IMHO the starlights and better cameras do not need as much light. I am thinking the "right" kind of light would help things a lot. That is what I am trying to find out. Which "type" of light is best for our IP cameras.
And this is far from trivial to figure out. Just like a digital photo camera, these security cameras use sensors that consist of arrays of photodetectors that incorporate color filters ahead of the individual sensors. Most use a
Bayer filter arrangement.
To create a color bitmap from the information the array produces requires software to interpret the data. And thus gets us into a subtle area of image processing that photographers deal with when shooting and then processing their photos. Initially, what the image sensor and analog-to-digital converters produce, as the sensor is "read out", is a file containing the "RAW" voltage readings from each of the millions of individual sensors in the array. Some have a red filter, some have a green filter, and some have a blue filter. The camera's internal firmware performs the "de-Bayering" and color balance adjustments on that "RAW" image data. The result is a color bitmap file. The "RAW" data is then discarded.
You'll notice that most cameras have settings for different ways to calculate white balance when performing the de-bayering calculations. Usually there are several manual presets, such as "daylight", "incandescent", etc., and then an "auto" setting.
For our security cameras, you might think that, as with casual photography, we simply want to get the most natural-looking colors with the least hassle.
But what you find when shooting more critical photo situations is that you cannot always predict the future. So choosing a processing preset is difficult or impossible. And as with shooting still photos using "JPG-Only" mode in a still photo camera, our security cameras are discarding the RAW image data, and we cannot go back in time and recover that data.
And because selecting a color balance preset, or using auto-white-balance requires applying non-linear adjustments to the RAW image data, we can end up with "clipping" or "zeroing out" of image data that wasn't clipped or zero in the RAW data captured by the camera's sensor. The result is that our color balance selection, in the camera settings, can reduce the dynamic range and lose details that the sensor itself actually captured. But again, that RAW data gets discarded, so we cannot go back and reprocess it with different color balance settings to recover that lost image data.
Many still photo cameras have the option to store the RAW image data so you can play with the processing settings after the fact, and get exactly what you want from each shot. And some video cameras offer different recording modes that use different tone curves, etc., that may provide better dynamic range.
But the security cameras I've played with don't offer anything like that. So we need to be careful when choosing the color processing settings. Settings (and lighting) that gives "off" colors may actually be preserving more dynamic range, and prove more valuable than lighting and white-balance combinations that produce more "pleasing" colors.
So, again, this is far from trivial. And what seems best to the eye, at first, may not really be capturing the most detail.
To get the exposure histograms that a still camera displays to better match what the sensor is actually capturing in the RAW files, some of us will adjust the in-camera JPG processing settings to optimize the histograms. And this results in the camera producing dull, bland, useless in-camera JPGs. But our RAW files are more accurately exposed because the exposure histograms, which are, annoyingly, always based on the in-camera JPG processing, end up better matching what the sensor is actually capturing.
One wonders if some similar tricks could be used to get bland, but higher dynamic range images from our security cameras when setting up lighting and camera settings. And if so, could we accept looking at the off-colors and bland dynamics for the sake of capturing the most detail. And would it be even better to have truly monochrome image sensors in these cameras, and give up color altogether.
Doing some research on the subject I came across this article. Looks like they are talking 400-700nm white light led's.
https://www.axis.com/files/feature_articles/ar_axis_raytec_summary_47904_en_1206_lo.pdf
That's a very interesting document. I need to read it all. But on quickly scanning it, I just want to say that if possible, I would place the external illuminators fairly far from the cameras to reduce the effect of brightly lighting bugs, snow, rain, etc., right near the camera. I know this would be difficult in a pole-mount situation.