Visual and non-visual properties of filters manipulating short-wavelength light

Abstract

Purpose: Optical filters and tints manipulating short-wavelength light (sometimes called 'blue-blocking' or 'blue-attenuating' filters) are used in the management of a range of ocular, retinal, neurological and psychiatric disorders. In many cases, the only available quantification of the optical effects of a given optical filter is the spectral transmittance, which specifies the amount of light transmitted as a function of wavelength.

Methods: We propose a novel physiologically relevant and retinally referenced framework for quantifying the visual and non-visual effects of these filters, incorporating the attenuation of luminance (luminous transmittance), the attenuation of melanopsin activation (melanopsin transmittance), the colour shift, and the reduction of the colour gamut (gamut reduction). Using these criteria, we examined a novel database of spectral transmittance functions of optical filters (n = 121) which were digitally extracted from a variety of sources.

Results: We find a large diversity in the alteration of visual and non-visual properties. The spectral transmittance properties of the examined filters vary widely, in terms of shapes and cut-off wavelengths. All filters show relatively more melanopsin attenuation than luminance attenuation (lower melanopsin transmittance than luminous transmittance). Across the data set, we find that melanopsin transmittance and luminous transmittance are correlated.

Conclusions: We suggest that future studies and examinations of the physiological effects of optical filters quantify the visual and non-visual effects of the filters beyond the spectral transmittance, which will eventually aid in developing a mechanistic understanding of how different filters affect physiology. We strongly discourage comparing the downstream effects of different filters on, e.g. sleep or circadian responses, without considering their effects on the retinal stimulus.

Keywords: blue-blocking filters; circadian; ipRGCs; optics; short-wavelength light; sleep.

Figure 1

Visual and non‐visual effects of filters. Luminous transmittance is given as the fraction of the luminance of daylight seen with filter (shown as the beige area) to without filter (shown as the white area). In this case, the luminance with the filter is only about a quarter (23%). Melanopsin transmittance follows the same calculation, except for the melanopsin photopigment. Because the cut‐off wavelength of the filter is outside of the spectral sensitivity of melanopsin, the attenuation is far larger compared to luminance (3%). Colour shift is the Euclidian distance between a specified white point seen with and without the filter in the uniform colour space used here (CIE 1976 u’₁₀v’₁₀ colour space based on the CIE 1964 10° observer.). The smaller the number, the smaller the colour shift and therefore, the closer the reproduction of the white point with the filter relative to without the filter. Gamut reduction is the reduction of the colour space which common surface reflectances inhabit, with and without the filter. In this case, the colour gamut with the filter is only around one fifth (20%) of the gamut without the filter. Here and elsewhere, the spectral power distribution was assumed to be of noon daylight (D65, daylight with a correlated colour temperature [CCT] of 6500K).

Figure 2

Exploring effects of cut‐off filters. Spectral transmittance of cut‐off filters is analytically modelled using the sigmoid function of the form $T(\mathrm{\lambda })=\frac{L}{1+{\mathrm{e}}^{-\mathrm{k}(\mathrm{\lambda }-{\mathrm{\lambda }}_0)}}$, where L is the upper asymptote, k is the slope, and λ₀ is the cut‐off wavelength. Left column: Varying cut‐off wavelengths (λ₀), with L = 0.9, k = 0.5. Middle column: Varying top asymptote L, with λ₀ = 550 nm, k = 0.5. Right column: Varying slope k, with λ₀ = 560 nm, L = 0.9. Small arrows point to the spectral transmittances at one extreme end of specific parameter choice, pointing out the corresponding retinal effect in the other panels as well.

Figure 3

Visual and non‐visual properties of spectral filters. Columns correspond to the three main filter categories we identified (Figure 2). Row 1: Spectral transmittances of filters. Row 2: Luminous transmittance vs melanopsin transmittance. Dashed line indicates equal reduction of luminance and melanopsin, as would be the case with a spectrally uniform neutral density (ND) filter. Row 3: Chromaticity diagram. The red cross indicates chromaticity of 6500K daylight (D65) and white squares indicate chromaticities of D65 seen through the respective filters. Row 4: Colour shift vs gamut factor. See Introduction and Figure 1 for explanation.

Figure 4

Pupil size effects of neutral‐density filters. Top panel: Predicted pupil size for a 32‐year old observer viewing a 150° field at varying luminances with both eyes. Bottom panel: Predicted retinal illuminance (luminance × pupil area) when viewed either through the natural pupil (dashed lines indicate maximum retinal illuminance given maximum difference in pupil size), or through spectrally uniform filters of varying transmittance (ND1.0, ND2.0, ND3.0).