Teams / DSLR Documentation and Reduction / IRIS: a simpler procedure !

IRIS: a simpler procedure !

Citizen Sky is now officially permanent part of the AAVSO. In the coming weeks we will be moving additional content to the AAVSO site and freezing this site as an archive of the 1st three years of the project. Please visit the new landing page for future updates.
By GMara on April 12, 2010 - 7:49pm
Teams:

Hello to all!! I have made a presentation of the procedure in IRIS to some friends but the argue that this procedure seems to confusing. One of them, that he is already active to the project and sending data to jeff Hopkins, told me that because of the small exposure time there is no great need for bias,dark,flat frames. So i went on to remove these steps and after many tests with our sample data and others I managed to simplify a little bit the procedure. So perhaps, this (http://www.citizensky.org/teams/dslr-documentation-and-reduction/iris-tu...) can be a beginner's tutorial and the one that we have now to become an intermediate tutoria (since we take into account more accurate steps).Perhaps, someone else can check this to see if it is correct. One point that still have not clarify is the adding of images. I see that both arithmetic and median works, but i suppose that median should lower the S/N. So is it better then to select only the arithmetic method?Regards,Grigoris

Needs for dark, bias...
By Roger Pieri on April 19, 2010 - 4:31pm

Hello Grigoris, I fully agree on not applying bias anddark, from my own testing with a 450D, dark current noise becomessomewhat higher than the residual read noise in the dark frame (stack of 20)only for exposures longer than 32 seconds at room temperature and 1600 ISO. For our typical 4~6 seconds exposure at 100 ISO (corresponding to the optimum electrons accumulation on 100 pixels at mag = 3~5 ) the dark current noise is not measurablebut the residual read noise of the dark frameremains significant.Under that conditions making dark process is worst than doing nothing ! Bias variations in picture for this type of imagers is just nothing, no need for processing, that would just add noise (in any case) Anyhow any gradient (or constant !) would be eliminated bythe background subtraction at the photometry stage. The only offset to consider is the systematic offset applied at the ADC: 1024 on the 450D. 1or 2 extra ADUs of "DC component" drift can be noticed at high ISO. This is just to subtract 1024~1026 to any pixel of the picture: a constant. It is needed to subtract that systematic offset only if a flat process is applied (before it !) In that case we shall be sure thefollowing of the process uses "signed integers" that to avoid to clip the negative part of the gaussian noise that would affect the background value. The Flat Noise (gain variation from pixel to pixel) is very low on present imagers and can be ignored. But theVIGNETTING of optics is high and SHALL be corrected (center to corner easily 30% on long focus and much more on short focus !) Dust could be an issue, I would at least recommend to carefully check if none could affect the process: the working area shall be clean. Arithmetic against median: I made tests (noise evaluation) about that and didn't findthe median being any better... I do prefer averaging pictures to reduce the random noise. I would underline thathere above statementsare only for recent imagers, mostly CMOS. Several years old CCD could show significantlyworst FPN (Fixed Pattern Noises, including dark currents) Yours truly, Roger

DSLR Documentation and Reduction

  • You must login in order to post into this group.

Wiki pages

My Teams

Not a member of any groups.

Powered by Drupal