Improving this Rho Ophiuchi

[celeron787]

Hi again, requesting help improving on this image of rho ophiuchi.
I'm getting some harsh gradient towards the lower right corner. Default gradient wipe doesn't give me a uniform background unless I push it all the way up to 96%.
There's significant walking noise in this image as I did not do any dithering. Something that I will have to explore manually soon.

FITS file: https://www.dropbox.com/s/tax0orwdzywqpgz/1.FTS?dl=0

Sony A6000
Samyang 135mm F2.0
Sky-Watcher Star Adventurer
Tripod

Bortle 8
ISO400
Light frames - 384 x 15secs
Dark Frames - 30 x 15secs
Flat Frames - 35
Bias Frames - 50

Deep Sky Stacker
StarTools - 50% Bin, Autodev, Crop, Wipe, Life, Contrast, Decon, Color, Denoise

[admin]

Hi,

Would you have any idea as to what caused the gradient?
As you say, it is really quite extreme. I don't think I've ever seen a real celestial gradient that strong at this sort of FOV... Can you confirm there was an extremely bright light source towards that area of the sky? Was the moon out? Was there a street light just out of view? What does your dataset look like without flats calibration? Any better?

As for the walking noise, though dithering is obviously the way to go, you are in luck! The very latest 1.7 version introduces walking noise reduction.

[celeron787]

Hi,

Would you have any idea as to what caused the gradient?
As you say, it is really quite extreme. I don't think I've ever seen a real celestial gradient that strong at this sort of FOV... Can you confirm there was an extremely bright light source towards that area of the sky? Was the moon out? Was there a street light just out of view? What does your dataset look like without flats calibration? Any better?

As for the walking noise, though dithering is obviously the way to go, you are in luck! The very latest 1.7 version introduces walking noise reduction.

Well, the target was pretty close to the top of a high rise building, that might be the source of the gradient.
I restacked without flats and it seems like that corner gradient is gone. Also, I don't see as much walking noise as my initial stack.
But the amount of vignetting seems more difficult to remove compared to the initial corner gradient.

Does it mean it's a flat frame problem?
I took my flat frame using an EL panel directly over the lens, no T-shirts/diffusers.

Single flat frame: https://www.dropbox.com/s/x3vqvvyzlfvpc ... 8.ARW?dl=0
Masterflat from DSS: https://www.dropbox.com/s/7m9eld5mugr9q ... 0.tif?dl=0
Stacked FTS without flats: https://www.dropbox.com/s/m56a0g7dpmoegb2/test.FTS?dl=0

Autodev - restacked without flats.jpg (108.35 KiB) Viewed 5223 times

[admin]

Thank you for uploading all that.

I can see banding in the non-flat-calibrated dataset, indicating it is likely of low bit-depth. It may be because of a combination of the very short exposure times and median stacking? It trips up Wipe pretty severely, as it sees the bands as detail.

Your master flat seems very dark and requires much stretching - flats should be exposed with their histogram peak around half way. Because the master flat is made up of frames that have been exposed way too short, it is also very noisy. This noise signature then gets applied to every frame you stack, causing the walking noise. As an experiment, you can import the master flat into StarTools (do not activate Tracking!), go into the Layer module and apply a Gausian blur (~10px). This eliminates the noise. Now save this as your new master flat and stack with that. You should see no walking noise this time around.

As for the gradient, I cannot be 100% sure, but it appears the master flat and stack w/o flats are mismatched with regards to lighting unevenness (the vignetting is much more off-centre in the lights). Are you sure you did not move the camera between acquiring flats and shooting your lights?

Ivo

[celeron787]

Thank you for uploading all that.

I can see banding in the non-flat-calibrated dataset, indicating it is likely of low bit-depth. It may be because of a combination of the very short exposure times and median stacking? It trips up Wipe pretty severely, as it sees the bands as detail.

Your master flat seems very dark and requires much stretching - flats should be exposed with their histogram peak around half way. Because the master flat is made up of frames that have been exposed way too short, it is also very noisy. This noise signature then gets applied to every frame you stack, causing the walking noise. As an experiment, you can import the master flat into StarTools (do not activate Tracking!), go into the Layer module and apply a Gausian blur (~10px). This eliminates the noise. Now save this as your new master flat and stack with that. You should see no walking noise this time around.

As for the gradient, I cannot be 100% sure, but it appears the master flat and stack w/o flats are mismatched with regards to lighting unevenness (the vignetting is much more off-centre in the lights). Are you sure you did not move the camera between acquiring flats and shooting your lights?

Ivo

Thanks ivo! I use Aperture Priority mode on my camera for my flat exposures. My flat was about 1/3 from the left in the histogram, that's usually what people recommend. But I'll try halfway the next time.

