The "No EQ" DSO Challenge!

[Filroden]

Indeed! Dew doesn't seem to be a problem yet but high clouds are! Forecast suggests it will clear soon and my target is low so I'm hoping I will get something! My first subs show some nebulosity so fingers crossed. 

Spacing is my next challenge. I've only got it rough at the moment and I need to improve it.

[Nigel G]

Good luck chaps. keep at it.

I have been experimenting with exposure times on M33, so far managed to get 10 x 60s, 4 x 120s, 20 x 45s and now bulking up on 30s.

All the kept subs are good subs, no trails in the 120's and they only took 20 minutes to get.

It will be interesting to see the final image when its finished.

Cheers

Nige.

 

[SteveNickolls]

5 hours ago, Nigel G said:

Any thoughts ?

Hi,

I would always try to image individual frames as long as you can to improve the S:N ratio for the fainter parts of your image. I would add the caveat that there's a personal decision to be made on what is bearable in terms of successful images which DSS will accept. If practice shows that a very few much longer exposures in a stack greatly improves the overall image then I'm in favour of it despite the big hit in wasted time. I have never tried imaging longer than 70 seconds thinking the time loss was more important but it would be worthwhile to now try. More importantly I've never stacked exposures of different times to see the benefit. Thanks Nige.

Since last December I've kept a record of the percentage success rate of images in DSS from different exposure lengths, really carrying on the technique suggested by Joseph Ashley in his book, but in this instance covering thousands of light frames not a small number. The results show the mount and imaging equipment do quite well over a wide range of exposure times but start to show a significant fall off at 50 seconds and an inconsistency in performance up to 70 seconds. I'm believe wind has some effect on the success of frames but mostly any reduction in acceptable frame numbers is down to the mount design and it's (in)capacity to perform long tracking. The average percentage of frames for exposure duration are listed below.

10 seconds 79%     Mostly taken early on in my imaging of DSO's

25 seconds 94%

30 seconds 90%

40 seconds 90%

50 seconds 68%

60 seconds 76%

70 seconds 43%

I'd expect the percentage rate of acceptable frames would continue to drop beyond 70 seconds but if the few successful ones provide a significant improvement in the final image then they are worth taking and adding to the stack.

Interestingly I have done a number of wide field imaging sessions taking images with my camera piggybacking on the mount and telescope and never had less than 100% frames acceptable to DSS (these are exposures between 40 and 60 seconds). Any trailing is probably better hidden by the wide field image. I believe wide field images could be taken well over the exposure limit of images using longer FL's provided the limits of field rotation are not grossly exceeded. Worth an experiment.

So thanks for this moment of thought provoking Nige.

General plea-can we now have a clear night please to practice?

Cheers,
Steve

 

[The Admiral]

Whilst I was ruminating on the subject of stacking I thought I'd have another look at the DSS manual, and by chance came across a theory part that I hadn't seen before. No doubt others have, but here's the link anyway: http://deepskystacker.free.fr/english/theory.htm

One of the details that caught my eye was the paragraph titled "Can I combine two (or more) resulting images?". Specifically this text, as I think it may have some relevance to Nige's question:

" Absolutely, the square root rule applies with a small twist.

When combining two images the SNR increases by 1.414 (square root of 2).
If both images have the same SNR then this is the same as doing a single stack. That does not mean that the combination is giving the same image, just that the SNR will be the same.

However if one stack contains more light frames than the other, the SNR of the two stacks will be different and the SNR of the combination will be lower than the SNR of a single stack containing all the light frames.

Thus by combining the result of a 10x1 minute stack with a single 1 minute frame the SNR is roughly the same as the one obtained by combining 2 light frames. This is due to the fact that when combining two images the noise is additive and the best image is damaged by the worse image in the process."

I'm still trying to get my head around this, but does it mean that if you stack, say, 100 x 30s subs and stack 10 x 60s subs, and then combine them, the resulting SNR will be dictated by the poor SNR of the 60s stack? If so, I think it means that the amount of time devoted to longer subs should be at least the same as to the shorter subs, in order for each to have the same SNR (or will it be the same???). But it seems a bit contradictory to me that adding more data should result in a poorer SNR though. I must be missing something (which, to be honest, wouldn't be surprising!). Also, I thought the noise would add in quadrature (i.e. the square root of the sum of the squares), not simply be additive, and the signals in the two images would be additive. So what would the resulting SNR be?

