What Does This Info. Mean?

Stefan Berg Nighttime Imaging 'N' Astronomy (N.I.N.A. / NINA) Acquisition Help Needed
14 replies•747 views
Mossyback avatar
Mossyback avatar
I'm starting to feel a little more comfortable with N.I.N.A. and I'd like to understand more of the information provided.

Here's the HFR box at the top of the screen. Some is self-explanatory, some not so much. I'd like some help understanding the data.

I'm using an SX694 mono camera with a filter wheel on an Esprit120mm refractor. The information below is from a 180 second exposure (Luminance).

What is this information telling me about the image and/or equipment set-up, image quality and exposure etc.?

Width 2750               Height 2200
Mean 1969.88           SD 64.31
Median 1969.00        MAD 12.00
Min 1878 (1x)             Max 32481 (1x)
#Stars 58                    HFR 2.13
Bit Depth16                HFR SD 0.33
Gain –                          Offset –

Hank
Well written Engaging
Reg Pratt avatar
Reg Pratt avatar
https://nighttime-imaging.eu/docs/master/site/tabs/imaging/
Christian Koll avatar
Christian Koll avatar
Hank,

as far as I understand, these mean the following:

Width, Height: The dimension in pixels of your light frames (images)
Mean: mean pixel value in your actual frame
Median: median pixel value in your actual frame
SD: standard deviation of you pixel values
MAD: mean absolute deviation of you pixel values
Min: minimum pixel value in your actual frame (number of pixels in your image with that value)
Max: maximum pixel value in your actual frame (number of pixels in your image with that value)
#Stars: Number of detected stars in your actual frame
HFR: half flux radius of your stars (similar to FWHM)
Bit Depth: Bit Depth of your camera (16bit)
HFR SD: HFR standard deviation
Gain, Offset: Gain and Offset of your camera (unclear why no values shown)

HTH & CS
Chris
Helpful
andrea tasselli avatar
andrea tasselli avatar
SD = Standard Deviation of the statistical distribution of pixel values in the image
MAD = Mean Absolute Difference, i.e. the absolute value of the difference between the model used to fit the PSF and the actual data
Mean = Average of the pixel values of the image
Median = Median of the pixel values of the image
Max = Maximum value of the pixel values of the image
Min = Minimum value of the pixel values of the image
HFR = Average Half Flux Radius, i.e. the radius of the PSF where half of the photon flux is accrued across all detected stars in the image
HFR SD = Standard Deviation of the HFR
#Stars = Number of Stars detected
Gain = Gain
Offset = Offset
Helpful
Mossyback avatar
Mossyback avatar
Thanks for the replies. Perhaps a better question would have been - are the data what one should expect, on average? The target was M13.
Well written
Linwood Ferguson avatar
Linwood Ferguson avatar
Most of these are highly dependent on circumstances – type of sensor, OTA focal length and focal ratio, seeing, light pollution, etc.  As a simple example number of stars can go down dramatically as light pollution goes up, and as altitude of the target goes down toward the horizon.

HFR (or maybe better FWHM if you use Hocus Focus as the focuser) is a little more constant indicator of focus quality and seeing.  It's still quite difficult to compare to other and different equipment and circumstances but to some degree HFR will be in the low single digits, maybe 2-4 or so, but still it is best as an indicator of your situation (e.g. if higher one night than another probably worse seeing, or you did a worse job of focusing). 

Eccentricity (also requires Hocus Focus) is a measure of circularity of stars, which is an indicator of tracking accuracy.  Anything in the 0.4 or even 0.5 or lower is good, as that goes above 0.5 there's usually some problems with tracking causing stars to be footballs. 

These measures (that I mentioned) are largely independent of target.  Not completely, but the star detection does not pay attention to diffuse subjects.

It's also worth noting that these analyses are aimed at auto-focus, and not at being comprehensive.  If you (for example) compare FWHM from Hocus Focus with Pixinsight after the fact, you will see very different numbers.  They will track relative (i.e. if you take two images where eccentricity goes up, PI will show it going up also) but not in an absolute sense.  Pixinsight may for example detect literally an order of magnitude more stars with its deeper (and slower) analysis.

As a general rule you want to compare all these settings to your own changes - whether equipment, procedure (e.g. autofocus), weather, etc. as an indicator of whether you improved certain things or not, and not get hung up on their absolute magnitude or trying to compare to others.
Helpful
Mau_Bard avatar
Mau_Bard avatar
andrea tasselli:
SD = Standard Deviation of the statistical distribution of pixel values in the image
MAD = Mean Absolute Difference, i.e. the absolute value of the difference between the model used to fit the PSF and the actual data
Mean = Average of the pixel values of the image
Median = Median of the pixel values of the image
Max = Maximum value of the pixel values of the image
Min = Minimum value of the pixel values of the image
HFR = Average Half Flux Radius, i.e. the radius of the PSF where half of the photon flux is accrued across all detected stars in the image
HFR SD = Standard Deviation of the HFR
#Stars = Number of Stars detected
Gain = Gain
Offset = Offset

Ciao Andrea,
maybe I am wrong, I always thought that MAD was just the MAD of the pixel values of the image, as an alternative variability indicator to SD.
Ciao!

Mau
Mossyback avatar
Mossyback avatar
Thanks, Linwood. I’ll try to relax and not get hung up in the numbers as “absolutes”.
Well written Respectful
andrea tasselli avatar
andrea tasselli avatar
Ciao Andrea,
maybe I am wrong, I always thought that MAD was just the MAD of the pixel values of the image, as an alternative variability indicator to SD.
Ciao!

Mau


It depends. If it is referring to calculated PSF values I suspect it's the definition I gave above, if for the whole image then it is indeed Mean Absolue Deviation -> Sum(value - mean_value) /number of pixels
andrea tasselli avatar
andrea tasselli avatar
Thanks for the replies. Perhaps a better question would have been - are the data what one should expect, on average? The target was M13.

I would bring you camera offset down a bit as 1878 is a bit high  try and get I down to about 400-500 at most…👍🏻

Min values have nothing to do with offset. On top of that it's a CCD camera, offset is fixed as it is gain.
Linwood Ferguson avatar
Linwood Ferguson avatar
Just keep in mind (re obsessing over these) that 200 vs 500 is about half a percent based on 65535 full well.  Reducing it is "right" but don't expect to see the difference (and do expect to redo your darks and dark-flats with any change in gain or offset).   The vast majority of DSO data is on the far left side of the histogram, and with reasonable length exposures you are going to saturate the cores of some bright stars in all likelihood.  Reducing exposure to save these cores will often yield a massive number of very short subs and may be more pain in that regard even if the result may yield slightly better star cores. 

Basically: Sweat the big stuff first.   Get nice round, tight stars, in good focus, and always collect more data than you might think you need (because more data fixes many other problems).
Helpful Insightful