Mark's Tuesnoad Horsehead

[MarkS]

Here's my Horsehead taken at Tuesnoad:

Canon EOS300D on a C11 (unguided)
180 x 30sec subs stacked (with flats & darks)
2x2 binned 3 times times in succession to reduce size


Here's a single 30sec sub:


I guess the problem is quite clear - the subs are far too short for the image data to rise above the background noise level of the CCD.   Compounded by the fact that the Ha sensitivity of the unmodified EOS300D is very bad.

[Ian]

Mark,

I was doing 60 second subs using my Art285 and 300mm camera lens and getting a little more, but not much than you. I think your analysis is correct that you need longer subs (as did I) which leads you into guiding. I didn't bother saving them because I then went on to try and get guiding working, which I didn't, and then gave up (it was rather late by then).

Having said that, while the image is probably not one for the calendar, it's still a fantastic achievement to capture something like the Horsehead, it's so incredibly dim, and the emission nebula behind is really only emitting Hbeta...

We tried to do the horsehead visually with the lightbridge and failed miserably.

[Mike]

For 30 second subs and without a Hydrogen filter of any kind that is a good effort. It will be interesting to see what you guys can get out of your cams once you get guiding sorted out.

[JohnP]

Mark - At least you got it which is an achievement in itself - well done. The horsehead is not easy. I have seen some great images though on various forums with unmodded DSLR camera's but you will need to up the 30 sec subs... Guiding here we come.....

John

[MarkS]

I've been doing a lot of technical analysis since taking this disappointing Horsehead image and come to some interesting conclusions.

1) Light pollution at Tuesnoad is approx 5% of light pollution in Sidcup.  In Sidcup the sky flux is 280 ADU/pixel/minute as measured on my Canon EOS 300D at ISO 800 F6.3 : at Tuesnoad it was only 15.  But even 15 is still far from ideal. (ADU is a digital unit in the RAW image from the CCD).

2) I was "lucky" that it was cold.  At a temperature of 20C I've measured the dark current of the EOS300D to be 40 ADU/pixel/minute at ISO 800.  This is much higher than the Tuesnoad sky flux and would be the main limiting factor for astrophotography.  However, on the evening I took the image the ambient temperature was probably 2-5C.   At 7C I've measured my ISO 800 dark current to be 10 ADU/pix/min, so on that evening it was probably half again.  I've read elsewhere that dark current halves for every 6C reduction ...

3) The striations on my Horshead image were due to CCD "patterning". This is a tiny effect (approx 1ADU) but for such a faint object it is very significant.  In the image below I've managed to remove most of the patterning.

4) The Canon EOS300D is very terribly insensitive to Hydrogen Alpha emission (extreme red) but reasonably sensitive to Hydrogen Beta (green/blue).  If I got the internal filter changed, apparently it would be 4x more sensitive to Hydrogen Alpha.  I'm very tempted to try this.

The overall signal to noise ratio in the image below is approx 4. This is very poor but agrees well with mathematical theory. The image is 150 stacked frames of 30sec each.

To improve the signal/noise ratio I need to:
1) Start guiding in order to increase the sub-exposure lengths i.e. 5-10 minutes instead of 30sec.
2) Maybe modify the EOS300D with a filter that passes Hydrogen Alpha.
3) Buy a dedicated astro camera with low dark current (for warmer evenings).

The image below is full frame but I've performed a 4x4 binning to reduce the image size and to improve the signal to noise ratio.  Note that the C11 does not illuminate the whole CCD at F6.3, hence the big disc-like artifact.

For anyone who's interested, I'm putting together a spreadsheet that calculates signal/noise given important parameters:
1) Sky flux (light pollution)
2) CCD dark current
3) CCD read noise
4) Flux of target DSO
5) Total exposure and sub-exposure time
I've come across others on the Web but they all seem to make an inaccurate, simplifying assumption somewhere.

[Fay]

Mark, I am very impressed with your image, especially taken with the Canon 300D. It is the first time I have seen this image, as I was ill at that time & missed this on the forum, as I wan't looking at the time.


Fay

[Ian]

wossesay Chris?

Mark, you nothing if you're not thorough mate

I think you've probably got as good a picture as you're going to get. Horsehead, unguided, no filters, long focal length, non-astro camera. I think if you wanted to make it any more difficult, you could try imaging it during the day

I'd like to see your spreadsheet, it might well take some of the guesswork out of exposure length...

[JohnP]

Hi Mark,

All very interesting stuff - sounds like you have been digisting some kind of astronomical image processing course over the festive period....

I'm curious - how exactly did you measure your signal to noise ratio? Obviously the signal part of the image varies all over the frame - did you pick a single location to make this measurement or do some kind of average on the whole frame...? As I understand it's normal with astronomical processing to measure the SNR using the background sky level but how do you do this with the Horsehead as none of the image is really background sky - you have the dark dust clouds at the bottom of the frame & then the 'curtain nebulosity' at the top.. I imaged the horsehead myself from Bromley just before Christmas using my astronomical camera & would be interested in doing a similar analysis.

You are right long exposures or a modified EOS will definitely help. When doing long exposures you need to be careful to not fill the well capacity of any pixel - this isn't easy if for example you have Alnitak in the frame - that's why I prefer imaging the horsehead with it out of frame. If I was you I would sort out my guiding/ long exposures first before going the route of modifying the EOS. I have seen some very good images taken with unmodified cameras.

Cheers,  John

[MarkS]

Thanks for your comment Fay.  I put the image there to canvas opinion on what I need to do to improve things.  However I'm stunned at the difference between a single sub (full of noise) and the final image.  Although I understand the maths, I still find it staggering what can be achieved with digital processing.  It's just like magic!

[MarkS]

John,

I agree with your comments.  There's only one good reason to modify the Canon and that's to improve the H-alpha  response.

How did I estimate S/N ratio?   I chose a specific area immediately around the horse's head itself.
Take the final image I posted as an example and open it in your chosen processing program.  The standard deviation of pixels in the red channel is approx 7 or 8 in the dark region of the head and more or less the same in the bright patch immediately behind the neck.  The average red channel pixel value in the dark region is approx 75 and just behind his neck is approx 110.  Assuming the dark region should be more or less black, the actual signal level is 110-75 = 35.  So the S/N ratio is 35 divided by 7 or 8 which comes out to around 4 or 5.

I'm realistic - with the Canon I'll never get a Horsehead image as stunning as Mike's.  But all the same, I'm very interested to see just how far I can push it.

[Rocket Pooch]

Hi MarkS,

I allways do a background measurement and sub profile to assess what the site is like, I'll dig out the numbers and post them up onto a map, but from memory with my HX it goes something like this

Home background ADU = 3,500 Gym lights off

Tuesnode = 1,200

Headcord = 700

Lydd = 400

Les Granges = 240, yes it dark there and there is still only place left for later this year, men only apply!!!

I tend to remove darks and work on my standard flats for the reading, the other things if the light gradient, Tuesnode is not good, mind you Chapel Farm Sports centre is worse.  John and I (when it stops raining) will be going to Lydd from March onwards in the Pikey2000 your more than welcome to join us.

These figures obviously tie in with your S/N points and comments as well. 

Chris

[MarkS]

Yes the Lydd/Dungeness area must be the darkest place in Kent, possibly in the whole South East of England (though avoid the immediate area around the power station).  I'd be interested in tagging along - I'm determined to get guiding soon so I can get some decent length subs otherwise the benefit of the darkness is lost.