• The forum software that supports hummy.tv has been upgraded to XenForo 2.3!

    Please bear with us as we continue to tweak things, and feel free to post any questions, issues or suggestions in the upgrade thread.

LED lighting

  • Thread starter Thread starter Deleted member 473
  • Start date Start date
Last edited:
I would be interested to see the '3 lamp' experiment repeated using different colour settings on the camera, and perhaps a second make of camera if available.
This is purely because as mentioned earlier the spectral distribution of the three lamps differs and digital camera detectors are also 'sensitive' (my camera has settings for daylight, fluorescent and tungsten for example).
I'd like to see if the different settings gave a different result in the pictures - though obviously what your eyes tell you is the ultimate arbiter.
 
The best you will be able to do is 1) remove the shades from the 3 bulbs, 2) Manually set white balance on the camera using illumination from all three bulbs, 3) Under expose any sample pictures taken after setting the white balance so as to move the 3 light sources away from any white saturation, this will accentuate any 'off whiteness' in the light sources. The resulting pictures will still need some interpretation as to which is the nearest to white, especially if they are of different intensities
 
Last edited:
The camera white balance is irrelevant - all that does is apply scaling to the RGB values post-sensor. You can do that just as well in post-processing. The globes are not tinted, and provide useful diffusion as well as illumination arc information. The exposure was set to bring the saturation down to acceptable levels.

You can see the colour of the CFL and the LED, simply by comparison with the GLS. The tints of the shadows also provide colour information.
 
Black Hole said:
Am I alone in thinking that buying an expensive piece of lab gear doesn't count as "easy"? Integration is exactly what the referenced item does for you!
Click to expand...

You claimed it was not easy to do. For many bulbs it is already done.

What you meant was that you could not do it, so you resorted to a method that measures something totally different. And a picture that was already on the Internet.

Why not repeat with narrow angle gu10s pointing at your camera?
 
Black Hole said:
The camera white balance is irrelevant
Click to expand...
It isn't irrelevant, if, in the worst case the white balance of the camera was pushed as far as possible towards the blue end of the spectrum as possible and then the picture was taken, any subtle differences in yellow levels would be lost and no amount of post-processing would get them back. Manually setting the white balance to be sum of the 3 bulbs sets a common white and underexposed pictures taken with this setting of an individual bulb will show how far from the 'common white' it is. This won't show which bulb is correct, but it will show a colour comparison
 
Last edited:
Ezra Pound said:
It isn't irrelevant, if, in the worst case the white balance of the camera was pushed as far as possible towards the blue end of the spectrum as possible and then the picture was taken, any subtle differences in yellow levels would be lost and no amount of post-processing would get them back.
Click to expand...
The operative word being "if".

Mike0001 said:
You claimed it was not easy to do.
Click to expand...
You claimed it is easy, the context being in an amateur/home environment not a professionally-equipped laboratory.

Mike0001 said:
you resorted to a method that measures something totally different.
Click to expand...
I haven't claimed to measure anything at all. What I have produced is a comparative study between a particular LED lamp and two samples of CFL and GLS lamps. I have not claimed anything that is not evident in the accompanying photographs, for anyone prepared to look at them with an open mind.

Mike0001 said:
And a picture that was already on the Internet.
Click to expand...
Now you've lost me.
 
My interest was more in the possible effects of interference between the camera and lamp (particularly the CFL and LED) spectral responses.

Cameras vary, even good ones, as this shows. Cheaper compacts may be far more ragged.
CCDSpectResp.jpg

(linked from http://publiclab.org/wiki/ndvi-plots-ir-kit)

For a typical fluorescent it looks like:
Fluorescent_lighting_spectrum_peaks_labelled.svg

(Linked from Wikipedia)

So you can see the Nikon and Canon red sensors are likely to 'see' that 12 peak differently.

