For LED, what color temperature?

Hi Folks,
For those of you using white LED’s for the light source, what color temperature are you using and why?

It seems some have moved away from the RGB LEDs and are just using a single white LED, and I’m trying to understand the pros and cons.


In my opinion, it seems that it would be reasonable to choose a color temperature similar to that of the halogen lamps used in projectors, around 3000 K.

The closest thing I have found has been a 3300 K lamp. However, after making an equalization adjustment of the color gains on the camera, I think good results are obtained.

1 Like

The film was projected with a lamp near 3200K color temperature, and most sensors are characterized with 3200K color temperature.

However, on the other side, most white LEDs have peaks and dips were radiant power changes drastically for certain wavelengths. This happens with all color temperatures, and it is a by-product of the color LED underneath the phosphor. Depending on the part/manufacturer, some particular temperatures have lesser peeks/dips, a flatter relative radiant power.

If going with white LED, I thought it would be more important (for me) to select an LED with less radiant power peak/dips, than to look for a specific temperature.

My selection, based on the above criteria was 5000K, but one cannot generalize that in all cases/manufacturers that would be the best option. I would recommend looking at the radiant power curve and temperature when making the White LED selection.


I am one of these guys which switched from narrow-bandwidth red, green and blue LEDs to white-light LEDs. Initially, I thought that I can optimize the color separation between the different color channels by a careful selection of the LEDs wavelength. I even tried to calculate these optimal wavelengths - only to figure out that the result was sub-optimal.

The reason for this is probably that the filter curves of the film’s magenta, cyan and yellow are unknown (at least to me). So you might end up with sampling narrow bands of wavelengths were the dyes have some dip in their response.

I tend to align myself more with Pablo’s (@PM490) approach - using a white-light LED with a mostly flat response. Specifically, I am using some Osram LEDs. In this way, you kind of averaging over possible dips in the dye’s response better.

The actual color temperature is probably not that important, for several reasons (but I am still looking into that by myself - so the following is only preliminary). I compared the results using the HQ camera with the normal color science and a color science specifically optimized for the Osram LEDs I am using - the results were visually indistinguishable (I will have to redo this, not sure any longer whether I made a mistake here). Provided, the correct white balance had been applied.

More importantly, in my experience, the film characteristic can change from roll to roll. I have footage from several rolls of Kodachrome from a 1981 San Francisco trip - were one roll features a funny greenish tint. One can correct in the post, but of course the colors will be anything than correct.

Furthermore, S8-film stock only knew two color temperatures: daylight and tungsten. If you were recording with different light (say fluorescent ones), you’d end up with quite some color casts. Again, potentially you can correct this in post production.

Also, once you start scanning different brands of film, you will notice color differences between the different film stocks. Kodachrome is quite different from Agfachrome, for example.

Ideally, you would have for each type of film a color chart target to adjust the colors in a perfect way. However, it is close to impossible to obtain such a thing for old film stock. Also, I think today, you would color grade the original scan anyway, to meet current audience expectations. So the final look will be anything than true to the original color anyway…