StarTools RGB Noise?

[Todd1561]

I first posted this on the StarTools forum earlier this week but haven't gotten any responses, so I'm trying here...

I'm starting to get more into imaging and have been trying to learn more about post processing methods.  I'm following this guide: http://astro.ecuadors.net/processing-a-noisy-dslr-image-stack-with-startools/  But I'm finding after the initial AutoDev (and amplified by the wipe module) I'm getting significant RGB noise I can't get rid of it even after I finish the process outlined in article.  Does anyone know what might be going on here?  My original capture was with a NexStar 6se, 0.5x FR, and ASI224MC.  I used SharpCap to do the capture and stacking and I exported a 32-bit FITS without any stretching which I'm putting directly into StarTools.  10s subs x 37 frames.  350 gain.  I did not use any dark or flat frames, which might be part of the issue.  I'm including the images below.

Thanks!

Noisy image:

Original:

Original FITS: http://toddnelson.net/misc/astro/rgbnoise-orig.fits

[wimvb]

Despite the very short integration time, you've already captured some outer dust of this galaxy. A very good start.

I don't know about startools, but the best noise reduction by far is: more subs, lots more subs. With an uncooled cmos camera at high gain, the final number of subs should be in the hundreds. At 10 s per sub, 200 subs is still only half an hour of integration time. More data will not only bring the noise down, but also increase the bit depth, allowing you to reveal more detail in the outer regions of this target.

[spillage]

Try manual dev first. Push it just a bit 75% maybe and adjust the gain a bit then use wipe at default, crop to remove stacking issues then redo dev and wipe again.

[Todd1561]

Thanks guys.  Wim, I'll keep that in mind the next time I'm out.  I was hopping from object to object snapping frames.  I'll focus on one for longer next time and really try to capture a lot of light.  After a while I wasn't seeing any improvement in the live stack so I stopped collecting, but it's probably that the extra data would come to life in the post-processing

Spillage, I took your advice and it seems to have really helped, see below.  There's still some noise but it's nothing like what it was before.  Probably a combination of your processing techniques and capturing more data will yield a pretty decent result.  Do you ever use the AutoDev feature?  I think all guides I've read for StarTools show using it over the standard Dev tool but from what I can tell the results seem lackluster or at least inconsistent.  I used it when processing a M42 stack and it worked pretty well.  Although, now I'm inclined to reprocess it with your method and see if I can eek out a better image.

[spillage]

Sorry back at home so can take my time a bit more.

I have found using dev instead of autodev gives me more control, although if you use autodev but do not save it will give you a better idea of where to crop the image if needed.

You can also use repair as mentioned by @almcl. You can mask the galaxy then invert the mask and in auto choose to replace old mask. It will then only mask the stars and not the subject. In repair use warp or one of the other ones depending on the results it gives you.

You have not listed your kit, are you guiding? if so you might want to dither with the dslr and it might be worth looking at astrophotography tool (I cannot recommend it enough) the unlimited trial version is amazing but for under £20 the full version is worth it.

This is just a quick run through so you should get a better result and as always more data helps

cheers

Mark.

 

 

[Todd1561]

Thanks again, very helpful.  My setup is listed in my original post.  Pretty basic, just the stock AltAz mount without guiding.  So I'm limited to 15, maybe 20 seconds of exposure depending on where in the sky I'm pointed.  Thanks for doing a quick processing on my image, it helps me know if I'm even in the ballpark or not.  Seems like my latest attempt was OK given my source data.  The next time I'm out I'll go for more subs and maybe push the exposure from 10 to 15-20 seconds.  I also built up a decent library of dark images at the end of this session.  They aren't used in this image because as far as I can tell you need to specify the dark file during the SharpCap live stack, but I'll try adding them next time.

[happy-kat]

I too only use autodev to quickly show how much I need to crop otherwise I use manual dev.

[spillage]

Give apt a quick look. Also make sure you use darks and flats and as you are not dithering bias frames using deep sky stacker.

Actually @happy-kat is the person that advised me not to use autodev.

[The Admiral]

2 hours ago, wimvb said:

but the best noise reduction by far is: more subs, lots more subs. With an uncooled cmos camera at high gain, the final number of subs should be in the hundreds.

I second that! Alt-Az imaging still requires a significant integrated exposure time as with other methods of imaging, except you have to take many, many more of short exposure subs.

Have a look at the looonnng thread "No EQ Challenge" and get a feel for what others are using.

