Mystery Streaking Help Needed.

[Spacehead]

Hi All,

I feel that with a lot of extreme persistence with my limited kit (tracking only) that I have made some really good progress recently - spending night after night in the garden till 2.30am then processing till 4.30am, then experimenting for hour after hour with different stack combos, DSS, Registax 6, Corel Paintshop X8.  Short on dark time at this time of year I was more than pleased when I landed about 70 subs of the Pelican (my goal is to better develop my skills in PA/tracking for the moment).  I am actually capturing fairly good Ha in my subs considering the light polution here in Manchester.

So I've hit a problem.

I have seen this before but never found out what causes it - but its a real shame because without this "problem" my images would be coming along nicely.

As you can see there are streaks in my picture.  The points I hope will help anyone point me in the right direction with correcting this issue are as follows.

The streaks are not present in the lights OR flat - no matter how I stretch or fiddle with them to try and bring the streaks out.
Stack with no darks and no flat - streaks do not show.
Stack with darks and flat - streaks show.
Stack with no darks and only flat - streaks show.
Stack with only darks and no flat - streaks do not show.
Stack with only flat - streaks show.

So I conclude that the processing problem lies with my flat file.

The flat was taken with my scope against a flat screen PC with a white blank image on it - there was a stretched t-shirt over the scope.
The flat - whilst taken at a different time than the lights - is a good representation of the brightness fading for my setup in that it works very nicely to remove the vignetting - just causes these smears.

It doesn't matter which software I use - registax6 or DSS - the streaks are still showing.

I use windows 7, RS6 and latest download of DSS.

Kit is modded (By Juan CheapAstroPhotography - SUPERB SERVICE AND BLOKE) EOS 1300D, SW200p, Tracking only, 1 min subs iso800.

I am processing JPG only at the moment - because I simply don't understand what on earth DSS does with CR2 files - they all come out grey.

I need a laymans terms "magic bullet" if there is such a thing for this problem - to set me off down the technical path to correction.

The final image is below.

My hair is thinning as it is at my age - so please anyone - help me out before I lose any more!!!

 

[wxsatuser]

Looks like correlated walking noise, it's a drift in the noise pattern.

If you can dithering, try it but forget the darks.
Darks are quite hard to match and may do more harm than good.

[geoflewis]

29 minutes ago, Spacehead said:

The streaks are not present in the lights OR flat - no matter how I stretch or fiddle with them to try and bring the streaks out.

I think that the steaking you are seeing is caused by tracking errors so when your stacking software aligns the stars any sensor noise creeps across the stacked image. You could check my theory by blinking through your sub images to see if the stars shift position along the direction of the streaks. A quick way to identify this is to blink the first and last images in your stack.

[wxsatuser]

BTW

Stick with RAW files, they might appear grey, it's upto you to get the colour out.

[wxsatuser]

4 minutes ago, geoflewis said:

I think that the steaking you are seeing is caused by tracking errors so when your stacking software aligns the stars any sensor noise creeps across the stacked image. You could check my theory by blinking through your sub images to see if the stars shift position along the direction of the streaks. A quick way to identify this is to blink the first and last images in your stack.

Thats correct.

The op could look at dx dy in DSS to see how much each frame is moving.

[Spacehead]

Hi Folks - thanks for this.

Yes - with tracking I have to limit myself to 1 min subs as there is naturally drifting over the period.
I can visually see this when I scroll through all the thumbnails making them flip through looking like a movie.

Ok - with the "streaks" also going in the direction of the drift, I am pretty much convinced that is the problem - additionally, I have just done the following.

1.  Stack a single image with the flat - no streaks.
2.  Stack the first and last image with the flat - no streaks.

Now with both of the above, I wouldn't expect "streaks" - but if it was possible, I am certain that there will be misplaced pixels on the number 2 test - one at the start of a given streak, and one at the end of a given streak - with the line between the two points being drawn and filled in for each sub as it is processed against the flat.

I am "with this".

But what can I do?  I know that other people do get this drift, they must - otherwise they would be rock solid in their tracking and would do 5 min subs with no problem.

So in terms of how I understand this now, what can I do to "pre-align" each and every sub (crop?) before I stack them?  Any ideas appreciated!!

[wxsatuser]

I assume your not guiding?

It may be a good idea to start guiding, it will make all the difference.
Obviously make sure your PA is as good as possible, some form of drift align will help.

 

[Spacehead]

I'd love to start guiding - but havent the cash ATM - I will try to pre-align the images with PIPP before stacking and see what happens with that.

[Spacehead]

Ok - it isnt drift which is causing this.

I have used pipp to pre-align the images, and they match perfectly now - with no drift AT ALL when blinking through 70 images.

I put those through DSS with the flat, and I get exactly the same effect.

I'm at a loss - because the stars are all ok - yet the Ha isnt????  How can this be?


 

[Spacehead]

Correction - there is still drift - gawd I am sorry - I am flapping around like a chicken with no head!!!

 

[geoflewis]

2 minutes ago, Spacehead said:

I have used pipp to pre-align the images, and they match perfectly now - with no drift AT ALL when blinking through 70 images.

I think that you are misunderstanding the sensor noise 'drift'. As your tracking drifts the stars move diagonally across the sensor, but sensor noise stays in the same place, so by aligning the stars the effect is that the sensor noise (in the background) appears to drift relative to the stars, which is what you are seeing in your final image. The best way to erradicate the sensor noise is to calibrate with darks and it seems that when you do this the streaking disappears...

