Calibration & Stacking - Pixinsight or DSS?

[Tim]

Having spent the whole afternoon trying to learn the method of calibrating registering and stacking in pixinsight, I thought I'd do a comparison side by side with the old favourite, drag and drop DeepSkystacker.

It took me hours to get the PI result, albeit that I didn't use the script.  It took minutes to get the DSS result.

Which side do you prefer? Left or right?? The data once stacked has had exactly the same treatment, just a linear stretch.

[Jannis]

The one on the left seems slightly sharper, but is having issues with what seems to be warm pixels. In the odd case that it's not, but stars, then it's showing a lot of stars wich isn't visable on the right image. There seems to be a slight contrast difference, but that can easily be corrected, so i'll ignore that part.

Personally i'd select the image on the right as my favorite, concidering the pixels vs sharpness.

[RikM]

I'll say the one on the right as well. There do seem to be fewer pixels requiring manual clean-up and the detail/contrast seems better. The odd thing is that the left hand images seems somehow smoother, which I like. 

[Tim]

and then there is this one...

[Tim]

I think when properly applied, the PI routines are very robust and will probably yield a better result. But it takes so loooooong to do in PI that i'm not entirely certain the marginal gains are worth the extra time investment........

What do you guys use?

[RikM]

I like that third one but I can't see past the differences in contrast really. It does seem to have more fine detail just back from the tip of the cone

I use MaximDL and normally use the SD Mask stacking option. I used to use DSS with either median kappa-sigma clipping or auto adaptive average depending on what gave me the best result. I did compare a comple of time between packages and Maxim seed to give both sharper detail and better noise control, so I stick with that. 

[harry page]

Hi

Off course I use pixinsight   for a rough work out I use AA5 ( very fast) but when it comes to a final stack I use pixinsight as it has better pixel rejection

and is very good at weighing the subs .

I always use the script as it is far easier and faster

Harry

[Tim]

Yeah Harry, once the script is understood and properly implemented it is a lot easier Your tute is great on it btw!

I did the first session in PI from scratch so that I could get a better understanding of all the steps and stages and settings involved, but it took literally hours. With the script, even adding in the cosmetic correction it came down to less than half an hour.

Cheers

Tim

[russellhq]

If it's not immediately obvious what's better quality wise, then I'd start looking at other factors i.e. time and cost.

My preference is the one on the right: the one with more contrast. But there's not a lot in it so it starts to get a bit subjective.

[narrowbandpaul]

Hi Tim,

I think the one on the right looks better, but that's very subjective. The only real way would be to try to estimate the SNR of the same area in each image. I would expect/hope that the PI would win, based on cost and sophisticated algorithms.

Paul

[ollypenrice]

Honestly I don't think it's all that subjective, myself. The one on the left is rattling with hot pixels which are pretty nasty. The one on the right is not only cleaner but has a greater dynamic range (more contrast) and is still cleaner. It should be the reverse. The one on the right looks as if it has had a further gentle iteration of levels and curves - meaning more noise - but in fact it has less. The close double in the cone of the cone (so to speak) are a tad clearer, too.

I ought to give PI stacking a go but I'm wearied by its inarticulateness and find engaging with it a real fag. On a couple of occasions very IT/PI types have come into my processing room and said, I'll show you. Great! They do this, that and the other and finally say, Well that's funny, it usually works but something's not quite right this time... Why can you just bung things in boxes in AA5 and hit go? You have to wait all of twenty seconds for a deep stack of full frame CCD subs to come out of the other end. If you don't like them you can change something and go again.

But, as I say, I should give PI a go. I belong to a generation that expects a manual. There is no PI manual (to my knowledge.)

I don't go anywhere near DSS. Worthy and free, but not for me.

Olly