Author Topic: Cant remove hot pixels using CosmeticCorrection, especially lot remains on R ch  (Read 168 times)

Offline axlns

  • Newcomer
  • Posts: 4
    • View Profile
Hello. I'm evalating PixInsights and I'm impressed by amount of features. really considering buying it after my trial is over.
I'm newbie, obviously, but I'm reading Warren A. Keller's book and that makes things easier to understand.

I however have an issue with removing hot pixels, I cannot get rid of all of them, and in fact I'm having lot of hot pixels after I apply CosmicCorrection to my lights. Especially lof of them in red channel.

I shoot using DSLR Sony a99 camera.

So I followed this process (just followed Keller's book)

1. Created master bias

2. Created master dark

3. Created superbias

4. Calibrated my flats with superbias and master dark

5. Now trying to get rid of hot pixels using CosmeticCorrections. I used master dark and auto-detect, and I cant get rid of all of the hot pixels, I think I tried almost every possible combination.
This is what I see on resulting image with corrections applied - https://www.screencast.com/t/1daxOKPdI
here I outlined problematic pixels   https://www.screencast.com/t/qk0Y329xHLp8
you will notice that these problems have the same patterns, it looks like it is always 2 pixels next to each other located on diagonal, in one or another orientation, here is closer look - https://www.screencast.com/t/tRRWxHhmaV

This is how my calibrated light image looks before corrections - https://www.screencast.com/t/TI5QmhE9
This is how it looks after - https://www.screencast.com/t/qr9fu0ANH8

After debayering - I see that most of these issues are on Red channel, but some blue hot pixels also presented - https://www.screencast.com/t/WxgQt2gwM6
This is entire debayered image - https://www.screencast.com/t/0M7F1bOn
This is red channel image - https://www.screencast.com/t/DEC9x6ei14
and closer look on hot pixels on red channel - https://www.screencast.com/t/UP3ykQGUkI

Maybe this will give someone a clue what I'm doing wrong? I don't think I should have this level of hot pixels after CosmeticCorrection applied, right?
I appreciate ANY help!

The problem with these hot pixels is that PixInsight seems to detect them as stars.

I extracted separate channels as I want to star align all of them separately and then stack separately and create artificial L channel as well.

now when I try to measure my individual images using SubframeSelector I see in the console that it detects lot of stars (like 1250)
I think will cause me lot of problems when I will to star align these images.

So, guys, I'm really need help here. What I'm doing wrong?



Offline dave_galera

  • PixInsight Addict
  • ***
  • Posts: 153
    • View Profile
    • QDigital Astro
One thing I have notices on the following image you posted:
       and closer look on hot pixels on red channel - https://www.screencast.com/t/UP3ykQGUkI

is that the stars are elongated in a bottom-left to top-right direction which is very similar direction to doubling of the hot pixels.
Dave

Offline axlns

  • Newcomer
  • Posts: 4
    • View Profile
One thing I have notices on the following image you posted:
       and closer look on hot pixels on red channel - https://www.screencast.com/t/UP3ykQGUkI

is that the stars are elongated in a bottom-left to top-right direction which is very similar direction to doubling of the hot pixels.
thanks Dave, for your input.
I think elongated stars are caused either by scope adjustment or by coma issue
here is screenshot of the central part of the frame, stars are more or less rounded - but hot pixels looks same - https://www.screencast.com/t/f57rOikMp43
this is red channel of debayered light frame

Offline axlns

  • Newcomer
  • Posts: 4
    • View Profile
Some more screenshots, not sure I understand whats going on, and if this is expected or I either doing something wrong or I have problem with my camera

This is non-calibrated, non-cosmetically-corrected light frame, debayered - https://www.screencast.com/t/g18HEdzEV
this is calibrated, cosmetically corrected light frame, debayered - https://www.screencast.com/t/8rpXxn3J8

notice how they having same looking hot pixels but different color, why?

this is calibrated, corrected, debayered light frame, channel R - https://www.screencast.com/t/BCsAIwaT
this is calibrated, corrected, debayered light frame, channel G - https://www.screencast.com/t/LHOSH3XbB
this is calibrated, corrected, debayered light frame, channel B - https://www.screencast.com/t/Fx9RvnM8b

note that on channel G and B these hot fixes are exactly 1 pixel, but on channel R they really look strange..

maybe somone know whats happening here?

one more thing. I'm processing DSLR RAW images, which are not debayered yet, and the book saying I need to select CFA option in CosmeticCorrection
I'm getting much more hot pixels when I select CFA - https://www.screencast.com/t/A5YZh1QY
and this is how it look when CFA is not selected - https://www.screencast.com/t/mICxj590
as you can see when CFA is not selected I'm getting much better results in terms of hot pixels.
maybe I should not check that option? even so it says I should - https://www.screencast.com/t/6nRcYhF8uZP

update:
so I decided to keep going, I aligned frames and integratred them, hoping hot pixels will be removed in the final image, unfortunately that did not happned
I attached result. all those hot pixels are presented

when I used DSS to stack images - I did not have this issue, no hot pixels on final image.

so I really want to know what I'm doing wrong, I really want to keep using PixInsight. So any help is appreciated
« Last Edit: 2019 September 04 09:51:13 by axlns »

Offline axlns

  • Newcomer
  • Posts: 4
    • View Profile
I have uploaded 7z archive with
  • few lights frames (not touched, just as they were from the camera)
  • master dark
  • master flat
  • master and super bias
here is the link to the archive - https://www.sugarsync.com/pf/D7775956_09685078_23158

I'd appreciate if someone will try to check these files in Pix and share his ideas what I'm doing wrong. thanks!