Question

I have written an application to compute the FFT and I would like to know a good way to visualize the spectrum in color.

Unfortunately my naiive attempts have not yielded good results. I would like to achieve something like this:

image

I am currently using this for colorization

float r = std::log( std::abs(source[u][v]) + 1 ) * 0.2f;
float g = std::log( std::max(std::abs(source[u][v]) - 10.0f, 0.0f) + 1 ) * 0.2f;
float b = std::log( std::max(std::abs(source[u][v]) - 100.0f, 0.0f) + 1 ) * 0.2f;

Answer 1

I propose the following approach. First, choose a maximum value (the one to represent “red”) and scale all values in the [0, 1] interval. Next, use this value for the hue and use a HSL to RGB transformation. The Python code would look like this:

import colorsys
from PIL import Image, ImageDraw

w = 400
h = 20
im = Image.new("RGB", (w, h))
draw = ImageDraw.Draw(im)

for x in range(0, 100):
    value = float(x)/100

    r, g, b = colorsys.hsv_to_rgb(1-value*3/4-0.33, 1, 1)
    color = int(r*255), int(g*255), int(b*255)
    draw.rectangle((x * 4, 0, (x + 1) * 4, h), color, color)

im.save('example.png' , 'png')

With the result: