Challenging Targets

[ICallHimGamblor]

I have been steadily getting better at StarTools, and it is a fantastic tool. But a lot of the targets I shoot tend to be dimmer and more challenging, especially with light pollution and general inexperience. All the tutorials or threads I have seen deal with much easier, larger and brighter targets, and I guess I'd like to know some tricks for coaxing good data out of more challenging targets.

For example, I have a TIF of NGC 1961 here:
https://drive.google.com/file/d/0By0hG9 ... sp=sharing

It is three hours of lights from my Celestron 9.25 and Canon T3i. For some reason it is giving me fits, way more than any other similar galaxy, and I think it is due to my use (possibly 'misuse') of the Wipe module and the galaxy's rather low surface brightness and delicate structure.

There is an odd gradient in the stack that I am positive comes from an odd wrinkle in my t-shirt flats, but the galaxy is small enough and far away enough that I can crop out the chaos before I start processing.

Can you point me to some tricks for images such as these, or am I at the border of what light pollution will let me do?

Thanks.

[admin]

That's some challenging data for sure!
Teasing out faint detail absolutely requires good flats, and as you point out, the presence of gradients and unevenness is currently readily apparent.
At this point I would look at improving acquisition; focus, guiding, good flats and bias frames.
Light pollution makes it really extremely hard to image faint objects, as they barely stand out from the signal. Do you use a filter at all?
LP filters mess with your colors, but if you image the colors separately you can combine an LP-filtered luminance stack with a non-LP-filtered color stack and get the best of both worlds.

[ICallHimGamblor]

Hmmm... I thought I was far enough away from the oddness of the flats that cropping it out would solve the problem until I could retake them. Is that not correct?

Yeah, I have a light pollution filter that seems to help a bunch, and I take subs so that the histogram is 1/3 from left like is often discussed at cloudynights. I get 180 seconds each.

I use BackyardEOS to focus on a nearby star just before imaging and I autoguide with PHD. I don't know how I can get better at those. Suggestions? I get minimum FWHM and autoguide. What more is there?

[admin]

Hmmm... I thought I was far enough away from the oddness of the flats that cropping it out would solve the problem until I could retake them. Is that not correct?

Yeah, I have a light pollution filter that seems to help a bunch, and I take subs so that the histogram is 1/3 from left like is often discussed at cloudynights. I get 180 seconds each.

I use BackyardEOS to focus on a nearby star just before imaging and I autoguide with PHD. I don't know how I can get better at those. Suggestions? I get minimum FWHM and autoguide. What more is there?

It could also be that the stacker has trouble with coma your optical train seems to be exhibiting. Your stars seems elongated, lacking a nice central core. The star shape is never the same across the image. Any detail in the galaxy then too is similarly 'smeared out' and difficult to recover. The data seems to be rather oversampled still, requiring a good amount of resolution reduction before 1 pixel really corresponds to 1 unit of detail in the image. This could be due to bad seeing too I guess?

If the signal you're after is very faint and buried in just a small amount of your CCDs dynamic range, you may get better results from stacking using averaging (this increases dynamic range), rather than picking a 'best' value from the stack (ex. median stacking).

[ICallHimGamblor]

Very interesting. Thanks a ton, you've given me a lot to think about.

[admin]

Very interesting. Thanks a ton, you've given me a lot to think about.

Would love to hear how you get on with this as I bet I'm sure this is something other people might run into as well. You're probably looking at an optimization you can make to the acquisition stage (or possibly pre-processing), rather than something that post-processing can/should solve.