight rope walking,

746
00:42:55,520 --> 00:42:57,959
Speaker 1: I stepped on the tight rope and it was really

747
00:42:58,040 --> 00:43:00,960
Speaker 1: something between a tight rope and a slide rope, and

748
00:43:01,040 --> 00:43:03,279
Speaker 1: as I put weight on my leg, my leg began

749
00:43:03,320 --> 00:43:05,880
Speaker 1: to shake. Now that caused the rope to move around,

750
00:43:06,320 --> 00:43:09,560
Speaker 1: and I tried to catch my balance but found myself overcompensating,

751
00:43:09,680 --> 00:43:12,160
Speaker 1: so I would shift and then the rope would move

752
00:43:12,160 --> 00:43:13,839
Speaker 1: the other way, and I try and shift again, and

753
00:43:14,040 --> 00:43:17,840
Speaker 1: it was just getting worse and worse, and it caused

754
00:43:17,840 --> 00:43:20,520
Speaker 1: my leg to shake more, and instead of studying myself,

755
00:43:20,560 --> 00:43:22,160
Speaker 1: I just found the lower half of my body was

756
00:43:22,200 --> 00:43:24,399
Speaker 1: going crazy and not in a fun way. So, after

757
00:43:24,440 --> 00:43:27,839
Speaker 1: approximately fifteen seconds of trying this, I realized that tight

758
00:43:27,920 --> 00:43:29,719
Speaker 1: ropes were probably just one of those things I wasn't

759
00:43:29,719 --> 00:43:33,520
Speaker 1: meant to experience, and I stopped. Now, the image stabilizer

760
00:43:33,560 --> 00:43:36,560
Speaker 1: system kind of behaves the way my legs did on

761
00:43:36,640 --> 00:43:39,680
Speaker 1: that tight rope in these situations. So you've got your

762
00:43:39,760 --> 00:43:43,239
Speaker 1: image stabilized camera on a tripod and your image stabilization

763
00:43:43,360 --> 00:43:46,440
Speaker 1: is turned on. A little motion within the system itself

764
00:43:46,719 --> 00:43:49,319
Speaker 1: could be picked up as a camera shake, so then

765
00:43:49,360 --> 00:43:51,520
Speaker 1: it tries to compensate, but because the camera is not

766
00:43:51,600 --> 00:43:55,000
Speaker 1: actually shaking, it's on a steady tripod, uh, it then

767
00:43:55,120 --> 00:43:58,439
Speaker 1: starts to detect the solution for that shake as its

768
00:43:58,440 --> 00:44:01,960
Speaker 1: own problem, and it becomes this feedback loop. And this

769
00:44:02,000 --> 00:44:04,760
Speaker 1: can actually introduce motion, blur and jitter in your photos

770
00:44:05,160 --> 00:44:07,680
Speaker 1: even and your video even though you've got your camera

771
00:44:07,719 --> 00:44:10,680
Speaker 1: mounted to a stable tripod. So you might want to

772
00:44:10,760 --> 00:44:15,480
Speaker 1: turn off image stabilization in that process unless you plan

773
00:44:15,560 --> 00:44:18,200
Speaker 1: on doing a panning shot. So panning shots can be

774
00:44:18,239 --> 00:44:20,600
Speaker 1: tricky to you want a nice even pan You're either

775
00:44:20,640 --> 00:44:22,960
Speaker 1: going left to right or right to left, and you

776
00:44:22,960 --> 00:44:26,680
Speaker 1: want to avoid introducing any jitter in the image. Vertically. Well,

777
00:44:26,760 --> 00:44:30,040
Speaker 1: some image stabilizers have a panning mode so that way

778
00:44:30,040 --> 00:44:34,000
Speaker 1: they compensate for vertical movements, but not horizontal movements, and

779
00:44:34,000 --> 00:44:36,239
Speaker 1: those systems will remove some of that jitter while still

780
00:44:36,239 --> 00:44:39,560
Speaker 1: allowing for a nice smooth panning motion. With both optical

781
00:44:39,560 --> 00:44:42,640
Speaker 1: image stabilizers and mechanical image stabilizers, it took a lot

