Light too dark to fast after sunset in winter about hass-circadian_lighting

[claytonjn]

I'm familiar with f.lux; like many others I'm sure that was my first real exposure to the effect of color-temperature. F.lux was actually the first point of reference I used when trying to decide on a behavior for the component. The main question I have in regards to how f.lux works is whether the "stepped" graph is intentional for optimal effect, or simply a result of computer screens not being able to achieve the same incremental transitions that bulbs can. Behind the graph is a sinusoidal curve which is what CL currently follows, and it seems to me that they are trying to follow that curve but can only do so with a few tiers.

I haven't looked at their research page in a while so I will go through that again when I have the time, but again, for the most part it seems to be just research "proving" that lighting effects wellness, without any actual data on what the optimal behavior truly is (other than that blue light at night is bad, which is both vague and has been challenged recently https://www.manchester.ac.uk/discover/news/researchers-discover-when-its-good-to-get-the-blues/)

Another thing to keep in mind is that light emanating from a screen is very different than room lighting. Yes, technically all light is reflected from a surface but light from a screen reflects off one specific surface directly into your eyes whereas room lighting reflects off all surfaces in the room, greatly diffusing and changing the wavelengths.

This is all good discussion and I hope that some good conclusions can come out of it!

[Joshfindit]

I'm familiar with f.lux; like many others I'm sure that was my first real exposure to the effect of color-temperature. F.lux was actually the first point of reference I used when trying to decide on a behavior for the component. The main question I have in regards to how f.lux works is whether the "stepped" graph is intentional for optimal effect, or simply a result of computer screens not being able to achieve the same incremental transitions that bulbs can. Behind the graph is a sinusoidal curve which is what CL currently follows, and it seems to me that they are trying to follow that curve but can only do so with a few tiers.

What f.lux is doing on the computer (and I suspect on mobile devices as well) is applying a colour profile to the monitor (and, I suspect, updating that profile by way of the always-running app. Though I could find no direct evidence of that currently, it's how it used to work based on lra/mackup#464 ). When it goes through the 'long' transition of 30min I don't notice any stuttering or difficulties following a curve, but as you mention screen colour temps work a lot differently than lights.

I suspect the stepped curve they have implemented is more about not overwhelming the end users with a complicated interface when typically what they expect initially is "if I download this I will sleep better"

I haven't looked at their research page in a while so I will go through that again when I have the time, but again, for the most part it seems to be just research "proving" that lighting effects wellness, without any actual data on what the optimal behavior truly is (other than that blue light at night is bad, which is both vague and has been challenged recently https://www.manchester.ac.uk/discover/news/researchers-discover-when-its-good-to-get-the-blues/)

Not to jump in here and pretend to be the know-it-all expert, but in my personal opinion as someone who's been paying attention to f.lux and 'screen' effect on both sleep and alertness since at least the 2010, I'd say that the recent study (I have not read it yet, but after some digging I found it) on blue light may have a point , but only in isolation. For my own purposes I find the most benefit in taking artificial blue light out of the equation so that it doesn't interfere with the actual sunlight (no blue light after sundown, for example)

While I see the point they are making that dusk and predawn have blue light components (hello Rayleigh scattering), I think they are probably locking mice up in completely black boxes with only artificial light and then sensationalizing the title of "blue light is actually good" to get attention as opposed to actually saying that from a holistic view of health.
They also seem to be arguing for manipulating human circadian responses with artificial light which I personally think is premature and problematic.

Honestly, I feel like this project is easily at a level where the f.lux team would be open to (or even comfortable with) collaborating. They already have a Philips hue tie-in in the app, and might shed some insight in to which exact science relates to room-lighting colour temperature.

[claytonjn]

I'm open to anything that fits within the core mission of this integration, which is to use lighting to help maintain internal circadian rhythms in the most effective way possible.

I know I've posted a couple comments recently that probably make it appear as if I'm not open to change or suggestions and that certainly is not the case at all. I'm welcome any input into improving this as much as possible. I'm by no means an expert in health or lighting and my research has been mostly superficial because there doesn't seem to be a whole-lot of detailed studies on this topic. There's lots of evidence that lighting has a major impact on wellness, but not a lot of conclusive information regarding what optimal lighting actually is.

I believe there are, in general, one of three different behaviors that the end user is looking to achieve. In my home I mostly use the default configurations, so behavior follows sunrise and sunset - this does result in lights starting to dim as early as 5pm, but for us this is not an issue. There is a second behavior that many users want, which is the same but also takes into consideration current weather, so more than following the season the lights follow the actual current conditions - often it is suggested to add the option to use a lux sensor for this. The third behavior is to have hard-coded sunrise and sunset, so the lights behave exactly the same, year-round. I honestly do not know which of those is best for the primary goal of maintaining a healthy circadian rhythm - I tend to think it would be the third option, but that's not much more than an educated guess.

At the end of the day, I'm completely open to suggestions and improvements, but they need to be backed by some science or at least some logic I can get behind - CL is not meant to be utilized simply for aesthetic purposes. I also have to keep in mind the complexity of configuration - I would be much more open to additional options if/when I get CL set up to be added from the integrations section in the frontend.

[claytonjn]

I haven't thought this through fully, but perhaps simply having the ability to configure different "sunrise" and "sunset" times for color temperature and for brightness would work? That way for your use case you could have color temperature follow sunrise/set and have brightness hard-coded always from 20:00-7:00 for example.

Or, to be honest, maybe brightness should be removed altogether and a different integration should be created/used for that purpose.

[Joshfindit]

I highly suggest looking at the work done by the f.lux team as they have been instrumental in creating the discussion around artificial light and circadian rhythm .

For example the graph proposed by @burkemw3 looks very similar to the default graph in the current version of f.lux:

Credit: https://justgetflux.com/news/pages/macquickstart/

[Joshfindit]

(As a side note: personally I like this because when it goes down to 1800K, I know it’s definitely time to get off the computer)

[burkemw3]

I have a good first impression of having a sunrise/sunset config for color temp and one for brightness. I'm hoping to hack that together and try it out myself for a little bit.

I understand the projects goals and your evaluation of changes better now. Thank you for elaborating. I am also no expert on rhythms. I am primarily going on the evaluation of using CL in my household. This is an issue we're having.

A random internet article ¹ does say sleep needs frequently increase in the winter months. The same article recommends at most 9 hours sleep at night. Taking the Chicago example again, the time between astronomical twilight (no more illumination from the sun) is 11.5 hours long. Longer than the recommended 9 hours of sleep. I want some light brightness to keep me awake long enough to get enough sleep (e.g. 9 hours), while also jarring me (e.g. light too blue in the evening).

I'm not expecting this to convince you now. I hope you'll think about and/or research it more. I'll resume attempting to convince if I like the behaviour of my hack

Footnotes

1. https://www.sleep.org/articles/changing-sleep-needs-2/ ↩

[ahmadtawakol]

I believe there are, in general, one of three different behaviors that the end user is looking to achieve.

There is a fourth behaviour that I actually came here to suggest before finding this thread.

I also found that the lights were dimming too quickly when winter started, but I managed to fix it with a few hacks that I can't say I'm very proud of, but oh well.

What would work better for me is a light that starts at 100% at sunset and dims all the way down to 0% at sunrise. This would break the brightness during the day, but I set my lights to 100% while the sun is up and use this plugin after sunset only. I think this would work better for both winter and summer without needing to make any changes to our automations. I also feel that this is the most "natural" approach that mimics nature a little bit better.