The masterflat is what DSS churned out. It's always very dark in all my flat frame stacks. Is it also a DSS setting?

I did shift the camera to face the lens up so I can put the EL panel over it. But the imaging train remained intact throughout the whole session. I thought it was alright to shift the position of the imaging train as long as the focus remains the same.

[admin]

Thanks ivo! I use Aperture Priority mode on my camera for my flat exposures. My flat was about 1/3 from the left in the histogram, that's usually what people recommend. But I'll try halfway the next time.

The masterflat is what DSS churned out. It's always very dark in all my flat frame stacks. Is it also a DSS setting?

I did shift the camera to face the lens up so I can put the EL panel over it. But the imaging train remained intact throughout the whole session. I thought it was alright to shift the position of the imaging train as long as the focus remains the same.

Hmmm... It sounds like you did everything right! AP mode, 1/3 from the left should be fine too. For the flats, see if the blur helps with the walking noise. To clarify; you can obviously move the scope + camera + etc. as a single unit carefully just fine, but you should not, for example rotate the camera.

All that said, under heavily polluted Bortle 8 skies, it's entirely possible the gradient is just a real gradient from a local light source like the building you mentioned.

In that case, what you already managed in terms of gradient reduction is indeed the way to go; a very high aggressiveness will be needed to mitigate the extremely steep gradient.
Try Aggressiveness 94%, Drop Off Point 0%, Corner Aggressiveness 98% and Precision 1024x1024.

[celeron787]

Thanks ivo! I use Aperture Priority mode on my camera for my flat exposures. My flat was about 1/3 from the left in the histogram, that's usually what people recommend. But I'll try halfway the next time.

The masterflat is what DSS churned out. It's always very dark in all my flat frame stacks. Is it also a DSS setting?

I did shift the camera to face the lens up so I can put the EL panel over it. But the imaging train remained intact throughout the whole session. I thought it was alright to shift the position of the imaging train as long as the focus remains the same.

Hmmm... It sounds like you did everything right! AP mode, 1/3 from the left should be fine too. For the flats, see if the blur helps with the walking noise. To clarify; you can obviously move the scope + camera + etc. as a single unit carefully just fine, but you should not, for example rotate the camera.

All that said, under heavily polluted Bortle 8 skies, it's entirely possible the gradient is just a real gradient from a local light source like the building you mentioned.

In that case, what you already managed in terms of gradient reduction is indeed the way to go; a very high aggressiveness will be needed to mitigate the extremely steep gradient.
Try Aggressiveness 94%, Drop Off Point 0%, Corner Aggressiveness 98% and Precision 1024x1024.

Sorry, quick question. For applying the Gaussian Blur, I import in the masterflat -> Select Linear -> Click track, no noise reduction -> Layers -> Gaussian filter 10px. Is that correct?
How do I stack using the masterflat in DSS? DSS isnt reading the edited masterflat properly and is giving me some compatibility error.

[admin]

Sorry, quick question. For applying the Gaussian Blur, I import in the masterflat -> Select Linear -> Click track, no noise reduction -> Layers -> Gaussian filter 10px. Is that correct?

Yes, that should work.

How do I stack using the masterflat in DSS? DSS isnt reading the edited masterflat properly and is giving me some compatibility error.

Hmmm.... That's odd. It shouldn't really be a problem. Any other message? Perhaps it doesn't like the color (3-channel) encoded file? Try something like Imagemagick (or photoshop) to save the file as a 16-bit grayscale file?
It'd be interesting to find out why the flats are noisy like they are though.

[celeron787]

Sorry, quick question. For applying the Gaussian Blur, I import in the masterflat -> Select Linear -> Click track, no noise reduction -> Layers -> Gaussian filter 10px. Is that correct?

Yes, that should work.

How do I stack using the masterflat in DSS? DSS isnt reading the edited masterflat properly and is giving me some compatibility error.

Hmmm.... That's odd. It shouldn't really be a problem. Any other message? Perhaps it doesn't like the color (3-channel) encoded file? Try something like Imagemagick (or photoshop) to save the file as a 16-bit grayscale file?
It'd be interesting to find out why the flats are noisy like they are though.

Ok, I tried a few things, all failed to stack except for the original master flat.
Here's a screenshot of how DSS is reading these files.

Capture.JPG (42.1 KiB) Viewed 5138 times

1. Original masterflat

2. Linear source in Startools -> Gausian Blur -> Save

3. Open in Photoshop -> Gausian Blur -> Save

4. Non-linear source in Startools -> Gausian Blur -> Save

[admin]

Ok. Looks like DSS might encode some special fields to indicate a file is a Master Flat.
However, maybe you can "stack" the modified master flat as if it is a stack of just one flat frame, so that DSS makes a master flat out of that?