Phase shift calculations and "% as bright as daylight"

  • I've trying to make sense of the calculations behind the phase shifting effect and the concept "% as bright as daylight". Unfortunately the "what is this" button has not make me brighter.
    Could you please elaborate on this? What is the timing assume of the light exposure for the calculation of the phase shift? What does it mean that for instance an m-lux of 60.1 (ipad 4 at default settings) has a 43% as bright as daylight? Are numbers normalized to a certain threshold?
    Thanks in advance!

  • f.lux team

    This is based largely on exposures of 4-6 hours (not short ones) after spending the day in the office.

    So you can think of phase shift here as the difference to your internal clock between seeing this light all night vs spending time in the dark.

    We show the percentage to compare the physiological effect vs a saturating one like daylight.

  • Dear Herf, thank you so much for your quick response.
    So to confirm, what it shows is then a phase delay since it corresponds to exposure to that light before bed, and although the light pulse might be long allowing for some advances, it might still not compensate for the delaying effect of bed before bed time. So this would be exposure to an iphone before bed time for several hours and no filter that gives according to your table 70 lux at aye level. And this compared to saturating bright light it already achieves 43% of that bright light effect, correct?

    You mentioned that the phase shift is based on long exposures after spending the day in the office. I am not aware of many studies that have investigated phase shifting effects of light after a day of exposure to typical office lighting. Are these f.lux own calculations based on studies on phase shifting effects of light from dim to a light condition?

    And one more thing on this output page. It shows me that I have 60.1 m-lux and that it measures 7541K but that this feels like 5334K, I am not sure I understand this. Why would it feel warmer than what it actually is? It also shows 0.98 m-lux/lux but if the table is correct and at eye level I had 75 lux I do not come to the 60.1 m-lux reported for the iphone?

    Sorry about all these questions, but I found the website very fascinating and tons of info and I want to be sure I understand it all in order to discuss this site with colleagues.

    Thanks in advance. Kindly, Marina.

  • f.lux team

    Oh, you're at Groningen! I should try to publish all of this so people can replicate it.

    For the main phase shift calculation, you could think of Zeitzer 2000 (90 lux background, 6.7hr stimulus), though we have made adjustments based on subsequent studies and spectrum. Zeitzer's fits also indicate a greater than 24h period (i.e. "zero light" maps to a phase shift), so we are subtracting this daily shift and reporting "phase shift relative to darkness".

    Then our "% daylight" metric compares this phase shift amount to the result from a saturating pulse ("daylight brightness").

    In real life, a history of bright daylight exposure might make the body less sensitive, and shorter duration exposures would likely require more light in order for the system to saturate. But for office lighting and screen lighting, and especially for younger populations (<18 years), this makes a good starting point.

    With the "feels like xxK" metric I have tried to explain the spectral effects of light in terms of the color temperature for a blackbody. So it is showing the equivalent blackbody in terms of M/P (melanopic/photopic) ratio.

    This is pretty easy to replicate: If you make a table of all blackbodies (computing M/P ratio for each), then with M/P ratio for a new light source, you look it up in the table. This might explain how the light source affects you. Since people have started to think of "color temperature" as a substitute for melanopic efficacy, we like to point out that (a) this is not a great idea overall and (b) to give an intuitive treatment of the number. Hopefully this metric helps with both.

    We're aware that "melanopic" response not the whole story for non-visual measurement. But in dealing with RGB screens, it is one of the more difficult metrics to minimize, because it extends far into the green. So if the only spectral tool you have is to reduce color temperature, we felt in the beginning that it might serve as a good basic guideline.

    I am discouraging people from over-optimizing their spectral engineering with "only" melanopsin in mind, but the site doesn't explain that idea very well. We have some references to cone contributions and some other metrics.

  • f.lux team

    Regarding the iPhone data, in the "Color" tab you can see 61.66 photopic lux.

    And then M/P ratio should be 0.98.

    For the melanopic calculation, we are using the "equal energy" calibration from Lucas et al 2014. Since Illuminant E has an equivalent CCT of about 5455K, this checks out.

Log in to reply

Looks like your connection to f.lux forum was lost, please wait while we try to reconnect.