Walking noise?

Colin Clackson

The attached image is a three hour integration using a 1600MM pro and taken through an LXtreme filter. None of the sub frames show this patterning, nor the Darks, dark flats or flats. It looks like walking noise to me but I dither 16 pixels between each frame. PDH2 is set to dither in RA only. Could this be the issue?  The object is NGC7129 and the second pic is the entire frame.

Funny thing is I didn't get a similar effect when I photographed the same object through RGB and Lum filters

Any ideas?

Colin

BlockHead

Do you measure any zeros in your frame?

(see my video on use of the Pedestal)

-the Blockhead 

Colin Clackson

Thanks for the response, Adam. 

I used your pixelmath approach to identify pixels with a zero value but found none. So I'm guessing its not an over subtraction. Any other thoughts?

BlockHead

It does look like walking noise...but it is SO vertical. I am suspicious.

You did not, by some chance, use the Linear Defect Correction in WBPP?

You should not use this.

-the Blockhead

Colin Clackson

The linear defect correction option is greyed out so it wasn’t used.

BlockHead

Ok... so how much are you dithering between frames?

-the Blockhead

Colin Clackson

My capture software is set to dither to a maximum of 16 pixels (see attached image). Blinking the uncalibrated frames shows a random shift of the stars along a straight vertical line There is no movement horizontally. 

I've also attached an image of the masterlight after full calibration, cosmetic correction (3 sigma), alignment and integration, an image of the low rejection and an image of the high rejection. All three images are matched to the same area.  I've never seen a high rejection result that included so much rejection of the background.

Lastly, attached are screenshots of my WBPP settings

Colin Clackson

Actually. Ignore the images. I mixed them up with some others. The high and low rejections of the masterlight are not the ones I uploaded.

Colin Clackson

Ok. Let's start again. I revisited your video on pedestal's and realized that I applied the pixelmath method of locating pixels with zero values to an integrated image rather than a calibrated image. There are indeed lots of zero-value pixels. I ran the WBPP script with the Output Pedestal Settings value at 500 but it didn't make any difference except the low rejection included a lot of back ground

BlockHead

Yeah, I am stuck on your dither. If you indeed did not have any dithering in the horizontal direction..this would definitely account for the issue. The amount of dither is perfect... but you need to randomly dither in both directions. Is one of your axis not communicating? (maybe you disabled a relay?)

Fixing the NB images with the pedestal will help you even if it is not the root of this problem.

-the Blockhead

Colin Clackson

I bllinked the raw images with a star at the vertical edge of the frame and it definitely doesn't wander horizontally. 

My guiding program is set to dither only in RA so maybe that's the issue, although I've never encountered the issue before and have always dithered only in RA. Do you think the nature of the filter is a contributing factor?

I really appreciate you taking the time with this.

BlockHead

No... I really think this is dither. When you dither in only one direction... if the pointing of your system is good.. you will have correlations in position (integer amounts of moment). When you later align on the stars... you will see shifted correlations between pixels (faintly). This is exactly what you see. I really think that random dithering... which really means both axis... will solve this issue.

-the Blockhead

Colin Clackson

Thanks Adam. I’m sure you’re correct. I wonder whether this data is salvagable if it is combined with additional data that has been dithered in both axes. Do you have any thoughts on that?

BlockHead

Absolutely. You will need to combine at least the same number of frames...(perhaps even just half as many, I am actually not certain). But I am certain that adding more data will definitely help.

-the Blockhead