Stupid Question About Lights

[gnomus]

I had a bit of a mixed session last night.  Clouds kept rolling overhead.  When they did so, PHD2 would lose the guide star.  Usually it was able to pick it up again after the cloud passed, but it lost it altogether towards the end.  So of the 30 frames I shot, only 8 were good.  Even amongst those, some were excellent with good initial colour whilst some were a little orange (I think this is from cloud passing over and reflecting back what light pollution is in the area).

Anyhow this got me thinking.  If I understand correctly, the main reasons  that I take multiple subs is so that I can increase the "Signal" to "Noise" ratio.  This works primarily because there is more signal than noise on each subframe.  Ideally, I am trying to take the 'same' picture multiple times.  That being the case, why don't I pick out only my top quality frames and make copies - perhaps multiple copies?  I could take 4 copies of my 8 good frames which, combined with the originals, would give me 5x8=40 subs for stacking.  

Has this beginner misunderstood something?

  

[JemC]

Hi gnomus

I'm just a beginner myself and I'm sure people with more experience than me will probably explain better than I can,

I understand what you are saying but the way I understand it is... if you where to just copy your best 8 you would only be copying the data you have from those and every time you copy one you lose a little of the quality anyway, you would also not be adding anymore data which is basically what you want to be doing in order to get better images, but like I said I'm sure it can be explained better,

I do apologise if what I've said is rubbish

Regards

JemC

[D4N]

You do not increase the signal to noise ratio by doing that.

It is a ratio so by increasing the noise at the same rate as the signal you achieve nothing. Multiple subs work because the noise is random so with enough subs it cancels itself out. Copied noise won't do this.

TSED70Q, iOptron Smart EQ pro, ASI-120MM, Finepix S5 pro.

[johnrt]

Indeed the noise you eliminate by stacking is random, by copying subs you are copying the noise too and removing it's randomness.

[hjw]

Noise is a random fluctuation around a given value (let's say brightness of a given pixel) It can be higher or lower than the real value. By taking the average of a large number of measurements (frames) the higher and lower variations cancel each other out. This actually holds true for both the signal and the background. If you stack the same picture over and over again the value for a given pixel will always be higher or lower and there can't be any cancelling out effect. The quality of the image remains unchanged.

[gnomus]

Noise is a random fluctuation around a given value (let's say brightness of a given pixel) It can be higher or lower than the real value. By taking the average of a large number of measurements (frames) the higher and lower variations cancel each other out. This actually holds true for both the signal and the background. If you stack the same picture over and over again the value for a given pixel will always be higher or lower and there can't be any cancelling out effect. The quality of the image remains unchanged.

Thank you all for your clear (and similar) explanations.

[ollypenrice]

Yes, it's simple maths. If you take a pixel which reads 32000 ADU, add up this value ten times and divide by ten you get 32000 - exactly the same as your starting point. Shame! 

Olly

[gnomus]

Yes, it's simple maths. If you take a pixel which reads 32000 ADU, add up this value ten times and divide by ten you get 32000 - exactly the same as your starting point. Shame! 

Olly

As you say...

I processed my 8 Lights along with 25 Bias frames (no darks or flats).  These were M33 - 5 mins (guided) at 400 ISO through an Edge HD 8" (f/10) with a 0.7X reducer using my FF Nikon DSLR.  A few issues that I keep running into are:

1)  What is the 'halo' that I keep getting?  I attach a full-frame view of the stack that has just been DBEd and stretched in Pixinsight.  I cannot get rid of this halo and that means I need to crop heavily.  I am ending up having to chop out some of the galaxy.  (I know a need a wider view).  My dew-shield was constructed out of a blue camping mat and extends out some 16" or so from the front element.  Could the shield be giving rise to this, or is this just a function of shooting full frame through the scope I have?

2)  Damn my sensor is filthy - I do keep cleaning it.  I will shoot some flats next time.

3) I suspect I am not getting enough data at this exposure.  Should I be shooting ISO 800 (instead of ISO 400) through my 'slow' telescope (even with the reducer)?  Or is it just the case that I need more subs?

I'll also attach the fully processed, cropped shot to show the best I can currently do with the 8 frames I have. 

[D4N]

Wow that's quite a gradient, dunno what's causing that but you definitely need some flats. It looks like there is some nice data hidden in there though.

Sent from my iPad using Tapatalk

[Scott]

I've not seen gradient (or whatever it is) that bad before and sorry, can't help you with it, but I will say that for 40min exposure time, thats quite an impressive amount of detail. obviously it's quite noisey but more exposures and calibration frames will go a long way to helping there. good luck trying to keep the dust out of the dslr... it's a losing battle me thinks. maybe a clip filter (can you get them for nikon?) would help.

[ollypenrice]

The vignetting is clear proof of the need for flats. I don't know how big a circle the Edge reducer produces (the Celestron one is bigger than the Lepus ruducer's which is very small) but since the galaxy will fit on the chip without the reducer, by the looks of it, why use it? If the galaxy will fit on the chip at native FL just do that. Reducers don't speed up capture of a target that fits on the chip at native. (Beware of the F ratio myth.) Without reducer I'd have thought the Edge should cover full frame.