Ian

[bottletopburly]

Quick play in startools

[wimvb]

1 hour ago, Todd1561 said:

Wim, I'll keep that in mind the next time I'm out.  I was hopping from object to object snapping frames.  I'll focus on one for longer next time and really try to capture a lot of light.  After a while I wasn't seeing any improvement in the live stack so I stopped collecting, but it's probably that the extra data would come to life in the post-processing

I think that most of us start out that way: as sky-tourists. I also used to hop from target to target, capturing three or so on any night. But for my latest project, I collected 11 hours of data over 5 nights. Since I get best results west of the meridian, I sometimes still image two objects per night, but spend several nights on each. The only time I do short runs is to evaluate if a candidate is worth pursuing. If it is, I spend multiple nights on it. If not, I haven't wasted much time.

[wimvb]

1 hour ago, Todd1561 said:

maybe push the exposure from 10 to 15-20 seconds

The asi 224 is a low read noise camera that shines at short exposure AP. As long as you keep the exposure long enough to clear the read noise floor, you can increase the number of subs and not the exposure time. In your case that will keep tracking errors under control. But at high gain you basically have an 8 bit camera, instead of a 12 bit one. The only way to regain those lost bits, is to shoot more subs. But there's no shortcut to quality; to get a good image, you need to invest the time. Total time on target is measured in hours, rather than minutes.

One of the people to pioneer short exposure AP with low noise cameras, is Emil Kraaikamp. He combined a large aperture scope (40 cm) with short (1 s) exposures.

http://astrokraai.nl/viewimages.php?t=y&category=7

[Todd1561]

1 hour ago, wimvb said:

I think that most of us start out that way: as sky-tourists. I also used to hop from target to target, capturing three or so on any night. But for my latest project, I collected 11 hours of data over 5 nights. Since I get best results west of the meridian, I sometimes still image two objects per night, but spend several nights on each. The only time I do short runs is to evaluate if a candidate is worth pursuing. If it is, I spend multiple nights on it. If not, I haven't wasted much time.

I agree, as I'm just getting into this it's all so fascinating and I can't help but jump around to see what else is out there.  As I settle down and figure out how best to manage my limited time I know I'll focus more on specific targets and squeeze a little more data out of each session.  You all have been incredibly helpful, this forum is great.  This is a hobby with a very steep learning curve and all the help just in this thread alone has been invaluable. And thanks @bottletopburly for running my image through ST, looks great, much less noisy than 2nd attempt.

[bottletopburly]

Ivo replied to your post on startools forum ?

[Todd1561]

I had some time and decent weather last night to get back out and take another "whirl" at M51 (sorry).  This time I did 557 frames @ 10 seconds, so about 1.5 hours of total integration time.  Quite a bit more than the 6 minutes last time!  I couldn't do more than 10 seconds, M51s location in my sky yielded too much field rotation at anything longer.  I still haven't done any dark frame subtraction because I found out in the process that's only a feature of the pro version of SharpCap, so I'll need to decide if I want to invest in the full version.  I did however save each individual frame this time so I could try using DSS with my dark frames.  I again used StarTools but this time didn't have any of the RGB noise with AutoDev, so it must be all the additional light frames helped in that regard.  Although, I still liked the result using the manual Dev so I used that again anyway.  Overall I'm pretty happy!  I'm sure there's still a long way to go, but I like the progress so far.

First attempt back in April without RGB noise after suggestions here:

Last nights attempt:

[happy-kat]

That's a big step change well done.

Did you use the autosave fits file and did you make sure to not align rgb in DSS when stacking.

 

[Todd1561]

Thanks.  I don't think I used the autosave file, I just clicked the button in SharpCap to export a 32-bit FITS.  Is that not what I should be doing?  From what I understand when you export the FITS in SharpCap it doesn't include any of the stretching.  I used SharpCap to do the stacking live, I didn't use DSS for this image.

[happy-kat]

Oh I see I hadn't realised you stacked in sharpcap.

[tomato]

A bit late to this, but +1 for using manual dev rather than autodev in StarTools. Also, it has been said many times but there is no substitute for good data and lots of it. I too started out catching lots of targets in one night, just to see how they looked on camera, but I quickly learnt detail comes with patience and under UK skies, perseverance.

The quality of my images is definitely proportional to the total integration time, the best I have managed is around 4 hrs, but most DSOs benefit from 2 or 3 times this integration time.

[Galen Gilmore]

Nice job! 

It seems you are getting really good advice from others, but I think that you may want to give a different stacking software a go. Deep Sky Stacker (DSS) is a free DSO stacking software that is really good. Although it is not perfect and has some bugs occasionally, it really gives you good results. It is easy to use as well.

http://deepskystacker.free.fr/english/index.html