For LEDs the distribution can be different again:
Spectral-Power-Distributions.jpg

(Linked from http://www.slrlounge.com/gear-review-fiilex-led-light/)

In this case the blues of the two cameras will likely give different results.
 
What is the matter with you people? This is not an exercise in photometry.

The quality of the light is not on the cool side to my eye, and the camera output is giving a good representation of the visual appearance otherwise I would not have presented it. So either you think I am not being honest about it or you think there is something wrong with my visual system. All I was doing was demonstrating that the Aldi LED lamp gives an output at least equivalent to a 40W GLS, is not blue, is not confined to a narrow beam, and therefore does not make the room like a cave, so that if anyone felt the desire they could also buy some informed by my experience.

While I would love to take time out to build a spectrometer and calibrate it somehow, it isn't going to happen. What I will do is report on the "dimmable" function when I have tested it.

BTW, in case you want to know: the camera is a Fuji FinePix HS50EXR (bridge), but I couldn't find a spectral response curve for it.
 
Black Hole said:
The operative word being "if".
Click to expand...
So you accept that your statement in #84 was incorrect then, that the nearer the white balance of the camera is to the summed colour of the 3 bulbs, the more accurately you will be able to detect any differences in colour between the bulbs, making the white balance of the camera far from irrelevant
 
Did I say that? I don't think I said that. It may well be possible to accentuate any differences in tint between the bulbs by adjusting the camera white balance, but that is still possible to do in post processing because the channels are not saturated (or at least are not saturated everywhere over the diffusion of the globes).
 
Last edited:
Have you ever tried to correct an image that was incorrectly white balanced in the camera by using post processing?, once yellow levels have been reduced to very low numbers any small differences are lost, multiplying them back up can never restore this.

Example, with 24 bit colour, the 3 bulbs are :-
red 249 + green 237
red 225 + green 230
red 240 + green 227

If you bias away from yellow by dividing all red and green values by 25 you get :-
red 9 + green 9
red 9 + green 9
red 9 + green 9

You can post process the '9's as much as you like, but you won't get the correct figures back
 
Last edited:
Black Hole said:
What is the matter with you people? This is not an exercise in photometry.
.....
So either you think I am not being honest about it or you think there is something wrong with my visual system.
Click to expand...
Sorry. Your initial 'test' and analysis was fine. I was just curious as to how different the results might be using a different camera or settings.

I don't think there is necessarily anything wrong with your vision, but colour perceptions do vary from person to person even though they may not be classed as 'colour-blind' ... but that's a whole different can of worms.
 
Ezra Pound said:
Have you ever tried to correct an image that was incorrectly white balanced in the camera by using post processing?, once yellow levels have been reduced to very low numbers any small differences are lost, multiplying them back up can never restore this.

Example, with 24 bit colour, the 3 bulbs are :-
red 249 + green 237
red 225 + green 230
red 240 + green 227

If you bias away from yellow by dividing all red and green values by 25 you get :-
red 9 + green 9
red 9 + green 9
red 9 + green 9

You can post process the '9's as much as you like, but you won't get the correct figures back
Click to expand...
This implies the camera would also have to divide the red and green channels by 25 when doing the white balance correction internally, resulting in the same problem.
 
"Dimmable"???

In theory, maybe. In practice, all the dimmers I can find have a minimum load requirement of 40W or more. Still, I'll connect one up and see what happens. If necessary I can include a GLS in the fitting to provide a dummy load (for experiment only).
 
Black Hole said:
This implies the camera would also have to divide the red and green channels by 25 when doing the white balance correction internally, resulting in the same problem.
Click to expand...
Sigh : You've lost the plot here, by setting the white balance on the camera correctly you are ensuring that it DOES NOT divide the red and green channels at all to obtain maximum sensitivity for yellows
 
Er, no. How do you think a digital camera adjusts for colour balance? What is the likelihood of having to divide a channel by 25??
 
You must log in or register to reply here.
Back
Top