782
00:44:42,680 --> 00:44:44,880
Speaker 1: of engineering to figure out precisely how to make the

783
00:44:44,920 --> 00:44:47,640
Speaker 1: internal mechanisms respond in such a way to preserve the

784
00:44:47,680 --> 00:44:51,919
Speaker 1: integrity of an image while simultaneously removing jitter. And while

785
00:44:51,960 --> 00:44:54,360
Speaker 1: the basics could be found in much older technologies and

786
00:44:54,440 --> 00:44:58,319
Speaker 1: use for centuries, like the gimbal, getting that precision and

787
00:44:58,360 --> 00:45:00,719
Speaker 1: response time to a level that is useful in a

788
00:45:00,800 --> 00:45:04,000
Speaker 1: dynamic use case such as taking video or photos required

789
00:45:04,480 --> 00:45:07,680
Speaker 1: a whole lot of engineering. So I'm really impressed at

790
00:45:07,719 --> 00:45:11,000
Speaker 1: this technology. But we still have one more variation to

791
00:45:11,040 --> 00:45:13,719
Speaker 1: talk about because some systems don't use any moving parts

792
00:45:13,760 --> 00:45:16,800
Speaker 1: at all to create image stabilization. Instead, they try to

793
00:45:16,840 --> 00:45:20,280
Speaker 1: beat the problem with software in a post process solution.

794
00:45:20,719 --> 00:45:23,360
Speaker 1: So how does that work? Well, we'll find out, but

795
00:45:23,480 --> 00:45:26,320
Speaker 1: first let's take another quick break to thank our sponsor.

796
00:45:34,239 --> 00:45:37,680
Speaker 1: Some digital cameras, particularly some smartphones, have what is called

797
00:45:37,840 --> 00:45:42,400
Speaker 1: virtual image stabilization or electronic image stabilization, or sometimes just

798
00:45:42,800 --> 00:45:47,040
Speaker 1: post process image stabilization, and these systems don't use moving

799
00:45:47,080 --> 00:45:50,399
Speaker 1: parts to adjust sensors or lenses in order to keep

800
00:45:50,440 --> 00:45:54,000
Speaker 1: the image nice and steady. Instead, they use software. So

801
00:45:54,200 --> 00:45:57,759
Speaker 1: what's going on here, Well, from a high level perspective,

802
00:45:57,880 --> 00:46:01,480
Speaker 1: a program attempts to reverse any shake found in an

803
00:46:01,520 --> 00:46:04,959
Speaker 1: image algorithmically, and some do this in a pretty basic way.

804
00:46:05,200 --> 00:46:08,719
Speaker 1: Imagine you have a video open on YouTube. So let's

805
00:46:08,719 --> 00:46:12,240
Speaker 1: just say you've got a regular video. It's not full frame,

806
00:46:12,480 --> 00:46:18,800
Speaker 1: it's just the video in YouTube's desktop application. Now, imagine

807
00:46:18,840 --> 00:46:23,839
Speaker 1: that the actual video footage extends beyond the borders of

808
00:46:23,880 --> 00:46:26,160
Speaker 1: that video frame. I mean, even if you were looking

809
00:46:26,200 --> 00:46:28,440
Speaker 1: at it in full screen mode. Imagine that the video

810
00:46:28,440 --> 00:46:32,360
Speaker 1: itself would extend a little bit beyond every single border,

811
00:46:32,560 --> 00:46:35,040
Speaker 1: just a touch. So what you're seeing is not the

812
00:46:35,080 --> 00:46:38,720
Speaker 1: full frame of video. It's a section, a cropped section

813
00:46:38,920 --> 00:46:41,600
Speaker 1: of that video. The edges are cut off because those

814
00:46:41,680 --> 00:46:44,280
Speaker 1: edges provide some cheat room, a little bit of buffer

815
00:46:44,880 --> 00:46:49,320
Speaker 1: for the purposes of image stabilization. Now, this approach relies

816
00:46:49,360 --> 00:46:52,959
Speaker 1: on some assumptions. The program does not necessarily know which

817
00:46:53,000 --> 00:46:55,600
Speaker 1: elements of a video you're really interested in, so it

