M31 - Same data - different processing (DSS/Photoshop vs PixInsight)

[peroni]

I've just purchased a copy of PixInsight and since the weather is so foul I thought I'd dig up some old data.

This was the first use of my telescope for imaging taken 1st Jan 2013. ED80 scope on HEQ5 mount. Canon 1000D DSLR attached with no FF/FR. No guiding.

38 lights at 60 second exposure ISO800

20 bias, 20 darks, 30 flats

All thrown into DSS

Processed in Photoshop with Noel Carboni tools. Resulting image finished 23rd Jan 2013. See below: 

I used the same data taken 1st Jan 2013 but reshot the bias files. 

252 bias taken to make a superbias in PixInsight

Stacked in PI, processed in PI. Image finished 15th Nov 2015, see below:

PI is a steep learning curve. I followed the Light Vortex Astronomy tutorial here for PI workflow.

A lot of effort to produce the PI image (16 hours maybe!!!)

Was it worth it? I'm reasonably happy for only 38 mins total exposure.

What do you guys/gals think?

[Uplooker]

Hi peroni,

Thanks for you're post. I am on the verge of getting PI. I appreciate that it is a steep learning curve. I think you have done very well. Personally, I think the image with the PI processing is better. It is, by all accounts, a very powerful tool.

Well done.

Sent from my Fone

[pyrasanth]

I've been dipping my toe into PixInsight & the tools are really powerful. I visualize the program as a suite of tools and that helps me get my head around it- easier to think in small parts. A useful tip. When you have produced your master frames ready to be combined by the LRGB color combination run the Multiscale Median Transform tool over the images.

I used the settings in the image below for the Crab Nebula picture that I have just posted. The settings I used are in the first picture & the effect in the second which is pretty dramatic- the left image is before running the tool & the right afterwards.. Images combine much better with little noise & I think it is better to remove on each component frame before the final combination. Feel free to play around with the settings in the tool as you may not have as much noise as me.

Hope this helps.

The tool in PixInsight

The effect

[peroni]

Thanks for you're post. I am on the verge of getting PI. I appreciate that it is a steep learning curve.  Personally, I think the image with the PI processing is better. It is, by all accounts, a very powerful tool.

I certainly don't regret the PI purchase. Great processing benefits seen on my first attempt. The tools within PI are fantastic. 

I too prefer the PI processed image. Better colour, more detail, less noise.

Can't expect too much from only 38 mins of data 

I've been dipping my toe into PixInsight & the tools are really powerful. I visualize the program as a suite of tools and that helps me get my head around it- easier to think in small parts. A useful tip. When you have produced your master frames ready to be combined by the LRGB color combination run the Multiscale Median Transform tool over the images.

I used the settings in the image below for the Crab Nebula picture that I have just posted. The settings I used are in the first picture & the effect in the second which is pretty dramatic- the left image is before running the tool & the right afterwards.. Images combine much better with little noise & I think it is better to remove on each component frame before the final combination. Feel free to play around with the settings in the tool as you may not have as much noise as me.

I forgot to mention that I'd used the LRGB processing method: Extracting the Luminance from the RGB data captured by the DSLR. I did use the Multiscale Median Transform tool (on both Luminance and RGB). It's an impressive tool and your examples show how good the results can be.

[sharkmelley]

I'm a great fan of PixInsight with its very powerful set of tools.

However given this example, if I had to choose one of those two versions of M31 to hang on my wall, I would choose the version processed with Photoshop and Carboni tools.  It just looks more natural.  It's difficult to put my finger on exactly why but the second version has a very "overprocessed" and disjointed look to it that is not aesthetically pleasing to my eye.

Just my opinion.

Mark

[RSM]

I'm currently using a trial version of PI and am likely to shell out for the full licence when that expires. I used two tutorials I found on you tube (search for Richard Bloch and Pixinsight) one for calibrating and stacking, the second for basic processing. I found a good explanation on YouTube on how to use DBE (search for Harry page) which is great.

[peroni]

I would choose the version processed with Photoshop and Carboni tools. It just looks more natural. It's difficult to put my finger on exactly why but the second version has a very "overprocessed" and disjointed look to it that is not aesthetically pleasing to my eye.

I'm split between both images.

I think the eye is drawn to the Photoshop version initially as it is lighter/brighter. But there is more detail and sharpness in the PixInsight version.

I think it is a case of going back to reprocess the PI version. Maybe stretching the luminance layer some more and being a little less hard with the HDR.

Sent from my GT-P5210 using Tapatalk

[Xiga]

Peroni

Those are lovely images, well done mate

I notice we have very similar equipment so i thought i'd asl your advice on how you processed your PS version. I recently took some M31 data myself that was in a similar ballpark to yours (28 mins total of 4 x 7 mins at ISO 800, taken from a dark sky location, absolutely no gradient was present), also stacked in DSS with Bias, Flats and Darks, and processed in PS using Carboni's tools. I'm new to processing so still have a lot to learn, but the best i could manage is the version below. I am finding the colour to be the hardest part for M31.

I will be giving the PI trial a go at some point (just waiting on getting more data first) but would like to improce my PS skills first before moving on.

Care to share your workflow?

[wimvb]

While I am a PI fan, and have no hands on experience with PS, I think you could get near identical images from your data either way. Imo it's the craftsmanship rather than the tool that determines the final result.

