https://www.dropbox.com/sh/f7dh6khukbmj ... _MA_a?dl=0
Link to 21 CR2 files as they came from the camera.
Maybe Mike , if you feel up to it , you can evaluate...
So these are pure, no processing at all on them
Crazy idea or not HOO approaches Soul neb
-
Mike in Rancho
- Posts: 1166
- Joined: Sun Jun 20, 2021 10:05 pm
- Location: Alta Loma, CA
Re: Crazy idea or not HOO approaches Soul neb
Hi Freddy,
So I fiddled around with your files on and off all day and then finished up tonight. Interesting stuff. Of course I don't have dcraw and wasn't sure about any Windows downloads I saw out there, but managed to make due. Sort of.
First I used ASTAP to convert them all to FITS, which it does by default. Unfortunately, that first iteration is still considered CFA. I thought no big deal, since in DSS you can simply uncheck demosaicking in the FITS settings. But no! Seems DSS will outhink you, and if the FITS says there is a CFA, you are getting color whether you want it or not.
I also tried a few grayscale conversions, which actually helped me find some things in the data, but still not optimal as that is post-debayering anyway.
So I went back to ASTAP, which actually follows your instructions if you tell it to turn off debayering, and stacked your CR2's there. The result properly comes out as mono/gray.
A few things, in case you were interested: Files 9800 and 9820 are not full 600s exposures, so I discarded them. Also, 9816 has the satellite trail.
Purely by accident, when doing single conversions manually, I broke the session into chunks and did some mini stacks. I also flipped through things to track your dithering, since if you let the hot pixels come though the pattern doesn't seem totally random.
Each set was just chosen by number, but I think the law of random number generation was working against you, as they maybe imparted some pattern movement. Notice that in each stack of 6 there are streaks that go different ways - both shadow streaks as well as other patterning. Not sure how exactly they are smearing about this way.
Oh and there's also Canon horizontal banding in the mix, which I'm sure doesn't help.
So I fiddled around with your files on and off all day and then finished up tonight. Interesting stuff. Of course I don't have dcraw and wasn't sure about any Windows downloads I saw out there, but managed to make due. Sort of.
First I used ASTAP to convert them all to FITS, which it does by default. Unfortunately, that first iteration is still considered CFA. I thought no big deal, since in DSS you can simply uncheck demosaicking in the FITS settings. But no! Seems DSS will outhink you, and if the FITS says there is a CFA, you are getting color whether you want it or not.
I also tried a few grayscale conversions, which actually helped me find some things in the data, but still not optimal as that is post-debayering anyway.
So I went back to ASTAP, which actually follows your instructions if you tell it to turn off debayering, and stacked your CR2's there. The result properly comes out as mono/gray.
A few things, in case you were interested: Files 9800 and 9820 are not full 600s exposures, so I discarded them. Also, 9816 has the satellite trail.
Purely by accident, when doing single conversions manually, I broke the session into chunks and did some mini stacks. I also flipped through things to track your dithering, since if you let the hot pixels come though the pattern doesn't seem totally random.
Each set was just chosen by number, but I think the law of random number generation was working against you, as they maybe imparted some pattern movement. Notice that in each stack of 6 there are streaks that go different ways - both shadow streaks as well as other patterning. Not sure how exactly they are smearing about this way.
Oh and there's also Canon horizontal banding in the mix, which I'm sure doesn't help.
-
- 9801-06.jpg (161.56 KiB) Viewed 3432 times
-
- 9807-12.jpg (161.25 KiB) Viewed 3432 times
-
- 9813-19 skip 16.jpg (158.17 KiB) Viewed 3432 times
-
Mike in Rancho
- Posts: 1166
- Joined: Sun Jun 20, 2021 10:05 pm
- Location: Alta Loma, CA
Re: Crazy idea or not HOO approaches Soul neb
And here's the entire stack, again stretched out to show what's there plus the banding. Things are getting better, but some of those opposing patterns are still baked in.
This was kappa 2.0 too, and the satellite trail is still somewhat there.
Of course these stacks are lights only, so they start out a bit bright (no dark or bias subtraction) and more importantly no flats.