818
00:46:55,680 --> 00:46:58,799
Speaker 1: has to make some guesses. Let's say that you're taking

819
00:46:58,800 --> 00:47:00,919
Speaker 1: a video of a kid running uh and the kids

820
00:47:01,000 --> 00:47:04,160
Speaker 1: running out though through the snow in a hedge maze,

821
00:47:04,280 --> 00:47:06,560
Speaker 1: and you're running after the kid. Maybe you're shouting out

822
00:47:06,600 --> 00:47:09,239
Speaker 1: encouraging words about this nifty hotel you've been hired to

823
00:47:09,280 --> 00:47:12,360
Speaker 1: look after, and your video footage would be really shaky

824
00:47:12,400 --> 00:47:14,600
Speaker 1: because you're holding the camera as you're running. You don't

825
00:47:14,600 --> 00:47:17,480
Speaker 1: have a steady camp, so every single run step you're

826
00:47:17,520 --> 00:47:20,839
Speaker 1: taking it's it's jittering the camera. But feeding the footage

827
00:47:20,880 --> 00:47:24,239
Speaker 1: through an algorithm can smooth things out a bit, and

828
00:47:24,280 --> 00:47:28,160
Speaker 1: the algorithm recognizes that the kid you're chasing is the

829
00:47:28,200 --> 00:47:30,319
Speaker 1: interesting thing in the frame. It's doing this through some

830
00:47:30,520 --> 00:47:33,080
Speaker 1: processing and figuring out which pixels are changing the most

831
00:47:33,120 --> 00:47:35,680
Speaker 1: and which ones are staying more or less the same,

832
00:47:35,800 --> 00:47:39,160
Speaker 1: and kind of drawing some conclusions based on that. So

833
00:47:39,160 --> 00:47:42,720
Speaker 1: while the camera shakes around, the algorithm repositions the frame

834
00:47:42,800 --> 00:47:46,319
Speaker 1: of view for the audience, so the kid remains more

835
00:47:46,400 --> 00:47:49,280
Speaker 1: or less in the same general area relative to the screen.

836
00:47:49,480 --> 00:47:51,600
Speaker 1: So you can think of it as like a a

837
00:47:51,600 --> 00:47:54,799
Speaker 1: picture in picture sort of thing, and the picture in

838
00:47:54,880 --> 00:47:57,759
Speaker 1: picture that frame is moving around with relation to the

839
00:47:57,800 --> 00:48:01,960
Speaker 1: rest of the the the the big picture view. But

840
00:48:02,600 --> 00:48:06,520
Speaker 1: if you just stare at whatever is happening inside that

841
00:48:06,600 --> 00:48:09,000
Speaker 1: picture within picture, it looks nice and steady, or at

842
00:48:09,040 --> 00:48:13,600
Speaker 1: least compared to the overall image. Uh. This is not

843
00:48:13,680 --> 00:48:18,680
Speaker 1: a perfect system, obviously, it can sometimes be very off putting,

844
00:48:18,960 --> 00:48:21,520
Speaker 1: but it is a common one that's in use in

845
00:48:21,600 --> 00:48:25,879
Speaker 1: post process image stabilization. The way this is done practically

846
00:48:26,040 --> 00:48:28,160
Speaker 1: has changed over the years, like the way that people

847
00:48:28,160 --> 00:48:31,520
Speaker 1: have actually designed the algorithms so that they could make

848
00:48:31,600 --> 00:48:34,720
Speaker 1: this happen. That has changed. So a very early version

849
00:48:34,760 --> 00:48:39,200
Speaker 1: of this would have a camera essentially the image processing

850
00:48:39,200 --> 00:48:43,200
Speaker 1: software identify a point in the background that was clearly

851
00:48:44,000 --> 00:48:49,440
Speaker 1: defined and lit. So let's say that you've got uh

852
00:48:49,480 --> 00:48:52,400
Speaker 1: an image of a person standing in a field, and

853
00:48:52,440 --> 00:48:57,680
Speaker 1: there's a some wooded forest as opposed to the unwitted

854
00:48:57,719 --> 00:49:00,560
Speaker 1: forest in the background, and there's a particular tree that's

