week 09

video

quantizing video colour: YUV

Y = luminance/brightness (changes more easily seen by human eyes)

UV = chrominance (colour/hue), not needed for black and white TV

4:1:1 method:

same brightness (4), but taking the average colour

* reduces storage

old TVs

interlaced display

* our eyes see phosphor dots on the screen
* an electron beam (gun) activates the dots
* the gun scans through the dots horizontally
* each pass = a field

2000s TVs

* 4k TVs have vertical scan lines
* standard definition resolution was 480 or 576
* now companies are beginning to make 8k TVs

2010 small devices

* Apple says small devices should be held 10-12 inches away
* screen resolution should be at least 300ppi to look crisp

retina display

* more pixels = crisper images
* 326ppi (300 is the limit for human retina)
* need to consider pixel density, viewing distance, and display size

editing before digital video

* had to copy from one tape to another
* had to load up tapes on a machine copy
* loss of quality after each copy made

editing video now

* can easily move clips around
* no loss of quality
* slight compression occurs in the camcorder when the video is captured

pre 1967

camcorders were not portable

1967

Sony came up with black and white camcorder (it was big)

1971

first cassettes

1982

combine videos, sound, recording and playback (camera recorder)

2001

most camcorders record at 5fps

2019

iPhoneX records 4k at 60fps

things to think about before exporting video

where?

* on the web compression is important
* on CD playback speed is an issue
* DVD video must be in mpeg2 format

who?

* will they be on different platforms
* how old will their computer be

why?

* videos have a lot of stuff (colour, sound, motion, etc…)
* 1 second video could be 133MB without compression

data and bit rate

* amount of video processed per second
* average bitrate = file size/length in seconds of video
* normally measured in bit (Mb = megabit and MB = megabytes)

compression: lower frame size

change the frame size for each frame in the video

compression: lower frame rate

* hold the frame on the page for longer
* depends on the type of video (not good for high action)

compression: pick a codec that does higher compression

codec: coder/decoder

* compression decompression software
* compresses a video or audio as it is created
* decompresses the video or audio as it is displayed to the user
* YouTube uses H.264
* often relates to file format
* sometimes can’t get a video if you don’t have to right codec

compression: lower picture quality

can be done with a codec

compression: lower the colour depth

* not popular because video looks best in 24 bit colour
* some compressors won’t compress colour

compression: play with the audio

not super helpful

spatial compression

* compresses each frame individually
* same are JPG compression

temporal compression (e.g., H.264)

* only saves info on keyframes
* all other frames just save the differences from the previous keyframe
* good when the difference between current frame and keyframe is small

lossy vs lossless compression (depends on codec)

lossless:

* looks for large blocks that are the saem and do run length encoding (RLE)

lossy:

* lower video quality but get better file size and bitrate

html5 (2008)

* a new standard for the html in web pages
* includes a way to watch video on a website that does not require a plug-in
* supports MP4, WebM, and Ogg (not Flash)
* goal was good, royalty free compression that handles hardware issues

Ogg, WebM, and MP4

* originally wanted to use codec .ogg but Apple didn’t want it
* WebM another royalty free file format used by Google and YouTube
* Apple and Microsoft use H.264 which creates .mp4 files

how to add video in html5

*

soft vs hard subtitles

hard: embedded in the video and cannot be turned on/off

soft: stored in a separate file and can be turned on/off (.vtt files)

downloading video on the web

* used to be the only option for viewing video
* progressive download: start watching the video as soon as it has enough bits to stay ahead of the download
* data is sent to be permanently stored on the machine

streaming video on the web

true streaming

* file is never permanently on the hard drive and plays as soon as it gets enough packets to stay ahead
* not as high quality video
* don’t have to wait for download
* uses buffering: download speed depends on bandwidth
* unicast: each user gets their own stream of video
* multicast: sends the same stream but people can’t pause

adaptive streaming

* when you upload a 4k video to YouTube, it makes several quality versions of it
* if the viewer picks “auto” it will select the best quality video depending on the current bandwidth
* people find buffer much more annoying than low quality

YouTube and adaptive streaming

* accepts several video formats
* uses many codecs, but mostly H.264
* convert every video into .mp4
* always compresses your video so it’s bets to always upload your highest quality video (but it’ll take a while)

bitrate and quality

* going from 1080p to 4k will not always improve quality
* the video with the least compression will have the best quality
* over 75% of videos are viewed on phones
* smartphones are capable of recording in 4k but cannot display in 4k (not enough pixels)