Having said that, the workflow in PI appeals to me as well as the fact that with PI I only need one tool for the job and not 2.

Just my € 0,02

Wim

sent from my mobile device

[Maximidius]

I know PixInsight is an awesome tool, but from the two images, I'd go with the first one hands down. 

[davew]

Hi Peroni,

Very reluctantly I have to say I don't like the PI rendition at all. Having said that though, it's your image so you do what you want to do with it.

The stars look like they've been stippled on and there is way more noise ( Depends what you class as noise ) in the background and shows as large red blotches. The fault with the PS process may be the fact it's too blue.

My first instinct was to walk away and say nothing but I decided to have a very quick play and show just how easy it is to turn your PS image into a fairly natural looking Galaxy with a reasonably neutral background. I've left the info pallet to show the background levels in the lower right of the image. I've left the layer history showing so you can tell I've done almost nothing. I even used Auto Colour !! How easy is that. The whole image could be made more red here by not lowering the layer so much.

Hope you don't mind too much and I also appreciate you may detest my contribution

Dave.

[gnomus]

I too prefer the PS version. It is though a little on the green side. Another way of achieving davew's effect might be to try some SCNR green reduction in PI on the PS image. (The SCNR defaults are usually fine).

PixInsight is a great tool, but I found that I started getting better results when I had more data, that is a much larger total capture time. Like you I followed the tutorials on Light Vortex and also Harry's videos. But I think these guys have a decent quantity of data - with less total capture time I was finding that the suggested settings produced too harsh a result. Thirthy eight minutes isn't a great deal. I bet the tutorial settings on 2-3 hours worth of data would have worked well.

But what do I know?

[Pompey Monkey]

I just spent a week at the Gite of a very well known and respected astrophotographer and SGL member: He uses both PI and PS, as well as other software, to produce his stunning final images.

I think each tool has its own strengths and weaknesses. Also different kinds of people get on with different interfaces to software.

I still struggle with all of it!

[MartinB]

To my eye the PI image is heavily over sharpened with halos, unnatural looking stars, harsh dust lanes and chrominance noise.  The PS version looks pretty good other than the colour balance problem.  You could produce exactly the same problems with photoshop!  The HDR tool can be brutal though if it's given too much rope.

I think fine faint details should stay that way, it is nice to be able to peer at an image and find what's there.  

[jtlandreneau]

I've found that if you don't use the right order of operations with PixInsight you can get into trouble fast. I think your Photoshop rendition is better, but the colors aren't quite right. The Pixinsight rendition ( I think ) was too aggressive with sharpening, particularly in way of the galaxy, but the colors are starting to be more on point.

I use both programs in conjunction with each other - I've found myself starting to transistion to PI more and more though... Keep at it!

[Fay]

The Photoshop version is the best, but too blue.

The PI version is over sharpened and too noisy. in the end is it what you are happy with, we all have different ideas

Fay 

[gnomus]

Given the appalling weather, I though I would have a play.  I've just changed the colour balance of the photoshop version...

The main issue is the lack of capture time.  There is only so much you can do with 38 minutes.

[Stub Mandrel]

I like the first image, the second one looks over-sharpened. The stars/globulars in front of the galaxy look like they have dark haloes.

When I have two images like that which highlight different aspects of the target, I layer one on top of the other and mix the transparency to get the best of both. This is the first image with 47% of the second on top, with a few small tweaks to saturation channels. A great set of data by the way.

[pyrasanth]

It is really interesting to compare the 2 images. PixInsight has not delivered the goods on this occasion. I prefer the PS image although the colour balance needs some work. The Pix image is far too red & grainy & something has definitely gone wrong in the process workflow.

I'm amazed at all the creative tools aka scripts that can be integrated in PS.

I only use elements of PixInsight at the moment. My lack of experience prevents me from getting the best from it but I like to dabble & slowly increase my knowledge. I would like someday to produce an entire project in PixInsight but am always drawn back to other tools which makes me feel "safe". PixInsight makes bold statements about never looking back once you've used it but I always have a little Photoshop Devil sitting on my shoulder prodding me with a fork.........

Clear skies to you......whatever processing tools you use.

[StargeezerTim]

I can't compare software because I don't have pixinsight... But I do have Photoshop and time (thanks to the weather), so had a go with your PS image:-

I do like PS and am getting more familiar with layer masks, LAB colour and using star masks. I also like that you can 'see' your changes as it happens and like the range of actions available. Seems quite flexible to me, though, as I say, I can't compare it.

[ollypenrice]

I agree with Mark, Martin and Fay. The Ps version has not reached the limit of the data and the PI version has exceeded it.

I process in both and refuse to be drawn into monogamy or fundamentalism. The PI tools I use are DBE (or sometimes ABE), Background Neutralization, SCNR Green and Local Histogram Equalization.  I can't imagine trying to manage without these.

But when it comes to selection tools give me Ps any day. It is so easy to run a global process like sharpening or noise reduction on a bottom layer and erase just what you want to erase from the top layer using colour select and, sometimes, simply your eyes. You can see what you are doing in real time. Trying to build a mask which excludes just what you want it to exclude is so difficult. Why bother when Layers/Eraser is so simple?

Olly