855
00:49:00,560 --> 00:49:05,160
Speaker 1: a nice sharp relief. Well, the image processing software might

856
00:49:05,480 --> 00:49:08,359
Speaker 1: focus on that tree and say, all right, we're going

857
00:49:08,440 --> 00:49:11,879
Speaker 1: to lock onto this section of the tree, and we

858
00:49:11,920 --> 00:49:14,120
Speaker 1: want that section of the tree to be in this

859
00:49:14,280 --> 00:49:19,640
Speaker 1: relative position in our frame of view for the duration

860
00:49:19,680 --> 00:49:22,920
Speaker 1: of this video. So you're hand holding the camera pointing

861
00:49:22,920 --> 00:49:25,719
Speaker 1: it at somebody who standing in the field chatting, and

862
00:49:25,800 --> 00:49:28,799
Speaker 1: because of the image processing, trying to keep that one

863
00:49:28,840 --> 00:49:30,920
Speaker 1: part of the tree in that one part of the frame.

864
00:49:31,280 --> 00:49:34,080
Speaker 1: It stabilizes the image even if there's a little bit

865
00:49:34,120 --> 00:49:36,640
Speaker 1: of jitter as you're holding it pointing it at the

866
00:49:36,680 --> 00:49:42,319
Speaker 1: person who's talking in the video. Uh, that works, okay,

867
00:49:42,480 --> 00:49:46,279
Speaker 1: it's it's a little primitive. You could actually do a

868
00:49:46,320 --> 00:49:49,160
Speaker 1: fairly nice static shot that way, but you can't move

869
00:49:49,200 --> 00:49:52,160
Speaker 1: the camera at all because if you start moving the

870
00:49:52,200 --> 00:49:55,880
Speaker 1: camera purposefully, i mean, beyond just the little jitters, then

871
00:49:55,960 --> 00:49:59,080
Speaker 1: that frame of reference is going to move dramatically and

872
00:49:59,160 --> 00:50:02,040
Speaker 1: the software can and handle that. It has to, you know,

873
00:50:02,160 --> 00:50:05,719
Speaker 1: keep an eye on a relatively stable shot. So this

874
00:50:05,760 --> 00:50:10,080
Speaker 1: is for static images static video. A slightly more advanced

875
00:50:10,160 --> 00:50:13,480
Speaker 1: version of that same approach would pick to reference points,

876
00:50:13,520 --> 00:50:15,440
Speaker 1: so one on either side of the frame, and then

877
00:50:15,440 --> 00:50:18,760
Speaker 1: it was essentially draw an imaginary line between those two points.

878
00:50:18,760 --> 00:50:22,000
Speaker 1: So let's say it's identified a tree on one side

879
00:50:22,400 --> 00:50:25,319
Speaker 1: and a bush on the other side, and said, all right,

880
00:50:25,760 --> 00:50:29,400
Speaker 1: based upon this setup, we want these two points to

881
00:50:29,520 --> 00:50:33,560
Speaker 1: remain the same general spots in your frame of view,

882
00:50:34,239 --> 00:50:37,239
Speaker 1: and the line that's between them will always be at

883
00:50:37,239 --> 00:50:41,160
Speaker 1: the same alignment no matter what. This would allow you

884
00:50:41,200 --> 00:50:44,040
Speaker 1: to actually have some rotation of the camera as well,

885
00:50:44,080 --> 00:50:49,520
Speaker 1: and the processor could could account for that and correct

886
00:50:49,560 --> 00:50:53,080
Speaker 1: for it. So, let's say you're holding a smartphone and

887
00:50:53,120 --> 00:50:56,759
Speaker 1: you're trying to take video of someone, and you're using

888
00:50:56,800 --> 00:51:00,520
Speaker 1: this particular version of image stabilization. If the outside of

889
00:51:00,560 --> 00:51:03,080
Speaker 1: your smartphone where to dip down a little bit and

890
00:51:03,120 --> 00:51:05,320
Speaker 1: the left side lifts up a little bit, thus you

891
00:51:05,320 --> 00:51:06,920
Speaker 1: would have a little bit of a tilt to the

892
00:51:07,000 --> 00:51:11,759
Speaker 1: video before you corrected it. Because of this particular form

