SNR increment of final stack with different Reference Image

[AstroScience]

Hi guys,
I had run Star Alignment process on 62 subs acquired during two nights and of course used reference image with the tool.
The SNR of final stack I got were 4.77-455-3.18 , it's a DSLR data.

Then I tried different Reference Image with Star Alignment and my final stack SNR increment came out this time something like 3.56-3.44-2.7 .

Is this due to a different amount of stacked (overlapped) areas of aligned images?
 Should I pursue that specific reference Image again or the central area of the image would still have same SNR increments no matter what reference image I use?

[pfile]

to find out you can crop each image to just the central region and then run the NoiseEvaluation script on both and see what the noise is like in each.

in imageintegration, the increments which are relative to the reference are just that - relative to the reference. so if the SNR of the reference image is low, the increment will be large. if the SNR of the reference image is high, the increment will be lower.

II also reports the SNR increase over the average of the input subexposures; this should be similar regardless of the reference image.

[AstroScience]

Hi, Rob
I didn't mean different reference image when stacking, I mean different image when Star Aligning.

I used Star Alignment tool with one reference image and then stacked the registered subs. The SNR increase over the average of input subs where high (4.44-4.33-3.23)
Then I used Star Alignment again, but chose different reference image to align to and once again stacked the registered subs with same integration parameters.
Only now the average SNR increase got much lower (3.33-3.21-2.4).

[Ignacio]

My guess is that choosing the "best" subexposure as reference in star alignment (best fwhm, darker background), helps with spatial resolution. But the choice of one reference sub over another will have negligible impact on SNR, as this depends mostly on integration time, optics speed, object brightness, and camera and sky quality. Its a bit like poor guiding/seeing.

Ignacio

[Carlos Milovic]

Maybe if the reference frame has the worst traslation/rotation, then the interpolation of the other frames would act as a low-pass filter, changing the SNR measurement.

[AstroScience]

So, which of those two SNR measurement should I relate to?
It seems that one of them is definitely not accurate...

[Ignacio]

Have you made the comparison by selecting the same image area, such that in both stacks there is 100% overlapping of frames?

Ignacio

[AstroScience]

No,
you think that one stack were more "stretched" by the StarAlignment tool than the other?

Should I overlay one above the other and see if star positions/forms matches?
Not sure I understand how to compare them.

[pfile]

> SA vs II reference.

sorry about that, i can not read.