Very cool. I’m the CEO of Innerscene (https://innerscene.com) and we make a commercial artificial skylight that uses some of these concepts. Actually the coelux ht25 model is almost identical to what you made but using smaller lenses and more LEDs - however the effect they were able to achieve still isn’t that great, the sun looks like a giant orb and once you get a few feet away you can make out a sun at all. We spent a lot of time working on perfect collimation and hiding lens edges and making sure the view into the sky was seamless and artifact free. I’d say the last 10% of that problem is 90% of the work. :). I think we successfully cracked the nut but currently using a lot of expensive parts so working on brining the cost down. If you search Innerscene patent many of our approaches are spelled out. We also spent a lot of time on simulation and software…
For me, the problem with this setup (and with most high efficiency LED lights) is the lack of red wavelenghts. Real sunlight has a substantial amount of energy in the very red end of the visible spectrum (700 nm) and also of course quite a bit in the infrared. These lamps have two spectral peaks: a narrow peak in the blue range, around 450 nm, a broader peak in the green, centered around 580 nm. That greenish peak falls off sharply, and has almost no energy in the red end.
The color sensitive cones in our eye have three peaks of sensitivity, the S cones in the blue range, the M cones in the green, and the L cones in yellow. The L cones are what your brain uses to see red colors, but they are actually pretty insensitive to deep reds like 700 nm. That’s why you THINK that LED lamps produce red, because they stimulate your L cones, but they do so without actually emitting much red energy at all!
Our bodies are sensitive to deep red light. The cytochrome in our mitochondria respond to it. There’s an experiment where shining red lights on the skin improved sugar metabolism. That makes sense, because we naked apes evolved under red-rich sunlight.
So these lamps may look like sunlight, but they’re missing some crucial wavelengths.
I saw "artificial sunlight" and thought "oh wow I'd love to see the spectrogram of the lighting solution this person came up with". I was disappointed to merely see "CRI 95+".
This is great video on the shortcomings of "CRI" - it explains in detail CRI, CRI extended, TLCI, TM-30, and SSI.
Brightness and color temperature are only two small parts of lighting - more people should start investigating the utility of taking their own spectroscopy measurements to figure out what lighting works best for them personally. My friends have very, very diverse opinions on what spectral distributions they like/hate, but they lack the language and experience to identify or communicate their preferences except for "Ooh I like/hate this bulb".
I mostly use LED bulbs to keep heat generation down (I pay for the heat twice in Houston: once to generate it and again for the A/C to negate it). But I always mix in a bit of incandescents / halogens (2400-3000K) which provide full-spectrum blackbody radiation to see ALL the wonderful colors in my world.
Really cool! I'm working on a lamp that gives you daylight levels of light indoors (albeit no raleigh scattering and columnated light). On the bright side (pun intended), it's 50,000 lumens instead of ~4500. https://getbrighter.com/
The brightness enhancement film is a transparent optical film. It consists of a three-layer structure. The bottom layer of the light-incident surface needs to provide a certain degree of haze by back coating, the middle layer is a transparent PET substrate layer, and the upper layer of light-emitting cotton. It is a microprism structure. When the microprism layer passes through the fine prism structure of the surface layer, the light intensity distribution is controlled by refraction, total reflection, light accumulation, etc., and the light scattered by the light source is concentrated toward the front side, and the unused light outside the viewing angle passes through the light.
So, it's similar to your design, but the grooves are very small.
EDIT: After reading the article, I see the OP calls out DIY Perks specifically - the OPs design is much more compact :)
> It's compact. The total size is 19cm x 19cm x 9cm. This is quite compact for a 5cm focal length and an effective lighting area of 18cm x 18cm. Reflective designs like the DIYPerks video or commercial products like CoeLux do not achieve this form factor.
> the main thermal issue when scaling up would be the cooling of the power supply itself, not of the lamp
If I would be scaling up that device, I would consider an ATX power supply. These are relatively large and typically include an active cooler inside, but they can easily supply hundreds of watts at 12V, often have an on/off switch on the back, are relatively inexpensive (at least unless you need much more than 500W of power), and are available everywhere. Usually, you just need to connect the PS_ON wire with the ground to make them turn on once powered.
I would pick a CDM bulb any day over the alternatives, including LED, unless power consumption is an issue.
