Pixinsight and PS - why both?

[ollypenrice]

Surely if the hardware works perfectly, the imaging is done properly and the software processing techniques are applied all over the image (including using masks for careful selection), the images are more scientifically accurate (rather than aiming for artistically pleasing through artificial blending/layering). I don't see astrophotography as much of an art, rather a scientific endeavor for perfection. Apparently the lead developers of PixInsight have the same opinions!

There's always PixelMath if you want to go crazy anyway.

Not so. Once you stretch the image (which PI facilitates very cleverly) you change the relative brightnesses and any photometric value goes out of the window. Some would say that this removes all scientific value. I'm going to disagree and give an example. The nature of the 'bridge of light' between M51 and its small companion was, at one time, much debated. I feel that this image provides compelling evidence, albeit a bit late, that M51 is a foreground object since you can see the blue of the tidal arm extending beyond the companion.

PI is not more scientific than Photoshop. It is just more mathematical at the user interface. Its assortment of masks, its wavelet sharpening and HDR tools, its noise reduction, etc etc are the stuff of graphics programmes and not of scientific research. Nothing wrong with that, nothing at all. But your data may say that the background sky is reddish in the lower right and greenish in the upper left. Believe it if you like. But Pixinsight gives you the chance to tell the image that it is really all neutral dark sky background. That can't be science!

To do science with a CCD you need a camera without an anti blooming gate. They are available, but you'll get long bleeds dropping down from all the bright stars. However, they are far more truly linear in their response than blooming gate cameras. Astrophotography is whatever you want it to be. I choose to do art with it. For science pick the right camera and don't do anything in Curves.

Olly

PS Ironically, if you want a scientific programme, have a look at Astro Art which prides itself on having a range of scientific photometric tools built in. What's in a name?

[NickK]

Woo Olly - attacking the fundamentals of the sci argument relying on the CCD Your argument is highly logical Captain!

Seriously, most amateur astronomers don't need that.. even for the science. What they do however end up doing is coming up with ideas or observations that fit around those limitations.

I would say that APers are interested in seeing the targets rather than doing science; and that's what a CCD + PI/PS can give you.

[MartinB]

^^^^

Some folks just use PI I notice. But some use both. What's the consensus, pros and cons of each pls? I am using shall we say, ahem, a "less than legal" version of CS3 at the moment. I am not comfortable with that fact.

Steve

Steve, for the sake of all our reputations could you confirm that you have rectified this irregularity?

[kirkster501]

Steve, for the sake of all our reputations could you confirm that you have rectified this irregularity?

Hi Martin, yes indeed. I have removed it and going to buy Pix Insight I think; as I said I was not happy having improper software like that. Having a think about it before buying the PI....

[kirkster501]

I should also add that I removed it in the process of rebuilding my PC and using a SSD. Boy does the machine run faster with a SSD !!!!! . Need to look into Pixinsight to see if it is disk intensive, swapping to disk etc.

Looking forward to giving GIMP a go when it gets 16 bit support as well !

[MartinB]

[NickK]

PI will scale.. give it an i7 with 16GB RAM and a good SSD/HD combination.

And SSD is good for a working location - but you'll want an HD too for big storage. My old MBP had small 120GB SSD, now the 500GB is feeling it with all the images.

[NickK]

Double post..

[Tim]

Personally, I dont care if astro images are only art. I never add or take away stars if I can help it, and never add more detail than is contained in the data, but the aim of the game is a pretty picture, nothing more.

Pictures from PS in many cases end up prettier than pictures solely from PI. Maybe PI pics ARE more 'scientific', but they don't have the same visual appeal in many cases, at least not to my eye.

If you think that PI can cope just as well as PS with layers, then you can't have seen layer masking in action, for me it is the single best feature of PS.

[fwm891]

Download Annies actions from the tutorials and have a play..... (scientifically of course)

[ollypenrice]

Tim, do you really mean 'only art?' Art is amongst humanity's finest acheivements, I'd say. I don't mean Damian Hurst but Rembrandt might get a look in.

I don't aim for pretty pictures. My aim, not my acheivemet, is to produce beautiful and informative ones. For the 'beautiful' bit I rely mainly on nature...

Olly

[Tim]

Ah, my family history on both sides comprises of artists, poets and writers, you're preaching to the choir :-)

http://www.kipling.org.uk/poems_conundrum.htm

This debate always makes me think of this verse by Kipling

[RogerTheDodger]

I sometimes wonder though with all this processing, is it cheating? I mean, it's like turning a plain looking lady ( or bloke) into a Cover model by excessive work on the picture. How much processing if the image is " too much " ?

Sent from my iPhone using Tapatalk

Here is a raw image from the Hubble WFPC2 detector, do you think they should just leave it like this?

Of course, it can produce finer stuff...

[swag72]

I'm all for art and trying to make a pretty picture. Unlike many on here I am a bit of a heathen and have even been known to clone out the odd prominent star or two if it was out of control. Funny thing is, no one even noticed!!!

