How Video Works? - Speaker Deck

Slide 1

Slide 1 text

How Video Works?  Flavio Ribeiro, 2017

Slide 2

Slide 2 text

Flavio Ribeiro

Slide 3

Slide 3 text

Flavio Ribeiro

Slide 4

Slide 4 text

Flavio Ribeiro { , } / flavioribeiro http://flavioribeiro.com

Slide 5

Slide 5 text

No content

Slide 6

Slide 6 text

No content

Slide 7

Slide 7 text

Leandro Moreira / leandromoreira

Slide 8

Slide 8 text

Leandro Moreira / leandromoreira

Slide 9

Slide 9 text

Leandro Moreira / leandromoreira

Slide 10

Slide 10 text

what is video?

Slide 11

Slide 11 text

video is a composition of images called frames

Slide 12

Slide 12 text

video is a composition of images called frames

Slide 13

Slide 13 text

what is a frame?

Slide 14

Slide 14 text

Composition of Pixels composition   of pixels

Slide 15

Slide 15 text

but, what  is a pixel?

Slide 16

Slide 16 text

No content

Slide 17

Slide 17 text

this is a black pixel!

Slide 18

Slide 18 text

this is a black pixel!

Slide 19

Slide 19 text

this is a black pixel! white pixel here

Slide 20

Slide 20 text

how pixels are represented?

Slide 21

Slide 21 text

how pixels are represented? BITS

Slide 22

Slide 22 text

how pixels are represented? 1 bit (1 or 0)

Slide 23

Slide 23 text

how pixels are represented? 1 bit (1 or 0) 2 bits (11, 10, 01 or 00)

Slide 24

Slide 24 text

how pixels are represented? 1 bit (1 or 0) 2 bits (11, 10, 01 or 00) 8 bits (you got that, right?)

Slide 25

Slide 25 text

colors

Slide 26

Slide 26 text

colors

Slide 27

Slide 27 text

No content

Slide 28

Slide 28 text

pixels in displays TV CRT PC CRT LCD

Slide 29

Slide 29 text

pixels in displays TV CRT PC CRT LCD

Slide 30

Slide 30 text

let’s do math

Slide 31

Slide 31 text

grayscale image width height 4 x 4 x 8 bits = 128 bits = 16 Bytes

Slide 32

Slide 32 text

width height colored image

Slide 33

Slide 33 text

width height color colored image

Slide 34

Slide 34 text

colored image 3 x 4 x 4 x 8 bits = 384 bits = 48 Bytes

Slide 35

Slide 35 text

3 x 1920 x 1080 x 8 bits = 49,766,400 bits ≈ 6.2 MB

Slide 36

Slide 36 text

do you remember that…

Slide 37

Slide 37 text

video is a composition of images called frames do you remember that…

Slide 38

Slide 38 text

our eyes can detect motion at 24 frames per second

Slide 39

Slide 39 text

No content

Slide 40

Slide 40 text

No content

Slide 41

Slide 41 text

3 x 1920 x 1080 x 8 bits x 24 frames ≈ 149 MB

Slide 42

Slide 42 text

3 x 1920 x 1080 x 8 bits x 24 frames x 60 minutes ≈ .5 TB

Slide 43

Slide 43 text

It would take… 04 days and 15 hours to download it in a 10 Mbps connection 11 hours to download it in a 100 Mbps connection

Slide 44

Slide 44 text

what can be done?

Slide 45

Slide 45 text

No content

Slide 46

Slide 46 text

exploit our vision

Slide 47

Slide 47 text

exploit our vision reduce redundancy in space

Slide 48

Slide 48 text

exploit our vision reduce redundancy in space reduce redundancy  over time

Slide 49

Slide 49 text

exploit our vision

Slide 50

Slide 50 text

No content

Slide 51

Slide 51 text

the eye contains about 120M rod cells and 6M cone cells.

