Computers store all their information as numbers. They are only capable of reading and writing numbers. Therefore, when we convert pictures and sound to digital information we are really converting of colours or notes (sound at differing pitch) to numbers.
When we make these conversions (colours or sounds to numbers) we have two choices to make regarding the quality of the image or sound we want to record.
The first choice is, how many stages should we provide to record 'all' of the colours in an image or 'all' of the notes in a song. For example; two colours only gives us black and white, and 88 notes gives us all of the notes in a piano, but what about all the other colours or sounds? This aspect of digital media technology is called bit depth.
The second choice we have is, how many times (per second) should we record each separate sound? Technically, we call recording sampling: How many times (per second) should we sample the sound? Natually, this does not apply to still images as we only need one sample to record the total image, but it applies to music and video. This aspect of media technology is called bit rate.
In the end, the final quality of a stored image will be a result of the bit depth used and for recorded music, a result of both bit depth and bit rate used in the recording process.
Bit depth means the quality of the colour or sound in each sample. In images bit depth means the number of different colours that can be recorded.
First we will examine images. Later, we will examine how in sound, bit depth means the number of values of sound that can be recorded.
When you look at the images above you can see how after the first image, the bit depth has been increased to allow for more colours.
Bit depth (or colour depth) means the amount of bits (computer spaces) provided for the storage of the colour of each pixel. Some examples of different bit depth are:
- 1 bit = black and white,
- 8 bit black and white = greyscale,
- 8 bit colour = limited colours,
- 8 bit colour = 256 colours palleted = colours stored in a 'Colour Lookup Table' (CLUT) like we get in a GIF,
- 24 bit = 8 bits per colour (channel) = "true colour".
When exporting an audio or video file the computer needs to make lots of instances of sound and/or vision that can be played back in sequence. The individual moments when the computer ‘looks at’ or ‘listens to’ the sound or vision are called samples. We can control the number of samples per second.
Bitrate means the number of samples per second. The typical bit rate for a MP3 is 44.1 kHz. This means the computer samples the sound at just over 44 thousand times a second!
Let's look at some simple examples to see how this works. In these examples I am imagining that I am recording sound. You notice that instead of just a number of possiblilities as we had for an image, we now have a graph. On the left is the number of possiblie piches of sound (the bit depth) and going from left to right is time divided into a number of samples. (The bit rate). The sound will be recorded only on each pitch and at each sample. The computer stores these positions as numbers. There are no in-betweens.
very low bit rate
In the next example I have kept the bit depth the same but I have increased the bit rate. Now that we have twice as many samples, the tune might sound a bit better. The computer stores twice as many numbers for each second of music.
increased bit rate
Now to make the sound even better I am going to increase the bit depth. Note the column on the left shows double the number of pitch steps for which I can record the sound. The computer uses twice as much space to store twice as many numbers in this one.
increased Bit depth and bit rate
When exporting a video it is possible to specify constant or variable bitrate sampling. Setting a constant bitrate will
make the encoder sample the vision at the same rate regardless of whether it is necessary. This can be fast to export but
wastefull of file size because some still pieces of vision may not need so many samples as fast action does. Variable bitrate
sampling can be more efficient as it samples only when it needs to, but it takes longer to export as the encoder needs to make multiple
passes to build the file. It must 'look' at the whole movie first to note where the action is fast or slow (the first 'pass'), then make a second pass to actually record it using variable bitrate sampling. You will see multiple or dual pass variable bitrate encoding as an export option where this is available.