-
-
Notifications
You must be signed in to change notification settings - Fork 287
Incorrect log function in the pupil diameter computation of "Barten (1999)" contrast sensitivity function. #1022
Description
colour/colour/contrast/barten1999.py
Line 131 in 491b6ef
| return as_float(5 - 3 * np.tanh(0.4 * np.log(L * X_0 * Y_0 / 40**2))) |
The np.log function should be replaced by np.log10
See Beau's work at JOV that traces the log function back to the work by Crawford (1936)
With np.log10, the Barten (flat) curve in the example should evaluate instead to:
[0.021876435681538915, 0.014184813713158736, 0.00952448986054018, 0.006680496306613172, 0.004924641286022344, 0.0038228778810919495, 0.003118856932347983, 0.002662754642667843, 0.002367458781369199, 0.0021814667512135366]
With np.log10, the Barten (flat) and Barten (ramp) curves, evaluated at X0=60, L = [0.001, 0.01, 0.1, 1.0, 10.0, 100.0] are:
Barten (flat):
[0.06299067646573944, 0.021876435681538915, 0.008678146175280605, 0.004311183591087381, 0.0027590747863252907, 0.0021814667512135366]
Barten (ramp):
[0.12598135293147888, 0.04375287136307783, 0.01735629235056121, 0.008622367182174762, 0.0055181495726505814, 0.004362933502427073]
which fit closer to Figure 33 of ITU BT.2246 and also Figure 4 of the following paper:
S. Miller, M. Nezamabadi, S. Daly, "Perceptual Signal Coding for More Efficient Usage of Bit Codes," SMPTE Meeting Presentation, 2012