SGP and PixInsight

[sidelight]

Hi All,

Been a long while since me being home and clear skies have happened at the same time, in the meantime I have been playing with my software setup to make sure I understand as much as I can before I get out again.  I bought SGP earlier this year and have a problem and wondered if anyone else has come up against it.  

When I take a picture in SGP and save them as FITS format with the modded 450D and process it in Pixinsight it appears the red and blue channels are swapped in the resulting image.  Using the CR2 images and all is well.  I tried the bottom up option in PI but that didn't work, so as far as I can tell either SGP is making a mistake with the raw image and saving the red as the blue, or the image is 180 degrees out, anyone know which?   As the camera is in a cooler box so taking everything apart to put a lens on will take a while.

I wanted to use the fits images as that way SGP can populate fits headers nicely as it doesn't seem to play well with raw files from the perspective of plate solving etc.

Cheers

Ross

[glowingturnip]

Hi,  I don't know anything about SGP, but it sounds like an incorrect debayering setting in Pixinsight to me ?

The batch preprocessing script has a couple of debayer settings - first one should normally be set to VGA, and the second one has four choices, eg RGBG, GRBG etc etc.  Play around with each of those, on a daytime scene if you can, and you'll probably find that one of them is right, and the others look strangely wrong.  the CFA button, over on the right of the box should be checked.

[Earl]

Best place to ask will be on the SGP forum as they are very active, then The PI forum if its down to the Debayer settings.

[sidelight]

Thanks all,

PI is fine if I use raw images its just when using the fits files, I am trying to avoid saving both fits and raw images, so my money is on the way SGP is saving the files, I will try their forum.  I can get the correct cast when using BGGR as a matrix pattern, or rotating the image 180 degrees prior to debayer.  Both have the same effect, I  just want to ensure I am processing the red part of the bayer matrix as red etc.  

Cheers

Ross

[glowingturnip]

aha, sounds like the bayer offset is out by 1 pixel if flipping it 180 degrees works.

Suppose your camera matrix goes

GRGR

BGBG

GRGR

etc, but your debayering starts from row 1 rather than row 0, then the debayer routine sees

BG

GR

which debayers properly if you use BGGR

but when you flip it 180 degrees, you'll have

GBGB

RGRG

GBGB

and when you lose the first row you get

RG

GB

which will debayer properly with RGGB

Anyway, I'm not aware of a setting in Pixinsight for setting starting row offsets for debayering, though I do know that Maxim DL has one (you can set row 0 or row 1 and col 0 or col 1). Maybe there's a setting to do the same thing in SGP ?

[sidelight]

 Thanks Stuart,

I hadn't considered a row offset, and couldn't find one in PI.  I will reach out on the SGP forum although that appears to be down at the moment.

Cheers

Ross

[lensman57]

Hi All,

Been a long while since me being home and clear skies have happened at the same time, in the meantime I have been playing with my software setup to make sure I understand as much as I can before I get out again.  I bought SGP earlier this year and have a problem and wondered if anyone else has come up against it.  

When I take a picture in SGP and save them as FITS format with the modded 450D and process it in Pixinsight it appears the red and blue channels are swapped in the resulting image.  Using the CR2 images and all is well.  I tried the bottom up option in PI but that didn't work, so as far as I can tell either SGP is making a mistake with the raw image and saving the red as the blue, or the image is 180 degrees out, anyone know which?   As the camera is in a cooler box so taking everything apart to put a lens on will take a while.

I wanted to use the fits images as that way SGP can populate fits headers nicely as it doesn't seem to play well with raw files from the perspective of plate solving etc.

Cheers

Ross

Check the Bayer pattern first. 

A.G

[sidelight]

Hi A.G.

The bayer pattern for the Canon is RGGB, the only way it works (or appears to) is using a BGGR bayer pattern in PI to debayer.  I can confirm this by taking some terrestrial pictures, and to be honest if SGP is mixing up the channels when it saves the fit file its not a big deal, if its a problem with SGP then I am sure it will be sorted fairly quickly by its devs, and I have a workaround in the meantime.  I just want to ensure I am processing the channels properly 

Cheers

Ross

[lensman57]

Hi A.G.

The bayer pattern for the Canon is RGGB, the only way it works (or appears to) is using a BGGR bayer pattern in PI to debayer.  I can confirm this by taking some terrestrial pictures, and to be honest if SGP is mixing up the channels when it saves the fit file its not a big deal, if its a problem with SGP then I am sure it will be sorted fairly quickly by its devs, and I have a workaround in the meantime.  I just want to ensure I am processing the channels properly 

Cheers

Ross

This sounds like SGP is assigning the wrong values to the channels, hope you get a bug fix soon. Why don't you try APT? It does save as CR2  but PI can deal with that efficiently and it also has a lot more to offer.

Regards,

A.G