“
Philips daylight CDM lamps are extremely efficient ceramic metal halide lamps with a spectral output close to natural sunlight. As a result, plants form more lateral branches, have smaller inter-nodal spacing, more flowering sites, and larger root systems, culminating in strong, healthy growth and high-quality yields.
Philips CDM bulbs have an amazing operating life. They maintain their high output for a lifespan of 30,000hrs on average.”
Is there a reason why you went with traces rather than pours? I count 7 signals per board, and they're all meant to be low impedance. You could even expose copper on the back of the board to be used as ad hoc heat sinks without spending any extra money. The weird little triangle loops on the back really stand out to me, even though you probably don't need to worry about the impacts of a loop in your circuit.
I'd be interested in seeing the optical spectrograph of the LEDs. If you want to simulate sunlight you want a full-spectrum LED like a Samsung LM301 series LED which are popular in grow lights. Not all LEDs are created equal, and even the LEDs in many "grow" lights only show two sharp peaks at red and blue wavelengths. A full-spectrum LED will output colors across the visible spectrum of light. You can't tell by looking at them, so you can either buy ones from which you trust the manufacturer or do what I did and build a cheap optical spectroscope using a raspberry pi with a small camera attached, a spectroscope lens, and some python code. I'm sure there are guides you can find with a quick web search if interested in making one.
I did something similar last year, but I was more focused on spectrum and used high powered movie lights, since their spectrum output is much better and you get a lot of lux. You can get most of the parallel light ray effect by adding a hyperrefelector which conveniently comes with most lights. I then added an incandescent heat lamp to fill in the IR part of the spectrum and some dedicated UV LED lights for the UVA part. I'm not brave enough to put in a UVB specific lamp.
I’ve been thinking about the potential benefits of incorporating something like this into VR headsets. Since VR already controls the user’s visual perception, would adding a high-intensity blue light (not via the OLED displays but indirectly) allow for a more effective way to simulate bright daylight conditions? Achieving 10,000 lux at the eye would be much easier this way.
I've been down the artificial sunlight at home rabbit hole too (also inspired by the same DIY Perks video). Ultimately the solution I settled on was a much lower budget 300W equiv LED Corn style bulb (4000lm) with good CRI, and a 250w incandescent heat lamp (like you'd get for a reptile). I decided to focus on heat from the light source rather than the parallel rays. Total cost was around $75 (LEDs were $25 for 2, heat lamp was $10 for 2, plus $20 each for 2 lamp stands that clamp to a desk and can support 250 watts).
Really neat, thank you for posting. I took a different angle when trying to figure this out - my goals were high spectral similarity to the sun, high brightness, adjustable warmth, and low cost. The existing lower-cost solutions people have posted about end up with high brightness, but the lights are evil.
I ended up with a photography light that's /alright/. It's not nearly as bright as I want, and I can't automate changing the warmth. When I next take a crack at this, I'll look deeper into some of what you've posted about here.
https://store.yujiintl.com/ has broad spectrum LEDs/lamps/fake skylights that may also appeal to others interested in the OP--just noticed Yuji was mentioned in OP. I'm not affiliated with Yuji but have ordered some of their products and been very happy with their performance.
I cannot remember if I first saw this here on HN but, here is a detailed video of an incredible artificial sunlight rig build. It's big but the results look, well, amazingly real, at least to the camera.
Projects like this are a great demonstration of just how far hobby projects can go today by exploiting some fairly recent innovations. In particular, designing a custom lens geometry and having JLC manufacture small runs cheaply is especially impressive; and of course, what you can get out of a modern LED is just absurd.
I built a very bright light source using a lamp stand from Ikea and the brightest, narrowest beam LED reflector I could find from my big box hardware store. The project works and gave about 10,000 lux but had a few drawbacks. After reading a book for a while a got a headache. The slightest imperfection (read dirt) in the arms of the armchair becomes super obvious. When I look at my right arm holding the book, it's so bright I'm sure I can see veins and arteries. I ended up getting a lamp less bright and that's great for sewing/electronics/fine work but there are limits for how bright you might want to go indoors. I kept the too bright globe for demonstration purposes.
I remembered seeing, loving, and saving that original DIY Sunlight video back when it first came out, and I went to your article hoping for more, and that's what you gave us. Brilliant job, and thanks for sharing!
As someone more on the software side as well, I’m inspired to take on something similar.