893
00:51:11,800 --> 00:51:15,040
Speaker 1: of post process image stabilization, it would detect that change

894
00:51:15,280 --> 00:51:19,400
Speaker 1: relative to that imaginary line and correct for it. Uh,

895
00:51:19,520 --> 00:51:22,680
Speaker 1: depending on how dramatic your turn was. You might actually

896
00:51:22,680 --> 00:51:26,080
Speaker 1: notice this while looking at the playback, because unless you

897
00:51:26,160 --> 00:51:28,920
Speaker 1: crop further into the image, you might start seeing the

898
00:51:29,080 --> 00:51:33,280
Speaker 1: edges of where the cut off is for the video

899
00:51:33,640 --> 00:51:35,680
Speaker 1: popping up and it can be a little weird. You

900
00:51:35,719 --> 00:51:38,680
Speaker 1: may have even seen this in some videos online where

901
00:51:39,040 --> 00:51:41,480
Speaker 1: you start seeing an edge kind of creep into the

902
00:51:41,560 --> 00:51:44,360
Speaker 1: video a little bit. That's due to this post process

903
00:51:44,400 --> 00:51:48,080
Speaker 1: image stabilization, whether it was for jitter or for rotation,

904
00:51:48,600 --> 00:51:51,720
Speaker 1: and it can be very off putting. That's why most

905
00:51:52,320 --> 00:51:55,000
Speaker 1: people who use this, who are actual video editors, will

906
00:51:55,040 --> 00:51:57,920
Speaker 1: digitally punch in a little bit. They'll crop the image

907
00:51:58,160 --> 00:52:00,759
Speaker 1: so that they cut off those edges so that you

908
00:52:00,840 --> 00:52:03,319
Speaker 1: can't see that when it happens. Of course, the problem

909
00:52:03,360 --> 00:52:05,160
Speaker 1: with that is that you have to be at a

910
00:52:05,200 --> 00:52:08,319
Speaker 1: high enough resolution where when you digitally punch in, it's

911
00:52:08,360 --> 00:52:12,319
Speaker 1: not it's not a noticeable decrease in quality of the

912
00:52:12,360 --> 00:52:17,960
Speaker 1: actual image itself. For moving shots, visual image stabilization might

913
00:52:18,000 --> 00:52:21,720
Speaker 1: look at individual pixels and track how they change over time,

914
00:52:22,400 --> 00:52:25,440
Speaker 1: and they interpret that as motion. So if these pixels

915
00:52:25,440 --> 00:52:30,040
Speaker 1: are changing in ways where you see one pixel changing

916
00:52:30,360 --> 00:52:32,279
Speaker 1: and the pixel next to it has become the same

917
00:52:32,320 --> 00:52:35,120
Speaker 1: as the pixel that was to its left, and then

918
00:52:35,920 --> 00:52:38,880
Speaker 1: two pixels down it becomes what was two pixels to

919
00:52:38,880 --> 00:52:41,360
Speaker 1: the left, etcetera, etcetera. It can then start to interpret

920
00:52:41,360 --> 00:52:44,319
Speaker 1: this as motion and starts to work out where things

921
00:52:44,320 --> 00:52:46,960
Speaker 1: are moving, and then and editing, you can again crop

922
00:52:47,000 --> 00:52:49,920
Speaker 1: your video to remove those jolting edges and punch in

923
00:52:49,960 --> 00:52:54,600
Speaker 1: a bit. Some folks from Google and Georgia Tech actually

924
00:52:54,600 --> 00:52:57,600
Speaker 1: developed a new method of virtual image stabilization a few

925
00:52:57,680 --> 00:52:59,600
Speaker 1: years ago. So I guess it's not new now, but

926
00:52:59,680 --> 00:53:01,720
Speaker 1: it was a couple of years ago, and they published

927
00:53:01,719 --> 00:53:05,720
Speaker 1: their work in a paper titled Auto directed Video Stabilization

928
00:53:05,760 --> 00:53:09,719
Speaker 1: with Robust L one Optimal Camera Paths. The paper also

929
00:53:09,800 --> 00:53:12,839
Speaker 1: gets super technical. It is out there free for you