It would still be good to examine some flats and perhaps a daytime shot or two of something with a lot of contrast, to really make sure there are no defects from the mono conversion.
Curious also, did they put new firmware into the T3i for you? Kinda wondering since I'm presume factory stock Canon will scale what it thinks are the R and B pixels when it writes out the CR2s. Maybe not.
I was also wondering what your bias level is. I don't have experience with Canons, let alone what ISO800 would do to it, but you might have room to lower the exposures, maybe to 6 minutes?
That's one thing I would try for the next time out -- 18 max pixels should be fine for dithers (and you've already walked away by more than 40 pixels just in these few subs), but I don't think you have enough of them to really start washing out all these movements.
Anyway that's my interim prescription for now. Shorter subs, allowing for more dithers, and longer total integration, also allowing for more dithers. That way you'll get both better SNR plus the pattern noise should start evening itself out and eventually go away. I hope.
The Canon banding I don't know anything about though.
EDIT: Oh you might also check things like your settle down time, to make sure you haven't started taking the next image while the mount might still be smearing things about a little bit.
-
- 9801-9819 ALL.jpg (208.65 KiB) Viewed 3432 times
This was kappa 2.0 too, and the satellite trail is still somewhat there.
Of course these stacks are lights only, so they start out a bit bright (no dark or bias subtraction) and more importantly no flats.
It would still be good to examine some flats and perhaps a daytime shot or two of something with a lot of contrast, to really make sure there are no defects from the mono conversion.
Curious also, did they put new firmware into the T3i for you? Kinda wondering since I'm presume factory stock Canon will scale what it thinks are the R and B pixels when it writes out the CR2s. Maybe not.
I was also wondering what your bias level is. I don't have experience with Canons, let alone what ISO800 would do to it, but you might have room to lower the exposures, maybe to 6 minutes?
That's one thing I would try for the next time out -- 18 max pixels should be fine for dithers (and you've already walked away by more than 40 pixels just in these few subs), but I don't think you have enough of them to really start washing out all these movements.
Anyway that's my interim prescription for now. Shorter subs, allowing for more dithers, and longer total integration, also allowing for more dithers. That way you'll get both better SNR plus the pattern noise should start evening itself out and eventually go away. I hope.
The Canon banding I don't know anything about though.
EDIT: Oh you might also check things like your settle down time, to make sure you haven't started taking the next image while the mount might still be smearing things about a little bit.
Re: Crazy idea or not HOO approaches Soul neb
Phew Mike thanks , that is allready quite an investigation you did!
To start with i was not too picky on my lights meaning i just stacked them all. As this actually was a kind of experiment because i used the mono cam only for the second time.
I took 10 minutes frames because the Elf advice me that, Dubbelderp does this also. The Elf even takes 15 min subs thesame camera.
I believe he also mention that about 12 subs should be just enough for dithering...
I don't understand well the correlation between the parameter settle time in APT and the actually settle time. I see that APT reports sometimes 20 to even 35 seconds it takes for dithering...this might indeed cause perhaps issues...
EDIT : seems a wait time..in sec, guess i might take a few seconds more cause now i believe this setting is a bit on the low side.
But the more and more this stories goes further the more and more it seems something is going on with the dithering procedure...
Banding, well yeah Canon is a bit notorious for that,less on the 800D but more on the 600D/T3i, i guess you just cannot stretch it too far or it might appear.
To start with i was not too picky on my lights meaning i just stacked them all. As this actually was a kind of experiment because i used the mono cam only for the second time.
I took 10 minutes frames because the Elf advice me that, Dubbelderp does this also. The Elf even takes 15 min subs thesame camera.
I believe he also mention that about 12 subs should be just enough for dithering...
I don't understand well the correlation between the parameter settle time in APT and the actually settle time. I see that APT reports sometimes 20 to even 35 seconds it takes for dithering...this might indeed cause perhaps issues...
EDIT : seems a wait time..in sec, guess i might take a few seconds more cause now i believe this setting is a bit on the low side.
But the more and more this stories goes further the more and more it seems something is going on with the dithering procedure...
Banding, well yeah Canon is a bit notorious for that,less on the 800D but more on the 600D/T3i, i guess you just cannot stretch it too far or it might appear.