Slide 52

Slide 52 text

we’re better noticing traces than colors

Slide 53

Slide 53 text

we’re better noticing traces than colors

Slide 54

Slide 54 text

an alternative to RGB

Slide 55

Slide 55 text

an alternative to RGB

Slide 56

Slide 56 text

Y (luma) U (chroma blue) V (chroma red) an alternative to RGB

Slide 57

Slide 57 text

Y (luma) U (chroma blue) V (chroma red) an alternative to RGB

Slide 58

Slide 58 text

No content

Slide 59

Slide 59 text

12 bits per pixel 24 bits per pixel

Slide 60

Slide 60 text

reduce redundancy in space

Slide 61

Slide 61 text

lots of similarities

Slide 62

Slide 62 text

lots of similarities

Slide 63

Slide 63 text

No content

Slide 64

Slide 64 text

No content

Slide 65

Slide 65 text

repeat x times repeat y times

Slide 66

Slide 66 text

No content

Slide 67

Slide 67 text

patterns / vectors

Slide 68

Slide 68 text

correlations in time

Slide 69

Slide 69 text

correlations in time

Slide 70

Slide 70 text

00:22 00:23

Slide 71

Slide 71 text

00:22 00:23

Slide 72

Slide 72 text

00:22 00:23

Slide 73

Slide 73 text

00:22 00:23

Slide 74

Slide 74 text

frame 0 frame 1 frame 2 frame 3 103 kb 103 kb 103 kb 103 kb

Slide 75

Slide 75 text

#qacodec FRAME DIFFERENCE 103 kb 103 kb

Slide 76

Slide 76 text

#qacodec FRAME DIFFERENCE COST 4 kb

Slide 77

Slide 77 text

frame 0 frame 1 diff + reference 4 kb 103 kb

Slide 78

Slide 78 text

exploit our vision reduce redundancy reduce redundancy  over time we just described how a codec works

Slide 79

Slide 79 text

exploit our vision reduce redundancy reduce redundancy  over time codec

Slide 80

Slide 80 text

exploit our vision reduce redundancy reduce redundancy  over time is the short for COmpressor / DECompressor

Slide 81

Slide 81 text

beyond codecs

Slide 82

Slide 82 text

containers

Slide 83

Slide 83 text

No content

Slide 84

Slide 84 text

this is a container

Slide 85

Slide 85 text

this is a container video here

Slide 86

Slide 86 text

this is a container video here

Slide 87

Slide 87 text

this is a container video here audio here

Slide 88

Slide 88 text

this is a container video here audio here

Slide 89

Slide 89 text

this is a container video here audio here stuff

Slide 90

Slide 90 text

MP4 H.264 MP3 Subtitles, metadata, etc

Slide 91

Slide 91 text

WebM VP8 Vorbis Subtitles, metadata, etc

Slide 92

Slide 92 text

bitrate

Slide 93

Slide 93 text

bitrate

Slide 94

Slide 94 text

bitrate: amount of bits used to represent a given media

Slide 95

Slide 95 text

bitrate: amount of bits used to represent a given media

Slide 96

Slide 96 text

bitrate: amount of bits used to represent a given media

Slide 97

Slide 97 text

bitrate: amount of bits used to represent a given media

Slide 98

Slide 98 text

deliver  video on internet

Slide 99

Slide 99 text

progressive download

Slide 100

Slide 100 text

adaptive http streaming

Slide 101

Slide 101 text

No content

Slide 102

Slide 102 text

No content

Slide 103

Slide 103 text

recap

Slide 104

Slide 104 text

recap

Slide 105

Slide 105 text

No content

Slide 106

Slide 106 text

video

Slide 107

Slide 107 text

video images (frames)

Slide 108

Slide 108 text

video images (frames) pixels

Slide 109

Slide 109 text

video images (frames) pixels bits

Slide 110

Slide 110 text

video images (frames) pixels colors bits

Slide 111

Slide 111 text

video images (frames) pixels colors RGB bits

Slide 112

Slide 112 text

video images (frames) pixels colors RGB 24 frames bits

Slide 113

Slide 113 text

video images (frames) pixels colors RGB 24 frames bits lots of information

Slide 114

Slide 114 text

video images (frames) pixels colors RGB 24 frames bits lots of information compression

Slide 115

Slide 115 text

video images (frames) pixels colors RGB 24 frames bits lots of information compression exploit the eye

Slide 116

Slide 116 text

video images (frames) pixels colors RGB 24 frames bits lots of information compression exploit the eye reduce redundancy

Slide 117

Slide 117 text

video images (frames) pixels colors RGB 24 frames bits lots of information compression exploit the eye reduce redundancy correlations in time

Slide 118

Slide 118 text

video images (frames) pixels colors RGB 24 frames bits lots of information compression exploit the eye reduce redundancy correlations in time codec

Slide 119

Slide 119 text

video images (frames) pixels colors RGB 24 frames bits lots of information compression exploit the eye reduce redundancy correlations in time codec codec != container

Slide 120

Slide 120 text

video images (frames) pixels colors RGB 24 frames bits lots of information compression exploit the eye reduce redundancy correlations in time codec codec != container bitrate

Slide 121

Slide 121 text

video images (frames) pixels colors RGB 24 frames bits lots of information compression exploit the eye reduce redundancy correlations in time codec codec != container bitrate progressive download

Slide 122

Slide 122 text

video images (frames) pixels colors RGB 24 frames bits lots of information compression exploit the eye reduce redundancy correlations in time codec codec != container bitrate progressive download adaptive bitrate