When thinking about light output and in discussions I regularly feel the need to refer to this V (lambda) luminosity curve, that describes the relative sensitivity of the human eye for the different colours of light, as a fraction of the sensitivity to 555nm for which the eye is most sensitive. The V (lambda) function is also integral to the lumen unit, which takes the different sensitivities for different wavelengths into account.
For example, recently when I tested the output (in lumen) of a 660nm led (to be posted still), I wanted to know the sensitivity of the human eye for 660nm (deep red) compared to a common 625nm (red) led. The curve tells me that the eye is 5 times as sensitive to 625nm, so for 660nm you need 5 times the optical power to perceive the same brightness (and measure the same lumen) as 625nm.
Because on the internet you can only find low-res luminosity curves that are not easily readable, I made one myself, using the data for the CIE (1924) standard V (lambda) function, which is by my knowledge still the most used photopic luminosity function for measuring candela (and lux and lumen). Since then a few improvements were proposed because it became clear the curve was not entirely correct for the blue and red region, but still the old V (lambda)function is mostly used.
Photopic (black) and scotopic (green) luminosity functions.[c 1] The photopic includes the CIE 1931 standard[c 2] (solid), the Judd–Vos 1978 modified data[c 3] (dashed), and the Sharpe, Stockman, Jagla & Jägle 2005 data[c 4] (dotted). The horizontal axis is wavelength in nm.
(image and explanation from Wikipedia, link)
The dataset gives a coefficient per 1nm which is not practical for this curve, so I only used part of the dataset, per 5nm.
Here is the curve, mostly for my own reference, but maybe useful for others.
