PixInsight + Orion StarShoot DScI's + Debayering

[GvnrRickPerry]

something is funny here - after debayering the R and G channels still look undebayered in the Y direction. the B channel looks fine.

can you try a different capture software?

rob


edit: whoops, posted late. to be honest i don't think the debayering is correct in the other software either, the image still has a "screen door" look to it.

Hey Rob!

I guess you did see that I tried to use a different capture software last night and got kind of a different (but the same) issue? Yes, I agree that something still doesn't look right in the "correct" image, but I just chalked that up to noise and errors in my capturing process/guiding/etc. The final image, once it's stacked and denoised, etc., doesn't look too bad (but there's definite room for major improvement!) - you can see the final version below. There are about 36 subs at 2 minutes each, with a few 5 minute subs mixed in with it. I captured a bunch of 10 minute subs last night, and really wanted to integrate them into my final, but alas I cannot because of the debayering issues I'm now facing! I'm up against a wall here and don't know what to do! 

[GrzegorzP]

Hi Duane

I don't know this camera, but if it is this model:

http://astronomy.lt/index.php?main_page=product_info&products_id=202

it uses Sony ExView HAD 429AKL  sensor. According to this specification:

http://www.npk-photonica.ru/images/icx429akl2.pdf

429AKL has Ye, Cy, Mg, G color mosaic filters on chip. I am afraid any debayer method imlemented in Pix is not suitable.

Grzegorz

[Juan Conejero]

That's correct. CYGM mosaiced images are not supported by our current tool set.

[GvnrRickPerry]

Hi Duane

I don't know this camera, but if it is this model:

http://astronomy.lt/index.php?main_page=product_info&products_id=202

it uses Sony ExView HAD 429AKL  sensor. According to this specification:

http://www.npk-photonica.ru/images/icx429akl2.pdf

429AKL has Ye, Cy, Mg, G color mosaic filters on chip. I am afraid any debayer method imlemented in Pix is not suitable.

Grzegorz

Hello Grzegorz,

Thanks for the info! You know, I was wondering if it used that weird debayer pattern... I am looking in my manual right now though and mine uses a different chip - my manual says it's the Sony ExView HADTM ICX429AKL Color

Does the HAD"TM" make a difference, or does that chip still use the CMYK debayer method?

Also, my other application has a CMYK debayer option that I tried on the images captured by my new software, and they're not debayering correctly using CMYK... I figured maybe there's an offset or something???

Let me know if this helps me out at all! I'm hoping my chip doesn't have the CMYK debayer filter!
Thank you!
-Duane

[pfile]

the "HADTM" is likely a cut and paste error. i think that says "HAD (tm)" as in trademarked.

in your other app it's possible that the CMYK setting is "backwards" (like the difference between RGGB and BGGR). are there multiple options?

rob

[GvnrRickPerry]

the "HADTM" is likely a cut and paste error. i think that says "HAD (tm)" as in trademarked.

in your other app it's possible that the CMYK setting is "backwards" (like the difference between RGGB and BGGR). are there multiple options?

rob

Makes sense.

You might be onto something there with the "CMYK being backwards" - I can't find a way to change that though... Here's the manual debayer settings screen I'm prompted with (so you know what options I've got here):

[msmythers]

Just thinking out loud but could this be handled as separate channels by splitting the cfa to the 4 components(CMYK) images using the SplitCFA tool. Do your pre-processing on each channels just like they are from a mono ccd with each filtered channel and then use PixelMath to convert to RGB. The conversion to RGBH looks straight forward from what I've seen doing a simple google search.


Mike

[GvnrRickPerry]

Just thinking out loud but could this be handled as separate channels by splitting the cfa to the 4 components(CMYK) images using the SplitCFA tool. Do your pre-processing on each channels just like they are from a mono ccd with each filtered channel and then use PixelMath to convert to RGB. The conversion to RGBH looks straight forward from what I've seen doing a simple google search.


Mike

Oh good stuff! I tried that just now, but am getting this error in the console:
Wrong image dimension: odd width [752] or height [581]

I'm assuming this is because the pixels are not square... Is there any way to make the pixels square in PI?

[Carlos Milovic]

To fix the error, you may try deleting one row (with the Crop process), either at the top or the bottom. See which one works better.

To fix the non-square pixels, you should deform the image to make it square. For that you'll need to use the Resample process. Unchech the preserve aspect ratio option.

[pfile]

the "HADTM" is likely a cut and paste error. i think that says "HAD (tm)" as in trademarked.

in your other app it's possible that the CMYK setting is "backwards" (like the difference between RGGB and BGGR). are there multiple options?

rob

Makes sense.

You might be onto something there with the "CMYK being backwards" - I can't find a way to change that though... Here's the manual debayer settings screen I'm prompted with (so you know what options I've got here):

well, you can always rotate the image by 180 degrees before you attempt to debayer... or maybe it's just flip over the X axis - i can never keep that straight. essentially i'm talking about the difference between reading a fits file in using up-bottom vs. bottom-up.

rob

[msmythers]

I looked at the Sony ExView HAD 429AKL pdf referenced by Grzegorz and on page 12 it shows the matrix for the sensor. SplitCFA will not work for this sensor.


Mike

[pfile]

yeah - in fact the pixels seem to get added together in hardware. i think this chip may have been designed to be compatible with some kind of television color signaling? seems to be focused on Y and chroma difference.

rob

[msmythers]

Rob


Probably low light surveillance cameras was the original target. That's such a massive market.



Mike

[GvnrRickPerry]

Hello all,

I've been getting things to work with capturing and debayering with the other software... Just got my larger camera back this week, but now there's a hurricane about to pass through our area (it'll be a tropical storm by the time it gets to us in Texas) so that kind of sticks me inside for a few days. I'm about to take apart my CGE mount to try and fix some noise that's been happening inside it... I'm a little worried.

I did notice a new script - Batch CMYG Debayer - I played around with it a bit and couldn't get the settings "perfect", but I bet I could if I played around with it enough. It's kind of difficult to use as a batch script, but I'm not sure what process it links back to so I can play with it manually (or if it's even possible to do that).

I'm about to start an adventure in OSC + Ha testing when it clears back up again, so this will probably go on the back burner for a while.

Thank you everyone for your help!
-Duane