white balance question

[brisguy]

I guess I am still confused about how to set white balance in SC. I thought you were supposed to either set it by eye on a known source or use the auto white balance. However, this results in some strange looking histograms and although I am not certain, seems to potentially cause premature clipping in Red or Blue. I also noticed that my planetary photos, after color balance adjustment in Registax, show obvious clipping and shifting of the curves for Red and Blue and leave only Green at the higher values.

Last night I decided to try setting all the colors to 100 and figured I would then set the white balance later. In SC4, the image has a green tint as expected, but after stacking in Deep Sky Stacker, the resulting TIFF file looks perfectly adjusted for color. Furthermore the stacked image histogram looks fine in Gimp with all the colors taking up the entire value range and equally balanced. I double-checked and confirmed I did not choose any color channel alignment or white balance settings in DSS.

So my questions:
1. Do the color channel settings affect what is captured or only what is displayed in SC? I thought it was the latter, but my experience indicates otherwise.
2. Any ideas as to why DSS produces what appears to be a white balanced result?

[admin]

Hi,

the colour balance sliders in the camera control area affect the saved data as well as the display. My preference is not to adjust these and to tinker with white balance later in the processing, but other people like to adjust here.

My experience with DSS is that it tends to produce a very washed out image by default - I usually have to increase the saturation. Perhaps that is why the WB seems reasonable?

cheers,

Robin

[brisguy]

Thanks for the information. I was going to say that should be in the manual, but then I found it. I have to say it is unexpected (and unwanted) since white balance is usually a post-process in photography and does not change captured data.

However, now I have a related problem/question. I took your advice and wanting to avoid losing any data during planetary capture, set all my color sliders to 100. However, to my disappointment, the histogram still showed a 'combed' response - indicating that there were several zero values in the capture area that should have had higher values. I did not capture a histogram at the time, but it is easy to reproduce: I simply set my capture to 8-bits and the problem returned. I capture a snapshot and CSV file for both 8-bit and 12-bit capture.

To make things even more confusing, although the 12-bit histogram looks nice and smooth, the CSV file shows regular intervals of zero values even though the histogram display did not.

So to my questions:
1. Why does the image histogram in 8-bit RAW show combing even though I set all the color sliders to 100?
2. Why does the image histogram in 12-bit RAW NOT show combing, but the CSV file does?
3. What is the source of truth? The CSV file or the SC histogram display?

Please see attached files.

[admin]

Hi,

as far as I can tell, the neutral white balance values for SVB cameras is 128, not 100. This is why you were still getting histogram gaps when set to 100.

I expect there are probably also gaps in 12 bit mode, but they may not show in SharpCap as there is not space horizontally to show all 4096 histogram slots - SharpCap shows a maximum of 1024 I think, so the averaging of adjacent values to make that graph will tend to hide histogram gaps. The CSV file is not averaged, so shows the true histogram values.

cheers,

Robin

[brisguy]

>as far as I can tell, the neutral white balance values for SVB cameras is 128, not 100

Ah, OK I will try that. 100 is the default value provided by SC4. How would I determine the 'neutral white balance' level other than trial and error?

One more related question. I don't understand why the green channel is always the brightest on the histogram. I would think that having double the number of green pixels would cause the histogram to show a higher peak (but I suspect the values are divided by 2), but not higher individual exposure values. Why is that?

[admin]

Hi,

yes, the value of 100 was incorrectly set as default in earlier versions of SharpCap 4.0 - this is now fixed in version 4.0.8169.0 and later versions.

The green channel brightness is related to the peak sensitivity of the camera being in the green part of the spectrum - even if there are equal amounts of light in the R/G/B channels, the camera detects more of the green. The pixel count being doubled doesn't have an effect on this movement of the green to the right.

cheers,

Robin

[brisguy]

Sorry, one more (I think!)...

Given the higher sensitivity to Green, it means that when I try to set the optimum exposure (getting close, but not clipping) while leaving Red and Blue with the same setting as Green, it leaves Red and Blue underexposed. Doesn't it make sense to boost the Red and Blue or does that just boost noise as well?

BTW - Setting each channel to 128 does the trick in terms of getting a smooth histogram.

[admin]

Hi,

no improvement in the data by boosting the R/B channels - all the boost does is multiply each pixel value by some number bigger than 1 (probably the white balance number divided by 128). This leads to the gaps in the histogram and ensures that there is no extra data gathered by doing the boost.

cheers,

Robin