But more importantly, as someone who rented a windowless room for a year, I would have loved to make a smart light like this to wake up to in the morning.
This is one of the most satisfying DIY reads I've seen in a while. I love how you approached this as a software dev and essentially built a full stack for hardware iteration.
It you put out enough light, you can diffuse it more, such as by bouncing off the ceiling or a light-colored wall.
You can also use photography modifiers, like the various kinds of reflectors. Even white photo lighting umbrellas (which you shine a bright light through) help a lot.
Whether or not you add diffusing, you might want to make your light source mountable on a photo studio light stand, so that you can position and aim it however you want (just like you would with a studio strobe or hotlight).
Very cool. I do think the DIY perks version just pops a little nicer but it could also be the photography of the project. But the firm factor if your project is much nicer as well.
I think both projects would be better suited for different needs so again very cool. If you had lots of space the diy perks would be the way to go. If you only have a thin wall or ceiling then yours would be best. All in all thanks for posting this.
This project is fantastic, but you also deserve congrats for the 'Lens Maker'[0].
"If we have a compound optical system made of a series of lenses, mirrors, etc., we can treat each optical element as the layer of a neural network." Kudos.
Honestly this area is so interesting to me because of how incredibly strongly light (or lack thereof) affects in the winter. It also seems to affect me more strongly than most in regards to sleeping patterns. Too much blue light in the evening and bam, brain refuses to sleep until 2am or later.
So for me the biggest factor missing in these kinds of projects is a dynamic color temperature. While we get that from products similar to apples adaptive lighting, that’s missing in products like this. It seems we can only really have one or the other.
My dream is something like this build but with full adaptive and programmable color temperatures based on time and seasons.
Honestly I want to build something like this but my area of expertise is too far into the software domain, and not very far into the hardware or electrical engineering domain.
Very cool! I'd be curious to see an HDR version of it using bracketed exposures--it might give a better sense of how it actually looks in person. It seems really bright in the photos, so the shortest exposure would probably need to be very low to capture a good dynamic range for the HDR.
I've just mounted tons of full spectrum LEDs in my home office. Off the shelf, fairly cheap, easy to replace. My plants like it, except for the Goeppertias that keep dying on me
This is a great build and report. Even though you don't have great photo gear, you should try capturing a picture of steam in the light (as DIY Perks did).
> Kinda cool that you can see a lens flare effect in the shape of the lens grid array.
A lens flare is just a copy of the scene you are photographing, but strongly attenuated in brightness, and possibly rotated 180°.
> 100,000 lux
Your notation is correct, and there are other ways to write it. The SI unit "lux" has the symbol "lx". Your quantity can also be written as 100 kilolux or 100 klx.
Idiotic question, but can you have an "angled mode" that lets the godray travel during the day, following the real suns position, given the location and mounting direction of the light?
I tried making artificial sunlight at home
(victorpoughon.fr)622 points by fouronnes3 27 March 2025 | 233 comments
Comments
The color sensitive cones in our eye have three peaks of sensitivity, the S cones in the blue range, the M cones in the green, and the L cones in yellow. The L cones are what your brain uses to see red colors, but they are actually pretty insensitive to deep reds like 700 nm. That’s why you THINK that LED lamps produce red, because they stimulate your L cones, but they do so without actually emitting much red energy at all!
Our bodies are sensitive to deep red light. The cytochrome in our mitochondria respond to it. There’s an experiment where shining red lights on the skin improved sugar metabolism. That makes sense, because we naked apes evolved under red-rich sunlight.
So these lamps may look like sunlight, but they’re missing some crucial wavelengths.
https://www.youtube.com/watch?v=lH_owRxupC0
This is great video on the shortcomings of "CRI" - it explains in detail CRI, CRI extended, TLCI, TM-30, and SSI.
Brightness and color temperature are only two small parts of lighting - more people should start investigating the utility of taking their own spectroscopy measurements to figure out what lighting works best for them personally. My friends have very, very diverse opinions on what spectral distributions they like/hate, but they lack the language and experience to identify or communicate their preferences except for "Ooh I like/hate this bulb".
I mostly use LED bulbs to keep heat generation down (I pay for the heat twice in Houston: once to generate it and again for the A/C to negate it). But I always mix in a bit of incandescents / halogens (2400-3000K) which provide full-spectrum blackbody radiation to see ALL the wonderful colors in my world.
