More progress made today!
So I’ve been playing around with values from a forward FFT of the music data.
Here is my first attempt with calculating the colors from the FFT. Red is dependent on the amplitude of the first frequency band, green on the second, and blue on the third. The first three bands are the lowest frequencies, and thus generally have the greatest amplitudes. Although the colors were nice, everything was leaning heavy on the bluer side.
I realized that this was because the values for the RG channels were completely getting out of bounds. I adjusted the values with a scalar constant. The results were much much too dark and almost eerie:
I ended up fixing this by inverting the color of the stroke. Much better!
Alright, now at this point, the line opacity was at a constant 0.5. I decided to play with this value as well, making it proportional to the average amplitude of the frequency in the spectrum. I am very pleased with how this is turning out. I feel like the images actually capture their respective song quite well. I included two other output images of Ella Fitzgerald’s “Lullaby of Birdland” and Snoop Dogg’s “Gin and Juice” for additional contrast. I am excited that the differences in the song can also be seen visually through my system.
Thinking about differences actually led me to think about similarities. I decided to test this out by pulling two more songs off of Kishi Bashi’s album “151A” to see how they would compare. I have included the “Manchester” image again for reference.