Anyone got a clue?

Ian

Edited October 27, 2016 by The Admiral

[Filroden]

But why wouldn't you stack all light frames (regardless of exposure) in the same process to benefit from the greatest improvement to SNR?

i cannot see a reason to combine two previously stacked images over stacking their constituent subs unless, by chance, both stacked images contained equal number of subs and had the same signal (which probably means they had the same exposure length).

i combine all my images in a single integration, weighting subs with better SNR and FWHM/eccentricity). I have darks to calibrate subs of different lengths (they share common bias and flats). I'm sure DSS does something similar using the tabs at the bottom (lights and darks of same exposure are added to their own tabs?).

the one occasion I have stacked already stacked images is to create a pseudo luminance image from separate RGB images. But typically my RGB images have similar number of subs behind them.

[The Admiral]

43 minutes ago, Filroden said:

But why wouldn't you stack all light frames (regardless of exposure) in the same process to benefit from the greatest improvement to SNR?

Perhaps on reflection you are right. I think I was confusing using different groups as tantamount to having separate stacks (they would need to be put into separate groups because they would need to be associated with the appropriate length darks) . Presumably that's not the case and that they are all essentially stacked together. Presumably you are also saying that if you produced different stacks on separate nights, that you would re-stack the lot together rather than combining the separate results? PixInsight does give one greater opportunity to finesse these things though, which of course is why it is quite complex.

It still leaves me a bit confused about the DSS statement though.

Ian

 

[Filroden]

Well, after over 2 1/2 hours of imaging, I calibrated and reviewed the subs this morning and had to discard most of them due to cloud cover. I suspect I've still included subs affected by feint clouds as the resulting image is very noisy. Worst affected was my RGB where I ended up with very little data (hence the subdued tones).

This is only 45 mins of 30s L subs with about 5 mins each of RGB. I only processed it quickly as there isn't really enough data yet. Saturday is looking clearer and if so, I'll hopefully add much more data to this.

So, here's my first attempt at NGC1333:

[Filroden]

11 minutes ago, The Admiral said:

Perhaps on reflection you are right. I think I was confusing using different groups as tantamount to having separate stacks (they would need to be put into separate groups because they would need to be associated with the appropriate length darks) . Presumably that's not the case and that they are all essentially stacked together. Presumably you are also saying that if you produced different stacks on separate nights, that you would re-stack the lot together rather than combining the separate results? PixInsight does give one greater opportunity to finesse these things though, which of course is why it is quite complex.

It still leaves me a bit confused about the DSS statement though.

Ian

 

Indeed. I would always go back to the raw subs even if taken over many nights and stack them all together. If you were using darks, you may even need tabs for different nights as well as different exposure lengths as the temperatures could differ from one night to another.

I have to admit I also cannot wrap my head around the math. In my simple brain I think it's like averaging averages. You don't know the weight/standard deviation of each average so you cannot be sure of the quality of the final result. Say you had two results:

999 of 1000 people all agreed the Earth was round from a survey

combined with

1 out of 2 people agreed the Earth was round from a sample of posts in a forum

If you just combined the results (99.9% and 50.0%) you would think only 74.75% of people thought the Earth was round. You're degrading your result with the worst sample. You'd actually need to weight the larger sample massively. However, when combining two already stacked images, DSS has no information over the size of each stack so cannot apply that weighting.

P.S. And assuming the sampling was equally random in both results, the right approach would be to consider this as 1000 out of 1002 people thought the Earth was round, i.e. both data sets were combined before generating the result.

Edited October 27, 2016 by Filroden

[The Admiral]

It's coming on, but as you say you are rather limited in data at the moment. It looks like it's a tough target though, and quite small.

Ian

[Filroden]

1 minute ago, The Admiral said:

It's coming on, but as you say you are rather limited in data at the moment. It looks like it's a tough target though, and quite small.

Ian

Indeed. I think it's both feint and very small (and there is lots of dust in the area which I don't think I have any hope of capturing). Nonetheless, I'm please to get some colour differentiation across the nebula and this is the first time I think my flats have worked (I tried a new technique of putting up a white image on my large TV and using that as a light box, with sheets of paper over the scope to increase the exposure times to between 0.5 and 1.5 seconds).

[The Admiral]

Where there's a will......

