Histogram Question: 3 Peaks?

[feilimb]

I got my first ever 1.25" sized filter recently and wanted to try a trial run of capturing a flat to get an idea of how good/bad the vignetting might be. Here are the details of the setup:

SkyWatcher ED80 with 0.85 focal reducer

1.25" IDAS P2 LPS filter

ASI294 MC Pro at Gain: 120, Exposure time: 0.05 seconds

Light source: my homemade flats light source with LEDs (see: https://stargazerslounge.com/topic/325457-flatstub-a-rough-ready-diy-flats-solution/

NOTE: the scope was not exactly focused at infinity - not sure if this matters regards my query below on histogram)

 

Histogram (grayscale):

Histogram (debayered):

 

I'm a bit clueless when it comes to histograms and I was wondering if someone might be able to explain why I am seeing 3 peaks..?  I was trying to look at the histogram to get a rough idea for a good exposure length where the peak would be around the center along the horizontal, but I expected 1 peak instead of 3!?

FITS File (23MB): https://we.tl/t-FsBPF6wruC

[vlaiv]

Quite normal for OSC camera like ASI294 MC Pro.

Since you have color camera - there are some pixels that are blue, some that are sensitive to red and some that are sensitive to green.

When you shoot flats, or out of focus image or even focused image of the night sky that is not perfectly black (some level of LP) you will get "three-pronged" histogram - this is because pixel values "bunch" around a peak, but since you have three colors on sensor and their sensitivity is a bit different and also source light has different spectrum distribution - three peaks do not overlap and are spread by some amount (if pixel sensitivity was the same and light source was uniform - you would get one peak, or 3 peaks that would align so it would look like single peak).

Just for reference:

From this you can see that for ASI294 - green and red tend to be about the same peak sensitivity, while blue has less - so one would expect - green peak to be the left most, followed by red and blue furthest to the right - but this is only when illuminated by uniform light source. If your light source has for example very "cool" feel to it (meaning blue part of spectrum is dominant) - order can change.

You can also change this by changing "WB" settings on your camera - it will shift the peaks, and one way of color balancing the image is aligning all three peaks so their max is roughly at the same place in histogram.

Btw, don't worry about three peaks - make sure you don't get clipping on any of them (neither right nor left) and you'll be fine. Also - don't do "on camera white balance" - you'll do white balance in processing - capture as is and then adjust later unless you are doing any sort of EAA - then you can color balance on camera to get a good looking live image.

[feilimb]

@vlaiv thanks so much for your excellent reply.  It is reassuring to know that the peaks I am seeing are normal in my case, and yes the light source is very cool / bluish which explains red peak appearing left-most I guess (before the green peak).  When 3 peaks appear in a histogram is it possible to judge whether a particular exposure length is a 'good exposure length' for a flat image, or is simply seeing the presence of all 3 histograms (unclipped) sufficient to know that the flat is OK to use?

[vlaiv]

33 minutes ago, feilimb said:

@vlaiv thanks so much for your excellent reply.  It is reassuring to know that the peaks I am seeing are normal in my case, and yes the light source is very cool / bluish which explains red peak appearing left-most I guess (before the green peak).  When 3 peaks appear in a histogram is it possible to judge whether a particular exposure length is a 'good exposure length' for a flat image, or is simply seeing the presence of all 3 histograms (unclipped) sufficient to know that the flat is OK to use?

Just go with "normal" recommendation, but place right most peak to be there - that is about 85% - 90% ADU value.