Skyform - E.G.Ø www.egzero.org

SKYform
by E.G.Ø

Landscape Sampling
Pixelache
Helsinki 02.04.04

SKYform

This work borns after a series of reflections about landscapes and in particular about the line that divides earth and sky. The horizon, something that actually does not exist in nature, is graphed and mapped as a soundwave in the temporal domain. We are able now to "listen" to a landscape, we are able to sonify a line/concept not reacheable and infinite.

STEP 1
skyline extraction

First step is to consider a picture or a frame from a movie and extract the skyline. Let's consider a grayscale movie-frame in PAL format (720x576 px). We apply an edge detection algorithm to this image (this is a time-consuming operation).

STEP 2
mapping the data

The second step of the algorithm requires a data-mapping from image to sound. Considering a PAL frame, the software analizes each single column assigning a value of +1 if the edge-detected pixel is the first or a value -1 if the pixel is the last (number 576). So a single sample is:

val[col] = (576/2 - row)/(576/2)

where col is the considered column and row varies from 1 to 576. In the end we have all the values of the samples saved to a .wav file.

HOW DOES IT SOUND? 1
considering the samples as they are

If we just use the detected samples as they are (so that the values of the samples are the amplitude of a soundwave) we will ear a sound that is like a train of pulses at 25Hz (=25 fps) where each pulse is shaped as the skyline (pulsar synthesis). Since we have 720 samples each 1/25=0.04 seconds the sample frequency of the soundwave will be

720*25 = 18000 Hz

Aesthetical considerations

Pulsar synthesis is a particular kind of synthesis where you can hear the single pulses if the frequency is lower than 20 Hz, then you start to hear kind of low frequency modulated sound (the singles pulses are no more hearable). That's why mybe the sound as is it's not so musically interesting. One could work with a generated file of skylines and read that sample with different frequencies.

HOW DOES IT SOUND? 2
mapping samples into other domains

Different results could be obtained if we map the samples into other domains (e.g. into frequency domain) or if we use these samples to control other parameters of patches. For example it could be interesting to use the samples as a envelope of another sound.

	SKYform by E.G.Ø Landscape Sampling Pixelache Helsinki 02.04.04
	*SKYform* This work borns after a series of reflections about landscapes and in particular about the line that divides earth and sky. The horizon, something that actually does not exist in nature, is graphed and mapped as a soundwave in the temporal domain. We are able now to "listen" to a landscape, we are able to sonify a line/concept not reacheable and infinite.
	STEP 1 skyline extraction First step is to consider a picture or a frame from a movie and extract the skyline. Let's consider a grayscale movie-frame in PAL format (720x576 px). We apply an edge detection algorithm to this image (this is a time-consuming operation).
	STEP 2 mapping the data The second step of the algorithm requires a data-mapping from image to sound. Considering a PAL frame, the software analizes each single column assigning a value of +1 if the edge-detected pixel is the first or a value -1 if the pixel is the last (number 576). So a single sample is: *val[col] = (576/2 - row)/(576/2)* *where col is the considered column and row varies from 1 to 576. In the end we have all the values of the samples saved to a .wav file.*
	HOW DOES IT SOUND? 1 considering the samples as they are If we just use the detected samples as they are (so that the values of the samples are the amplitude of a soundwave) we will ear a sound that is like a train of pulses at 25Hz (=25 fps) where each pulse is shaped as the skyline (pulsar synthesis). Since we have 720 samples each 1/25=0.04 seconds the sample frequency of the soundwave will be **72025 = 18000 Hz* Aesthetical considerations** Pulsar synthesis is a particular kind of synthesis where you can hear the single pulses if the frequency is lower than 20 Hz, then you start to hear kind of low frequency modulated sound (the singles pulses are no more hearable). That's why mybe the sound as is it's not so musically interesting. One could work with a generated file of skylines and read that sample with different frequencies.
	HOW DOES IT SOUND? 2 mapping samples into other domains Different results could be obtained if we map the samples into other domains (e.g. into frequency domain) or if we use these samples to control other parameters of patches. For example it could be interesting to use the samples as a envelope of another sound.