[Stub Mandrel]

I use a patch of evenly painted wall that is illuminated by diffuse daylight on our landing. To eliminate any bias due to uneven lighting I rotate the telescope taking (usually) three photos in each of 8 or 12 positions.

[Bobby1970]

Posted on error

 

Edited October 27, 2016 by Bobby1970

[Nigel G]

Last nights results.

4x120s plus 10x60s plus 20x45 plus 60x30s, Matching darks, flat and bias. DSS and new found methods with StarTools. 150p Canon 1300D

I need to re stack without the 120's to see a difference, I think I will do a couple of different stacks to get an idea how effective the longer subs are.

My clip in LP filter arrived today

Cheers

Nige.

Edited October 27, 2016 by Nigel G
add filter

[Filroden]

12 minutes ago, Nigel G said:

Last nights results.

4x120s plus 10x60s plus 20x45 plus 60x30s, Matching darks, flat and bias. DSS and new found methods with StarTools. 150p Canon 1300D

I need to re stack without the 120's to see a difference, I think I will do a couple of different stacks to get an idea how effective the longer subs are.

My clip in LP filter arrived today

Cheers

Nige.

 

When you think that's only an hour! It's crystal clear with lots of globular clusters clearly visible. And I love the colour.

[Nigel G]

5 hours ago, Filroden said:

Well, after over 2 1/2 hours of imaging, I calibrated and reviewed the subs this morning and had to discard most of them due to cloud cover. I suspect I've still included subs affected by feint clouds as the resulting image is very noisy. Worst affected was my RGB where I ended up with very little data (hence the subdued tones).

This is only 45 mins of 30s L subs with about 5 mins each of RGB. I only processed it quickly as there isn't really enough data yet. Saturday is looking clearer and if so, I'll hopefully add much more data to this.

So, here's my first attempt at NGC1333:

Ken, that's looking real good for such short exposure time, its all there and crisp too, can't wait to see after more subs added.

Nice, I think I need to start saving £'s  

Nige

[Nigel G]

6 minutes ago, Filroden said:

When you think that's only an hour! It's crystal clear with lots of globular clusters clearly visible. And I love the colour.

Thanks Ken, I have a lot to thank Ivo of StarTools for this image, he has given us some great info. More will be posted very soon.

Cheers

Nige.

Edited October 27, 2016 by Nigel G
addition

[SteveNickolls]

5 hours ago, Filroden said:

So, here's my first attempt at NGC1333:

You can already see some of the dark area surrounding the object. Hardest thing imaging dark on black but perseverance will pay off.

Cheers,
Steve

[Stub Mandrel]

1 hour ago, Nigel G said:

Last nights results.

4x120s plus 10x60s plus 20x45 plus 60x30s, Matching darks, flat and bias. DSS and new found methods with StarTools. 150p Canon 1300D

I need to re stack without the 120's to see a difference, I think I will do a couple of different stacks to get an idea how effective the longer subs are.

My clip in LP filter arrived today

Cheers

Nige.

Shiny!

You can dig more colour out of that!

[Nigel G]

Just about had time to snap a shot or 2 before clouds came.

Difference between LP filter and no filter. 45s on M33.

Nige

[Yamez]

That is an incredible difference, what LP filter did you use?

Edited October 28, 2016 by Yamez

[The Admiral]

Interesting! Yes that is quite a difference Nige, but to my eyes the curly-whirly appears more visible in the one without! May be an illusion. Does the filter need more exposure?

+1 to which filter are you using?

Ian

[Nigel G]

Its the Astronomik CLS CCD deep sky & Light pollution filter Canon clip fit.

It will work better with the modded camera.

I need to get a set of subs to see how it performs.

The write up says, Good with un modded , very good with modded cameras which is the reason I chose this one. With un modded camera there will be added contrast,

 

There is a loss of detail in the same exp time for sure. Hopefully not a waist of money!

Nige.

Edited October 28, 2016 by Nigel G

[Filroden]

There is a slight loss of detail but the filter will let you do longer exposures (probably not much of a benefit to us) and should reduce noise when integrating as the noise in the background will be much reduced. I see that as a huge benefit. 

[The Admiral]

I guess it really comes down to whether the SNR is improved, increased exposure or not. I can imagine though there the would be more potential with EQ imaging where it would allow greater sub duration. Still, we are scratching at the limits so any benefit to be had is worth seeking out.

Ian