davidestevens

I'm assuming that, just as the ear is non-linear, and one needs to map a linear 0.-1. to a logarithmic curve for volume, a similar thing applies when controlling light intensity (in this case, LED lamps over DMX). Does anyone have a formula or know what such a curve would be? And is it different for red, green and blue?

mapping-0-1-to-light-intensity

Some drivers will map the 0-1 to 0-100% perceived output using a gamma curve.

Other drivers will not do that and the result will be linear and look odd.

The best is to check with the light that you are using and see if the driver is linear or not.

If the output is linear you will have to compensate with an expression (or vector expression). I found this interesting page on gamma:

http://www.poynton.com/notes/colour_and_gamma/GammaFAQ.html

Good question.
Some drivers will map the 0-1 to 0-100% perceived output using a gamma curve.
Other drivers will not do that and the result will be linear and look odd.
The best is to check with the light that you are using and see if the driver is linear or not.
If the output is linear you will have to compensate with an expression (or vector expression).  I found this interesting page on gamma:
http://www.poynton.com/notes/colour_and_gamma/GammaFAQ.html
G


http://www.lutron.com/TechnicalDocumentLibrary/367-2035_LED_white_paper.pdf

also found this interesting PDF.
http://www.lutron.com/TechnicalDocumentLibrary/367-2035_LED_white_paper.pdf
and in particular the diagram of page 9.


Thanks G. Downloading those articles now...

you could use the RGB to luminance (or luma? i keep mixing them up) colorspace conversion

 (green picture and patches in the middle of the site)

but that would probably be overkill for a led lamp.

https://cycling74.com/forums/rgb2xyz/ (green picture and patches in the middle of the site)

This is a really interesting question - I'd like to hear if you have any success...

Anecdotally, I've found that my own perception of brightness is definitely different between red, green and blue LEDs. Whether there is a scientific reason for this (perhaps like sound frequencies, we see certain light wavelengths better?), or it's more to do with me buying the cheapest LED lights I could get my hands on; I can't be sure.

I do find that when working with multiple lights, the DMX values I send to 'match' a bright red with a bright blue (for example) are noticeably different. The brightness curve is definitely exponential, rather than linear; for this I would point the finger at my cheap lights.

Here's where I've already gotten to with this...

http://eyemusic.org.uk/Education/seniorcitizens.html

http://eyemusic.org.uk/Education/mobilesensoryspa.html

What I want to improve is the relationship of hand movements to light intensities. At the moment it's a bit "light on, light off", though that is to some extent down to the way that people use the sensors. Even if one smoothly moves a hand up and down directly above the IR sensor, the controlled light doesn't fade up and down in a smooth, "visually linear" way. (It occurs to me that to test the lights, I should try using a fader rather than the sensors, so that I eliminate the noisy, non-linear control source!)

I've tried running the output control signals through estimated hand-drawn [table]s, but I find that it's actually pretty hard to tell (at least in the space I have to test in) how well it's working. So I was hoping there was a simple formula I could plop in somewhere that would be an official "linear to visual" curve transformation.

I do need to check, as geotruptrude suggested above, if there's any info on the driver response on my specific LEDs.

Here's where I've already gotten to with this...
http://eyemusic.org.uk/Education/seniorcitizens.html
http://eyemusic.org.uk/Education/mobilesensoryspa.html

What I want to improve is the relationship of hand movements to light intensities. At the moment it's a bit "light on, light off", though that is to some extent down to the way that people use the sensors. Even if one smoothly moves a hand up and down directly above the IR sensor, the controlled light doesn't fade up and down in a smooth, "visually linear" way. (It occurs to me that to test the lights, I should try using a fader rather than the sensors, so that I eliminate the noisy, non-linear control source!)
I've tried running the output control signals through estimated hand-drawn [table]s, but I find that it's actually pretty hard to tell (at least in the space I have to test in) how well it's working. So I was hoping there was a simple formula I could plop in somewhere that would be an official "linear to visual" curve transformation.
I do need to check, as geotruptrude suggested above, if there's any info on the driver response on my specific LEDs.


I map 'linear' FSR data onto LED brightness simply using a [pow] function, to override the off-then-on behaviour you describe. You can create many distortions, using this idea:

https://cycling74.com/forums/easing-and-tweening-how-to-make-a-tulip/

@noob_meister. Cool! thanks for that Brendan.

Mapping 0-1 to light intensity