Saul's Log - September 1, 2001
The other day I thought of a great analogy for what I do
when I compose music for the OMF soundtrack:
Imagine that my job is to paint a picture of a landscape or a person's face, and to make it as realistic, appealing, and rich as possible. Now, imagine I have to paint that picture using nothing but squares - no lines, arcs, smudging or smearing - just squares. Also, imagine that each little jar of paint costs me money. I'm only allowed to spend $10 on paint. Some of the simpler colours are pretty cheap, say 10 cents a jar. However, the nicest, richest colours cost up to a dollar, and for some reason, white and black each cost 50 cents. Also, some colours aren't available in stores, so I need to create them in my lab, which can take hours or even days. So, I have to work with a limited palette of colours.
"Okay", you might think to yourself, "when you need a curved line, just paint a huge number of small squares and it'll create the illusion of a curve. And when you need a colour you don't have, alternate two other colours and it'll create the illusion of that colour". One problem: each square of paint weighs 1 gram, and my painting can't weigh more than 1 kilogram or it'll pull its hook out of the wall. So now what?
Now I have to compromise. When I paint a beautiful face, I might spend two dollars on paint for the blue eyes, and another 3 dollars on paints to mix for the skin tone. This leaves me only 5 dollars for the hair, ears, lips, clothing, and background, so they'll have to be less rich. Say I use 300 squares making the face round, and another 100 squares making the smile just perfect. Now I have 600 left for the rest of the picture, so it'll have to be less detailed. In the end, the most important features of the face will look great, but on close examination the other parts may seem a bit off.
Now, to explain this analogy...
The paint represents the audio samples I use when I compose a song. The money represents RAM - each sample takes up a certain amount of RAM, and better-sounding samples require more RAM. The squares represent the fact that the samples must be played back as they are - unlike a live instrument, I can't buzz my lips differently or pluck the stringer harder. The canvas is the CPU, and the weight of the paint is how much of the CPU the music is using. The painting has to "share the wall" with a lot of other things, so it's limited to a maximum weight. Finally, when I paint a lot of squares to approximate a curve, that represents layering or doubling samples to create the illusion of reverberation, atmosphere, and depth in the music.
So, using our analogy, what's it like to do a live recording to an MP3 file?
Paint costs nothing, and weighs nothing as well. You can buy any colour imaginable and use it as much as you like. You can paint using any technique you want, mix colours as much as you like, and smear or smudge if you want to.
In other words... you can paint anything you can imagine.
By this point you're probably thinking, "why didn't these idiots use MP3's?"... The tradeoff, of course, is that our painting is *alive*. The canvas can fold and re-arrange itself into different shapes, and even hide or show different parts of the picture to change its meaning. In fact, the canvas can fold itself into an uncountable number of shapes. The number is literally about 650 digits long. In other words: it's a different painting every time you look.
So, instead of a single picture with great detail and depth, we have a great number of pictures with limited depth. Is the tradeoff worth it?
Well, it's debatable. One thing I do know: right now, we're the only gallery in town with folding paintings.
Imagine that my job is to paint a picture of a landscape or a person's face, and to make it as realistic, appealing, and rich as possible. Now, imagine I have to paint that picture using nothing but squares - no lines, arcs, smudging or smearing - just squares. Also, imagine that each little jar of paint costs me money. I'm only allowed to spend $10 on paint. Some of the simpler colours are pretty cheap, say 10 cents a jar. However, the nicest, richest colours cost up to a dollar, and for some reason, white and black each cost 50 cents. Also, some colours aren't available in stores, so I need to create them in my lab, which can take hours or even days. So, I have to work with a limited palette of colours.
"Okay", you might think to yourself, "when you need a curved line, just paint a huge number of small squares and it'll create the illusion of a curve. And when you need a colour you don't have, alternate two other colours and it'll create the illusion of that colour". One problem: each square of paint weighs 1 gram, and my painting can't weigh more than 1 kilogram or it'll pull its hook out of the wall. So now what?
Now I have to compromise. When I paint a beautiful face, I might spend two dollars on paint for the blue eyes, and another 3 dollars on paints to mix for the skin tone. This leaves me only 5 dollars for the hair, ears, lips, clothing, and background, so they'll have to be less rich. Say I use 300 squares making the face round, and another 100 squares making the smile just perfect. Now I have 600 left for the rest of the picture, so it'll have to be less detailed. In the end, the most important features of the face will look great, but on close examination the other parts may seem a bit off.
Now, to explain this analogy...
The paint represents the audio samples I use when I compose a song. The money represents RAM - each sample takes up a certain amount of RAM, and better-sounding samples require more RAM. The squares represent the fact that the samples must be played back as they are - unlike a live instrument, I can't buzz my lips differently or pluck the stringer harder. The canvas is the CPU, and the weight of the paint is how much of the CPU the music is using. The painting has to "share the wall" with a lot of other things, so it's limited to a maximum weight. Finally, when I paint a lot of squares to approximate a curve, that represents layering or doubling samples to create the illusion of reverberation, atmosphere, and depth in the music.
So, using our analogy, what's it like to do a live recording to an MP3 file?
Paint costs nothing, and weighs nothing as well. You can buy any colour imaginable and use it as much as you like. You can paint using any technique you want, mix colours as much as you like, and smear or smudge if you want to.
In other words... you can paint anything you can imagine.
By this point you're probably thinking, "why didn't these idiots use MP3's?"... The tradeoff, of course, is that our painting is *alive*. The canvas can fold and re-arrange itself into different shapes, and even hide or show different parts of the picture to change its meaning. In fact, the canvas can fold itself into an uncountable number of shapes. The number is literally about 650 digits long. In other words: it's a different painting every time you look.
So, instead of a single picture with great detail and depth, we have a great number of pictures with limited depth. Is the tradeoff worth it?
Well, it's debatable. One thing I do know: right now, we're the only gallery in town with folding paintings.