3 hours ago, Spacehead said:

Stack with only darks and no flat - streaks do not show

I do not understand why the problem reappears when you introduce flats, so are you sure that you are calibrating the lights with the darks and not the flats with the darks when you calibrate with both.

[geoflewis]

4 minutes ago, Spacehead said:

Correction - there is still drift - gawd I am sorry - I am flapping around like a chicken with no head!!!

No need to apologise, we've all been there and many, like me remain there at one level or another. It's what this forum is for ;-)

[Spacehead]

God this is complicated! - Ok - I think that what you mean is have I entered the light file into the light section correctly, and the dark into the dark section in DSS when I am selecting my files - and not got them the wrong way round?

I will double check everything with a quick stack.

[Spacehead]

But if it was sensor noise - wouldnt I be able to bring it out by just stacking the lights on their own?  If I do that - the streaking is not there.

[geoflewis]

1 minute ago, Spacehead said:

God this is complicated! - Ok - I think that what you mean is have I entered the light file into the light section correctly, and the dark into the dark section in DSS when I am selecting my files - and not got them the wrong way round?

I'm not familiar with DSS, but you can calibrate both your lights and flats with darks - those for flats usually being called flatdarks. I wondered if you'd loaded you darks as flatdarks by mistake. Darks need to match the lights and Flatdarks need to match the flats. Then of course there is always bias - sorry, yes this is comlicated....

[geoflewis]

4 minutes ago, Spacehead said:

But if it was sensor noise - wouldnt I be able to bring it out by just stacking the lights on their own?  If I do that - the streaking is not there.

Yes, I would have thought it would show in a stack of just the lights, so you've got me beat on that..... . Have you confirmed that the streaking matches the drift of the stars when blinking the individual sub exposures? How many flats did you stack, did you shoot flatdarks, to calibrate the flats?

[Spacehead]

I dont know what is meant by "calibrating flats" - to me at this stage, I only have one flat, it does work (reduces the inner glow) very effectively oddly enough - maybe I got lucky (on that front).

I am stacking 70 images - but the thing is - I DO have to take a step back because I havent actually addressed the drift now that I have discovered that PIPP hasnt aligned the subs for me.  Alignment and getting rid of the drift is my first step - as the direction of the drift through the subs when blinking matches the direction of the streaks.

My theory - DSS and Registax are applying the flat frame before alignment.
1.  Apply flat.
2.  Align shifts position of flat effect.

Somewhere along those lines......

I need to re-collate my info, be more accurate, stop flapping, chill out and concentrate!!!!  So I can provide you with more viable information.

[wxsatuser]

You really ought to have more than one flat.
Everything about stacking is reducing noise, this means plenty of subs, thats lights, darks, flats and bias.
Get at least 20 frames or more of each calibration file, make sure the darks match the temperature of the lights.
You cannot remove the drift after the fact, to get rid of drift you will need a better PA, drift align if possible.

[swag72]

I have had these sorts of streaks on a few occasions...... For me this is the data telling me that it's time to redo my calibration frames. So I redo my darks (40 or so of them), my bias (125 of them) and my flats (about 50 or so) and then normality is resumed. I use a calibration frames library and so redo them perhaps every 6-8 months.

[Stub Mandrel]

6 hours ago, Spacehead said:

Hi All,

I feel that with a lot of extreme persistence with my limited kit (tracking only) that I have made some really good progress recently - spending night after night in the garden till 2.30am then processing till 4.30am, then experimenting for hour after hour with different stack combos, DSS, Registax 6, Corel Paintshop X8.  Short on dark time at this time of year I was more than pleased when I landed about 70 subs of the Pelican (my goal is to better develop my skills in PA/tracking for the moment).  I am actually capturing fairly good Ha in my subs considering the light polution here in Manchester.

So I've hit a problem.

I have seen this before but never found out what causes it - but its a real shame because without this "problem" my images would be coming along nicely.

As you can see there are streaks in my picture.  The points I hope will help anyone point me in the right direction with correcting this issue are as follows.

The streaks are not present in the lights OR flat - no matter how I stretch or fiddle with them to try and bring the streaks out.
Stack with no darks and no flat - streaks do not show.
Stack with darks and flat - streaks show.
Stack with no darks and only flat - streaks show.
Stack with only darks and no flat - streaks do not show.
Stack with only flat - streaks show.

So I conclude that the processing problem lies with my flat file.

The flat was taken with my scope against a flat screen PC with a white blank image on it - there was a stretched t-shirt over the scope.
The flat - whilst taken at a different time than the lights - is a good representation of the brightness fading for my setup in that it works very nicely to remove the vignetting - just causes these smears.

It doesn't matter which software I use - registax6 or DSS - the streaks are still showing.

I use windows 7, RS6 and latest download of DSS.

Kit is modded (By Juan CheapAstroPhotography - SUPERB SERVICE AND BLOKE) EOS 1300D, SW200p, Tracking only, 1 min subs iso800.

I am processing JPG only at the moment - because I simply don't understand what on earth DSS does with CR2 files - they all come out grey.

I need a laymans terms "magic bullet" if there is such a thing for this problem - to set me off down the technical path to correction.

The final image is below.

My hair is thinning as it is at my age - so please anyone - help me out before I lose any more!!
 

Everyone will tell you this is caused by noise, PA drift etc. etc.

I get this on occasions with my 450D and I am convinced it only happens when there is thin, uneven, high cloud.  The direction of the lines credibly aligns with the direction of the cloud.

Try stacking just a few of you best subs, if it is noisier but less streaky, then it is probably caused by cloudiness in the poorer subs.