Shifting Pixels

[Robert Q. Kimball]

I have just returned from the Okie-Tex Star Party with lots of nice new data to process.  While I was at the "party" a fellow astronomer showed me how he processed data using AstroArt.  In the process he did something that was very strange but seemed to really work nicely.

During the process, he created a master flat. Then he pulled up a screen and told me was going to shift the data 1/2 a pixel and thus remove and color issues in the master flat.  (I may not be explaining this correctly)  After the procedure the flat seemed smoother.

Two questions: Does this make any sense?  Can it be done in PixInsight?

Bob

[oldwexi]

Hi Robert!
Yes, the process ChannelMatch allows move of Pixels on subpixel level. (For example 0.1 Pixel or more)
I am using this for bringing in the DSLR/OSC flat the different color shoulders (in the raw Image)
more together to remove the Color imbalance

You simply move the X- and Y-Offset a little (below 0.5) and you have the
same result as with Astroart.

Gerald

[oldwexi]

Forgot to add the processes.
http://www.werbeagentur.org/oldwexi/processesforpixelshift.xpsm

This includes 3 Processes
ConverttoRGB
ChannelMatch
ConverttoGray

As ChannelMatch works only with RGB Images you have to Change the RAW Gray Bayerimage to a fake RGB -
than do the Channelmatch
Than convert this corrected Image back to RAW Gray Image

When usig this processes have the Histogramm open so you can watch the Progress how the histogram bumps come together..

Gerald

[msmythers]

Gerald

Wouldn't this need to be done to all images for the set. The master dark, master bias and all lights before any pre-processing. Are you correcting a cfa issue I would think? If so it's a hardware issue that would be present in all images. I don't know, just a thought.


Mike

[oldwexi]

Hi Mike!
The flats, with foils or Skyflat, usually have another Color imbalance based
on the Input medium. The idea is to neutralize the imbalance of the Input medium (Sky, foil).
The lights calibrated at the beginning with neutral channel masterflat will be Color calibrated later on.
I preprocess the masterflat before  to keep them as RAW gray flats showing the histobumps together instead of separated.
And use the treated Master for calibration of lights. No preprocessing of the raw lights before calibration.

Gerald

[Harry page]

Hi

In astroart you half shift pixels to destroy the bayer matrix ( hence it looks smoother )

However you must not do this in pixinsight , Pix requires the use of raw OSC flats for correct calibration

colour problems are resolved in processing

Harry

[Robert Q. Kimball]

Thank you Harry.  I have watched all your YouTube videos and enjoyed them. 

I guess the problem with Internet Forums is that when you ask a question you get convincing answers on both sides of the question.  Gerald seems to think it will work in PixInsight and has a process to do it.  You say not to do it.

I think for now I'm going to avoid this technique.

Bob

[Harry page]

Hi

I asked the question a few years back

http://pixinsight.com/forum/index.php?topic=1757.15


Harry

[msmythers]

Harry,

That is why I brought up the need to do the same to the lights. You generate a pixel error between flats and lights during calibration if you don't do both. I still don't understand the need to completely color balance a flat before calibration. Any color imbalance in the stacked image should be able to be corrected during regular post processing.


Mike

[Harry page]

Hi

It is my understanding that the colour of flats is generally irrelevant as as long as there enough signal in each channel ( so whitish normally works )

as the post said each channel is treated separately .

I have never had a problem that can not be sorted in processing

Harry