I don't understand the basics of signal processing either. Any tips for good literature/videos about signal processing for image sampling like this?

> Looks like he made the classic log/linear error too.

It's converted from linear space to "gamma space" before writing the image, if that's what you mean:

> img = np.where(img <= 0.0031308, 12.92img, 1.055img(1.0/2.4) - 0.055)