Jump to content

Banner.jpg.b89429c566825f6ab32bcafbada449c9.jpg

pixinsight settings for cooled camera


Recommended Posts

My new ASI 071MC Pro saw first light yesterday. I did a quick setup with the asiAir and got my setup working in no time. Wow! 

So I started taking some lights on a couple of objects, and when the moon started rising, I added my new dual band filter (Ha - OIII), and get my first narrowband images. 

 

BUT... My Pixinsight must still be set on CFA images (I used to shoot with dslr) - because I keep seeing some sort of bayer grid in muy pictures, and the color seems off, too. 

I have debayering set at RGGB, and VNG algorithm. 

 

Here's 30 x 60s on M42... Can anyone tell me what settings I need to adjust? 

(I still need to tweak the right settings for my flats, so please don't mind the vignetting)

433926317_M42HA0IIItest.thumb.jpg.470a37867ed8df920520b492455a4f94.jpg

Link to comment
Share on other sites

I'm rather confused with this image of yours.

You say that it is integration of 30 subs? What sort of alignment do you use (if any) with PI? Do you dither?

Even if you don't dither, this sort of effect can show up in stack only if you have perfect guiding and no dither or for some reason you opted to do full pixel alignment (no fractional pixel shifts) and each sub is shifted by even number of pixels (I know, it's strange description, isn't it?)

Back on the topic. What are you trying to accomplish? Use OSC/CFA camera for mono with dual band nb filter or use it like regular color camera?

In first case - no need to debayer / demosaic image. You should treat it as mono. It would be good to apply flats - this will remove bayer pattern with narrow band filters. You can bin your subs - this will improve SNR. There are other techniques to further boost SNR of such images (inherently R, G and B pixels have different response to NB and therefore register different levels of signal - and have different SNR for same signal - simple stacking is not the best way to handle this, nor classic weighted stacking - that one expect SNR to be uniform across sub)

In second case - just process it like you would DSLR - regular OSC image. You will certainly have very strange color this way, but you can color balance it to proper red/blue combination for nebulosity (OIII and Ha) - like bi color version. Star color will be lost however. They will contain mostly red and blue and combinations of those colors (magenta and such).

What was your exact work flow to get bayer type pattern in resulting stack?

 

Link to comment
Share on other sites

The setup was not aligned perfectly. I have no fixed setup. Guiding was done with the phd part of the asiair app. The dec and ra lines stayed within 0 and 2". I did not dither this session.

When I debayered the image in the app, it showed a gorgeous red nebula. As one would expect. 

The workflow I used was the same as I do with my dslr images. Only here, I used a matching master dark, and my flats are not yet on point, I used 1s exposures and I'll possibly need 10s to get my histogram to 50%. 

Temp: -10, gain: 90.

batch preprocessing script - staralign the registered and debayered lights, imageintegration. 

I changed a setting in the cfa (raw) image settings, and did a manual debayer, now I did get color, albeit green and a slight hint of pink.. But still the same lines pattern. 

Link to comment
Share on other sites

If you have mostly green image - your bayer pattern setting is wrong. With dual nb filter you should get mostly red and blue.

Changing order of bayer pattern can swap colors and since pixels in bayer matrix have red and blue on one diagonal, and green on other - changing pattern changes diagonal - so what was to be red and blue gives most of the green.

 

Link to comment
Share on other sites

7 minutes ago, Wiu-Wiu said:

I'll have to look into the selective color calibration, that way I could get more "hubble" looking pictures. This combo looks a bit strange :) 

Can you do something like this using pixel math in PI:

Split each sub after calibration into 4 components by having each image have "every other pixel" in it?

First component would have pixels from 1,3,5 in both X and Y. Second component would have 1,3,5 in X and 0,2,4 in Y. Third component same as second but X and Y swapped and last component all even indices?

If you can do that, there might be a better way to do bi color image from this.

Resulting resolution of the image will be halved, but this is what I would suggest:

Stack only R sub pixels (or component made of only red pixels) into Ha image.

Stack B and G pixels (or again 3 components made of blue pixels and two times green pixels - each one in their own sub image) to OIII stack - this time be sure to use weighted average in PI - each sub having contribution based on SNR.

 

Link to comment
Share on other sites

3 minutes ago, Wiu-Wiu said:

Oh, wow, if there ever was any doubt in why math is more important than art, this is it :) ? 

I'll have to do some serious reading about that. I'm sure it makes sense, I'm just not sure I know how to do that :) 

Maybe it does not need to be that complicated. Do you have access to following features:

- Super pixel debayering?

- Split into channels

Almost same thing as above can be made using:

Calibrate frames then use super pixel mode to debayer - this should get you bunch of color subs each with twice less pixels in X and Y axis. Extract red channel and save as mono images. Do the same with green and blue channels. If you can split into channels - that is same thing.

Stack red channel subs to one stack

Stack both blue and green channel subs to another stack but make sure you: 1. equalize frames, 2. use weighted average (based on SNR)

You should get pretty much the same result as above with red stack being Ha and green + blue stack being OIII

Link to comment
Share on other sites

With your dual NB filter, you need to split channels. The problem is that if you do "normal" debayering, you spill over colour into other pixels. Ie, green pixels will receive red from averaging nearby red pixels, and red pixels will receive green and blue from nearby green and blue pixels, etc.

As Vlaiv wrote, try super pixel debayering. This way, the red information from Ha will only come from red pixels, and the teal information from Oiii will come only from green and blue pixels. There will be no interpolation.

Also, have you seen this?

https://pixinsight.com/forum/index.php?topic=7184.msg71479#msg71479

Link to comment
Share on other sites

I gave the super pixel debayering a go, but it didn't come out better than what I posted earlier. Could be blamed on my flats, though. Vignetting was through the roof, and as it was my first try, I only took flats of 0,5 and 1s, but I think I need 10s flats with this camera at these settings. 

The link is handy for the settings, but I'd have to read through it a couple of more times to know what everyone is talking about, as there are so many angles of input there. 

tbc! 

Link to comment
Share on other sites

4 hours ago, Wiu-Wiu said:

Vignetting was through the roof, and as it was my first try, I only took flats of 0,5 and 1s, but I think I need 10s flats with this camera at these settings. 

Adjust the exposure such that the average intensity (adu count) is in the mid range of your sensor. For 16 bit, about 30 000. Use a low gain setting, so that exposure time isn't too short. I usually set gain to 0. Also take dark flats to calibrate your flats.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.