930
00:53:12,880 --> 00:53:16,200
Speaker 1: to read, so feel free to seek it out if

931
00:53:16,239 --> 00:53:18,960
Speaker 1: you are technically minded. Here's the bit that I think

932
00:53:19,040 --> 00:53:22,280
Speaker 1: is really important. They lay out that this post process

933
00:53:22,360 --> 00:53:27,319
Speaker 1: video stabilization requires three steps, and the first is to

934
00:53:27,480 --> 00:53:29,920
Speaker 1: estimate the original camera path, which is the one that's

935
00:53:29,960 --> 00:53:32,000
Speaker 1: all shaky and stuff. It's the one you want to fix.

936
00:53:32,440 --> 00:53:36,799
Speaker 1: The second step is to estimate a new smooth camera path,

937
00:53:37,239 --> 00:53:40,600
Speaker 1: and the third step is to synthesize a stabilized path

938
00:53:40,680 --> 00:53:44,600
Speaker 1: by following the estimated smooth path. This is not easy

939
00:53:44,640 --> 00:53:47,200
Speaker 1: to do, and there are lots of different ways to

940
00:53:47,280 --> 00:53:50,239
Speaker 1: try and do it, and we're always seeing developments in

941
00:53:50,239 --> 00:53:53,840
Speaker 1: this space. However, that being said, even as this post

942
00:53:53,880 --> 00:54:00,480
Speaker 1: process stabilization technique advances and it evolves, I think most

943
00:54:00,520 --> 00:54:04,719
Speaker 1: filmmakers would argue that the optical image stabilization and the

944
00:54:04,800 --> 00:54:09,200
Speaker 1: in camera stabilization systems are far superior that you can

945
00:54:09,600 --> 00:54:12,959
Speaker 1: get some image stabilization that's all right for basic use

946
00:54:13,280 --> 00:54:16,680
Speaker 1: in this post process approach, especially if you're going to

947
00:54:16,719 --> 00:54:18,960
Speaker 1: just do something like share something online, like it's a

948
00:54:19,320 --> 00:54:22,200
Speaker 1: online video or you know something on Facebook or Twitter,

949
00:54:22,600 --> 00:54:24,360
Speaker 1: it's not that big a deal to have this post

950
00:54:24,400 --> 00:54:28,960
Speaker 1: process image stabilization then. But if you want something that

951
00:54:29,120 --> 00:54:33,280
Speaker 1: is more of a professional level, hands down, the argument

952
00:54:33,280 --> 00:54:35,320
Speaker 1: I have seen is that you should go optical image

953
00:54:35,320 --> 00:54:39,239
Speaker 1: stabilization if you can, in camera stabilization if you if

954
00:54:39,280 --> 00:54:43,239
Speaker 1: optical is too cost prohibitive, But either of them are

955
00:54:43,360 --> 00:54:49,560
Speaker 1: far more reliable and effective than post process image stabilization,

956
00:54:49,600 --> 00:54:53,200
Speaker 1: and they are not going to create the weird artifacts

957
00:54:53,239 --> 00:54:57,719
Speaker 1: that you might see using a software based solution. So

958
00:54:57,800 --> 00:55:00,600
Speaker 1: it turns out that in this case, the pchanical one,

959
00:55:00,680 --> 00:55:04,319
Speaker 1: the electronic mechanical one, might be better than the software one.

960
00:55:04,400 --> 00:55:06,640
Speaker 1: That may not always be the case. We may eventually

961
00:55:06,680 --> 00:55:09,920
Speaker 1: get to a point where they are advanced enough algorithms

962
00:55:09,960 --> 00:55:14,840
Speaker 1: and advanced enough cameras where it becomes a non factor,

963
00:55:15,200 --> 00:55:19,640
Speaker 1: but we're not there yet. So generally speaking, image stabilization

964
00:55:19,719 --> 00:55:23,880
Speaker 1: is all about trying to take out the frailties of

965
00:55:23,920 --> 00:55:28,560
Speaker 1: being human, trying to remove that little element of humanity

966
00:55:28,680 --> 00:55:32,279
Speaker 1: where we get that imperfection that creeps into our art.