The brightness enhancement film is a transparent optical film. It consists of a three-layer structure. The bottom layer of the light-incident surface needs to provide a certain degree of haze by back coating, the middle layer is a transparent PET substrate layer, and the upper layer of light-emitting cotton. It is a microprism structure. When the microprism layer passes through the fine prism structure of the surface layer, the light intensity distribution is controlled by refraction, total reflection, light accumulation, etc., and the light scattered by the light source is concentrated toward the front side, and the unused light outside the viewing angle passes through the light.
So, it's similar to your design, but the grooves are very small.
https://www.youtube.com/watch?v=6bqBsHSwPgw
EDIT: After reading the article, I see the OP calls out DIY Perks specifically - the OPs design is much more compact :)
> It's compact. The total size is 19cm x 19cm x 9cm. This is quite compact for a 5cm focal length and an effective lighting area of 18cm x 18cm. Reflective designs like the DIYPerks video or commercial products like CoeLux do not achieve this form factor.
If I would be scaling up that device, I would consider an ATX power supply. These are relatively large and typically include an active cooler inside, but they can easily supply hundreds of watts at 12V, often have an on/off switch on the back, are relatively inexpensive (at least unless you need much more than 500W of power), and are available everywhere. Usually, you just need to connect the PS_ON wire with the ground to make them turn on once powered.
It was discontinued for a while but I’m happy to see it’s back in production?
https://www.futuregarden.co.uk/philips-ceramic-discharge-met...
I would pick a CDM bulb any day over the alternatives, including LED, unless power consumption is an issue.
“ Philips daylight CDM lamps are extremely efficient ceramic metal halide lamps with a spectral output close to natural sunlight. As a result, plants form more lateral branches, have smaller inter-nodal spacing, more flowering sites, and larger root systems, culminating in strong, healthy growth and high-quality yields.
Philips CDM bulbs have an amazing operating life. They maintain their high output for a lifespan of 30,000hrs on average.”
I wrote about it here: https://metrep.substack.com/p/improve-your-focus-productivit...
Some people report much better mood in dark winter days:
https://meaningness.com/sad-light-led-lux
https://meaningness.com/sad-light-lumens
Buy a cheap lux meter if you will be doing this since otherwise you are flying blind.
I ended up with a photography light that's /alright/. It's not nearly as bright as I want, and I can't automate changing the warmth. When I next take a crack at this, I'll look deeper into some of what you've posted about here.
https://www.youtube.com/watch?v=6bqBsHSwPgw
As someone more on the software side as well, I’m inspired to take on something similar.
But more importantly, as someone who rented a windowless room for a year, I would have loved to make a smart light like this to wake up to in the morning.
You can also use photography modifiers, like the various kinds of reflectors. Even white photo lighting umbrellas (which you shine a bright light through) help a lot.
Whether or not you add diffusing, you might want to make your light source mountable on a photo studio light stand, so that you can position and aim it however you want (just like you would with a studio strobe or hotlight).
Does anyone know anything about that?
"If we have a compound optical system made of a series of lenses, mirrors, etc., we can treat each optical element as the layer of a neural network." Kudos.
[0]https://victorpoughon.github.io/torchlensmaker/
So for me the biggest factor missing in these kinds of projects is a dynamic color temperature. While we get that from products similar to apples adaptive lighting, that’s missing in products like this. It seems we can only really have one or the other.
My dream is something like this build but with full adaptive and programmable color temperatures based on time and seasons.
Honestly I want to build something like this but my area of expertise is too far into the software domain, and not very far into the hardware or electrical engineering domain.
See https://brilliantlightpower.com/plasma-video/
and scroll down to the
February 10, 2023 COMMERCIAL SUNCELL® INITIAL SHAKEDOWN
video.
500 W of luminosity in such a small package, as it turns out, seems close to a thermal limit, even with 90+% efficiency.
> Kinda cool that you can see a lens flare effect in the shape of the lens grid array.
A lens flare is just a copy of the scene you are photographing, but strongly attenuated in brightness, and possibly rotated 180°.
> 100,000 lux
Your notation is correct, and there are other ways to write it. The SI unit "lux" has the symbol "lx". Your quantity can also be written as 100 kilolux or 100 klx.
:-)
But my neighbors kept complaining 8-/
Haters!
Probably the same people that keep flagging comments on the articles I'm interested in...