Why no IC5146 (Cyg) or NGC 6946 (Cen)?

[SteveBz]

Hi Guys,

Here are two calibrated single shots of IC5146 (Cyg) or NGC 6946 (Cen).  IC5146 (Cyg) is the Cocoon Nebula and NGC 6946 (Cen) is a galaxy.  If I amplify the images I can just make something out, but they seem amazingly dim.  Processing is as follows:

200 mm Newtonian (f/5) with PHD2 guiding and Nikon DSLR D5000.

1 NGC 6946) or 2 (IC 5146) light frames of 240 secs @ISO 1600.

61 x BIAS (ISO1600).

37 x FLAT (ISO1600).

No darks.

Then Noise subtraction and white balance using gimp.

But at no point can I see a nebula or a galaxy, unless I really ramp up the signal amplification, when everything is noise.

What's gone wrong?  Is there just too much light pollution in West Sussex?  The signal was sooo... faint, I spent a long time just working out if I'd even got the right location, it was just plate-solving against the wonderful Henry Draper catalogue that convinced me.

Let me know what you think.

Regards,

Steve.

[alacant]

Hi

I think the images are fine. You simply need more of them.

Whilst you should be able to record galaxies, the nebula may need the camera to be astro-modified.

Cheers and HTH

 

 

Edited October 24, 2019 by alacant

[Ronclarke]

You need 20 +stacked light frames then stretch the image .

Ron

[coatesg]

Well, you've no problem with the fields, but the issue here is the amount of exposure - you just need much more of it.

Is the D5000 modded or not? If not, you'll potentially lose a lot of sensitivity in the HA part of the spectrum which strongly affects the image of the Cocoon - this will make nebula photography quite hard for you. 

Even with a modded cam, only having 4 or 8 min is not enough to ensure the signal to noise ratio is high enough to drag the object out of the noise (thermal, read and sky background). If you continue to take lots of subframes (ensuring the histogram is not pinned to the left), and then reduce and stack those subframes, then you'll find the image starts to become easier to stretch and display. 240 or 300sec should be fine for subexposures - you need to ensure that the background noise swamps the read noise to get an optimal length for each individual sub exposure. 

I'd also run at 400-800ISO instead of 1600 to preserve Dynamic range - see: http://dslr-astrophotography.com/iso-values-nikon-cameras/ which suggests 400 or 800 as optimal. Any higher ISO, and you will get the same results by stretching in post-processing. 

Also, beware of not making the background too dark when processing the images - rule of thumb is RGB=21:21:21 (out of 256) as a target background - it prevents you losing the faint stuff into the blackness.

 

Good luck!

[SteveBz]

19 hours ago, alacant said:

Hi

I think the images are fine. You simply need more of them.

Whilst you should be able to record galaxies, the nebula may need the camera to be astro-modified.

Cheers and HTH

 

 

Thank you for plate-solving these.  I do have a spare Canon 350D, that I could get modified. I'll think about that.

Regards

Steve. 

[SteveBz]

Hi Graeme,

I thought I knew this stuff, but there is so much in your answer that eludes me.

18 hours ago, coatesg said:

If you continue to take lots of subframes (ensuring the histogram is not pinned to the left),

What does "pinned to the left" mean?  I try to take images for as long as I can, while not swamping the image with light pollution.  Like this:

18 hours ago, coatesg said:

you need to ensure that the background noise swamps the read noise to get an optimal length for each individual sub exposure.

Maybe this is a similar point.  How should I measure this?

18 hours ago, coatesg said:

I'd also run at 400-800ISO instead of 1600 to preserve Dynamic range

OK, I did try ISO800 the day before, but without much success.  I'll try again. It looks like Sunday may be a clear night.

18 hours ago, coatesg said:

Also, beware of not making the background too dark when processing the images - rule of thumb is RGB=21:21:21 (out of 256) as a target background - it prevents you losing the faint stuff into the blackness.

You're saying I shouldn't truncate the image too severely.  Here is my calibrated and stacked image:

You can just make out the galaxy.  I use the ink dropper tool to choose the darkest area of the stacked and calibrated image to create a new layer which I subtract from the noisy layer (eg the one that looks like the one above).  So if it's sort of light blue colour  like this (it looks more than 21:21:21):

Giving this:

Then I again use the dropper to create a white-ish layer like this:

To create this:

Which is what I posted.

18 hours ago, coatesg said:

reduce and stack those subframes

What does "reduced" mean?

Thanks for your help Graham,

Regards

Steve.

[coatesg]

"Pinned to the left" is a really poor way of saying (sorry...) that you need enough exposure to overcome the read noise in the sensor. If the exposures are too short, then the read noise will have a more significant influence. It can be shown that, if read noise is negligible, then (practically) you don't need a longer sub exposure, you just need more of them [you get nearly the same Signal To Noise from stacking equivalent total lengths of exposures vs a single long exposure, but without losing dynamic range at the top end].

Reduced = calibrated (ie with daks, flats) - ie to remove as far as possible instrumental effects. 

 

Read noise can be measured from bias frames (shortest possible exposure with lens cap on) - it's a statistical measurement of the variation of pixel values between frames. Ultimately, there's no hard and fast rule as to how much greater the background count should be, but I would say yours looks overexposed (though the raw data is needed to properly show it) - here's the histogram of the first image:

 

The data is well over to the right, probablymuch further than it needs to be, and there is clipping shown by the spike at 255. If the peak of the image is about a third of the way from the left, then you're probably fine - you've then enough data to swamp the read noise, yet you retain dynamic range in the brighter portion of the image. It'll depend on sensor, your location, your telescope. etc.

 

I am a little confused by you method of post-processing though. From the calibrated stack, you should just be able to use Levels, which allows you to "stretch" the data, and to set the background level appropriately. 

Something along the lines of this levels, but you'd use a 16bit image as source rather than a jpg, and there's some colour casts and gradients to remove, the background isn't dark enough (noisy because of starting with 8 bits, not enough sub exposures in the stack, etc, etc.)

Edited October 24, 2019 by coatesg

[SteveBz]

22 hours ago, coatesg said:

Something along the lines of this levels, but you'd use a 16bit image as source rather than a jpg, and there's some colour casts and gradients to remove, the background isn't dark enough (noisy because of starting with 8 bits, not enough sub exposures in the stack, etc, etc.)

Hi Graeme,

I'd never seen the stretch function before, great.  It's exactly what I want.  I'm not sure what you mean by colour cast, but I can do the stretch by channel and that means we don't have the bright blue effect.  But what about the gradient? Is there a gradient tool in GIMP?  How do I deal with that?



So now I can just see it, I'm sure some more lights will help.

Regards,

Steve.

[coatesg]

I haven't dealt with gradients in PS or Gimp in a long time - one way is to make a suitable greyscale gradient across the field using the gradient fill, apply as a mask and then levels to reduce the gradient. 

Others might have better explanations!