I also think DBE can do a lot better than it has done here. The vital things to watch in DBE are;

- Crop all edge artefacts before using it.

- Don't use lots of background markers, use 5 or 6 per row at most.

- Be ultra careful to have no markers anywhere near the object(s) themselves. Keep them well away from M33.

- Examine the gradient map before applying it. It should show no sign of having picked up the structure of the galaxy. It should be wthout detail and really just be a gradient map, in this case showing the vignetting.

- Raise the tolerance parameters if some of the background sky lies outside the range accepted by the markers.

The dirt on the chip is going to have to be removed. There will be no data underneath it so flats can't do much about it.

Olly

[sharkmelley]

You certainly have a very interesting halo there. The reducer may well be introducing some gradient issues that do not happen without it.  I'm not familiar with the 0.7x reducer on the Edge but I do have experience with the 0.63x reducer/flattener on the (non-Edge) C11 which can cause a nasty structure to gradients that are very difficult to remove, even using flats.  However, I think that the way you have used DBE is giving it a extra structure that it wouldn't  have otherwise.  Take a look at Olly's suggestions

I can definitely understand why you want to use a reducer.  A non-cooled DSLR has thermal noise which you need to overcome.  At f/7 you end up getting a better signal to thermal noise ratio than you would do by using f/10 and binning the image down to the same size.  This is essentially because the target of your image at f/10 (e.g. M33) includes thermal noise from a larger number of pixels than at f/7 but the number of M33 photons remains the same in both cases.  That said, the difference will only be noticeable if thermal noise is the major component of noise in your images (i.e. greater than the noise from sky glow).   I would suggest imaging with and without the reducer to see if the difference in image quality is noticeable under your sky conditions.

I would be interested to see a light frame and a flat frame for the 0.7 reducer on the Edge to see how well the flat frame behaves.

Mark

[MiguelMJr]

Why don't you just take more subs tonight or tomorrow night or the next night? You don't need to take all your exposures in one night on an object that doesn't appear to change. Taking more subs next year would be just fine too.

Just keep collecting subs until you think you have enough.

Miguel

[gnomus]

........ but since the galaxy will fit on the chip without the reducer, by the looks of it, why use it? .... (Beware of the F ratio myth.) Without reducer I'd have thought the Edge should cover full frame......

Thanks again for your advice. The rather simplistic answer to your question is ... because it cost me a small fortune!! Seriously, I thought I was gaining 0.7 of a stop with it. If that is the 'F-Ratio myth', then I will try going without it - especially for the smaller objects it would make a great deal of sense. (And of ourse this strengthens the argument for getting a 'faster' scope.)

I will definitely take flats next time, and will prioritize this over darks, if time-limited by something. As a beginner, I had read a fair bit but clearly got the wrong end of the stick about some things.

I'll also do another wet-clean of my sensor.

Thanks again to everyone for their continuing patience.

[gnomus]

Why don't you just take more subs tonight or tomorrow night or the next night? You don't need to take all your exposures in one night on an object that doesn't appear to change. Taking more subs next year would be just fine too.

Just keep collecting subs until you think you have enough.

Miguel

Thanks Miguel. I may well try that. How do I get round the fact that I took no flats or darks last night, though?

[MiguelMJr]

Take your flats and darks tonight, just use the same temp and exposure setting as last night, things will be less complicated that way.

Miguel

[MiguelMJr]

To clearify... use the same temp and exposure settings as last night for your flats, darks and subs. Also compose the subs as close as possible to what you already have.

Miguel

[gnomus]

To clearify... use the same temp and exposure settings as last night for your flats, darks and subs. Also compose the subs as close as possible to what you already have.

Miguel

OK.  I'm taking the darks as I type.  I'll fire off some flats either later tonight or tomorrow morning.  I can do the flats indoors - right?  The only thing that I think I may have a problem with for the flats is getting the camera back on the telescope in the exact orientation that I had it last night.  It was roughly horizontally aligned with the dovetail on the scope.  But if it's even a little off it could give me unusable flats - am I right?

[Nathan_Pembs]

Temperature isn't important with flats, only your darks. For your flats you just the same ISO, keep your camera at the same focus point (easy if you can lock it) and orientation. I usually lock everything up and take the flats the next day, using a white shirt stretched over the end of the scope pointed at my laptop screen made white with a txt document opened up to make the screen all white.

[MiguelMJr]

Hey G,

Nathan is correct.

Darks are used to quantify the "dark noise" that is produced during exposure. The amount of dark noise is dependent on the temperature of your chip and the length of your exposure. You need to take your dark frames at the same temperature and for the same length as your lights. The chip shouldn't see any light during producing your darks so make sure that the scope (or camera) is capped and there are no light leaks.

Flats are used to remove imperfections in the optical train, things like dust spots and the such.

Hope this helps.

Miguel

[MiguelMJr]

gnomus,

Check out this thread, it may help.

http://stargazerslounge.com/topic/232142-help-needed-with-darkstemperature-with-dslr/

Miguel