967
00:55:32,880 --> 00:55:35,439
Speaker 1: And for some of us. That means that we get

968
00:55:36,040 --> 00:55:38,960
Speaker 1: an image that we actually really wanted to convey to

969
00:55:39,080 --> 00:55:41,320
Speaker 1: our audience. For other people, they may say, well, that

970
00:55:41,400 --> 00:55:43,800
Speaker 1: kind of removes some of the humanity from the art itself,

971
00:55:44,040 --> 00:55:46,600
Speaker 1: and I'm not here to make either argument. I think

972
00:55:46,640 --> 00:55:49,799
Speaker 1: that there are some amazing uses of image stabilization that

973
00:55:49,880 --> 00:55:55,680
Speaker 1: create really compelling effects and are great for storytelling, and

974
00:55:55,719 --> 00:55:58,239
Speaker 1: I think there are elements where you don't want that,

975
00:55:58,320 --> 00:56:01,360
Speaker 1: where you want something more natural, a stick and shaky,

976
00:56:01,440 --> 00:56:05,759
Speaker 1: to convey a specific kind of mood or tone, And

977
00:56:05,800 --> 00:56:09,799
Speaker 1: really it comes down to the intent of what you

978
00:56:09,800 --> 00:56:13,759
Speaker 1: are doing and the theme that you're going for. I

979
00:56:13,800 --> 00:56:17,759
Speaker 1: don't think that either is necessarily inferior or superior to

980
00:56:17,800 --> 00:56:21,719
Speaker 1: the other, and I enjoy plenty of work that incorporates

981
00:56:22,360 --> 00:56:28,680
Speaker 1: both types of elements, hopefully purposefully. Sometimes you get these

982
00:56:28,719 --> 00:56:32,319
Speaker 1: these effects just by happenstance because people just didn't know

983
00:56:32,360 --> 00:56:35,279
Speaker 1: any better, and that can still be effective, but it's

984
00:56:35,280 --> 00:56:37,200
Speaker 1: a little less special to me than when people go

985
00:56:37,239 --> 00:56:40,960
Speaker 1: into it knowing what they're doing. Thank you so much

986
00:56:41,200 --> 00:56:42,799
Speaker 1: for joining me on this episode. It was a lot

987
00:56:42,840 --> 00:56:45,600
Speaker 1: of fun to look into a more technical aspect. I've

988
00:56:45,640 --> 00:56:48,920
Speaker 1: got another one coming up soon. That's also extremely technical,

989
00:56:48,920 --> 00:56:51,280
Speaker 1: and also was a suggestion that was left in the

990
00:56:51,320 --> 00:56:55,200
Speaker 1: twitch dot tv slash text stuff chat room. Remember, on

991
00:56:55,200 --> 00:56:59,239
Speaker 1: Wednesdays and Fridays, I do live stream my recordings of

992
00:56:59,560 --> 00:57:02,480
Speaker 1: text so just go to twitch dot tv slash tech Stuff.

993
00:57:02,520 --> 00:57:05,480
Speaker 1: You can see the schedule there. Also, if you have

994
00:57:05,520 --> 00:57:08,279
Speaker 1: any suggestions for future episodes of tech Stuff, you can

995
00:57:08,280 --> 00:57:11,480
Speaker 1: send me a message. My email address is tech Stuff

996
00:57:11,560 --> 00:57:14,040
Speaker 1: at how stuff works dot com. People ask me for

997
00:57:14,120 --> 00:57:16,680
Speaker 1: it all the time. I say it in every single

998
00:57:16,720 --> 00:57:21,560
Speaker 1: episode tex Stuff at how stuff works dot com. Or

999
00:57:21,680 --> 00:57:24,080
Speaker 1: you can drop me a line on Facebook or Twitter.

1000
00:57:24,160 --> 00:57:26,920
Speaker 1: The handle at both of those is text stuff hs W.

1001
00:57:27,640 --> 00:57:29,720
Speaker 1: That's it for me. I'll tell it to you again.

1002
00:57:30,440 --> 00:57:38,960
Speaker 1: Releas for more on this and thousands of other topics

1003
00:57:39,040 --> 00:57:50,320
Speaker 1: because at how stuff works dot com.