[IanL]

For my 2p's worth:

- I think we all agree that there is a 'scientific' approach to the basic steps of calibrating, registering and integrating your images. The process (when done properly) is managed by reference to the numbers and you can home in on the best initial SNR in a rigorous manner. Not that everyone does it, but you absolutely can make two images from the same base data and prove that one is a better representation of "reality" than the other.

- After this you are making choices about which features of the image to enhance and which to suppress. There is scientific visualisation, which should involve the application of processing routines in a repeatable and measurable way to achieve the science goal. For example, dealing with all those blurry images from the early days of Hubble before they fixed it was science, not art.

- Conversely, most of us selectively enhance our images in pursuit of pretty pictures. We all use some of the same rigorous processes developed for scientific visualisation, but equally we go further than that. To my mind, the moment you introduce masks or layers to selectively enhance some features and suppress others you are in the realm of art not science. (The only exception I can think of is using layers in PS to create a colour image from multiple filters, but you could just as easily do that without layers outside PS).

- It is pretty clear to me that any result you could achieve with layers in PS could be achieved with image clones, masks and PixelMath in PI and vice-versa. What it comes down to is which processing approach works best for you.

- I think the more 'artistically' inclined prefer the idea of layers so that they can try different approaches to each layer and get immediate visual feedback about the results. The more 'scientifically/mathematically' tend towards a sequence of processing operations that they can apply in a sequence whilst measuring and refining their results. Many seem happy to use both at different processing stages (or even more packages in some cases).

Personally I prefer PI because that is the way my brain works. I like being able to 'program' a repeatable processing pipeline, stepping backwards and forwards through the process to tackle issues that occurred earlier in the process that only become apparent later on. I don't find photo/paint packages lend themselves towards that approach as it is harder to back up and re-apply the same sequence of steps again (both in terms of labour and in terms of repeatability).

[Gib007]

As a physicist, I agree with what is being said. I do have to say though, Photoshop is pretty weak software in my experience. The price paid for it is also not justified for its capabilities. I do think of masks as a kind of layer but when people speak of layers, I picture people painting over their image with a brush. I would totally disagree with that on principle!

[Gib007]

As an addendum, when I say "weak", I mean towards astrophotography specifically. A photographer friend of mine loves it for all his work. I always felt extremely limited with what I could do, even with plugins. PI was an incredible upgrade for me and necessary for me to continue developing as an astrophotographer. To top it off, the price of PI is must lower than Photoshop's (worse still is Adobe's new subscription-based business model).

[Gina]

As a physicist, I agree with what is being said. I do have to say though, Photoshop is pretty weak software in my experience. The price paid for it is also not justified for its capabilities. I do think of masks as a kind of layer but when people speak of layers, I picture people painting over their image with a brush. I would totally disagree with that on principle!

Firstly I don't agree that Photoshop is weak - I've found it excellent! I have never used the paint brush in PS with astro images! The most I have done in that line is to use the clone stamp tool to remove diffraction spikes on stars which are an artefact of the lens iris and shouldn't be there. But I do find PS very useful for enhancing particular aspects of the image and for combining narrow band and tri-colour images to produce a good representation of reality while looking nice as well. NB captures the nebulosity and the RGB is used to get better star colours than NB produces. I don't think this of as cheating - it's a way to capture aspects of the image the would be impossible otherwise.

Having said that though, I do like the colourful images produced by using the Hubble Pallette. Whilst this is false colour it does bring out detail that you wouldn't otherwise see.

Finally I would say - produce results that you like yourself. Yes, It's nice and rewarding to please others and get nice comments but really you want to please yourself and enjoy what you produce.

[Gib007]

Firstly I don't agree that Photoshop is weak - I've found it excellent! I have never used the paint brush in PS with astro images! The most I have done in that line is to use the clone stamp tool to remove diffraction spikes on stars which are an artefact of the lens iris and shouldn't be there. But I do find PS very useful for enhancing particular aspects of the image and for combining narrow band and tri-colour images to produce a good representation of reality while looking nice as well. NB captures the nebulosity and the RGB is used to get better star colours than NB produces. I don't think this of as cheating - it's a way to capture aspects of the image the would be impossible otherwise.

Having said that though, I do like the colourful images produced by using the Hubble Pallette. Whilst this is false colour it does bring out detail that you wouldn't otherwise see.

Finally I would say - produce results that you like yourself. Yes, It's nice and rewarding to please others and get nice comments but really you want to please yourself and enjoy what you produce.

I by no means think combining Hydrogen-Alpha, for example, with RGB as cheating. I actually do this on PixInsight myself and one of my best images of the Orion Nebula is a HDR one with Hydrogen-Alpha used as Luminance for an RGB image. All I'm saying is that having got into PixInsight quite a bit, with the flexibility of the likes of the PixelMath tool, I try to replicate something in Photoshop and all I see is a cumbersome GUI with tools that I want/need simply missing. If I were on the market for astrophotography processing software, I could not by any means justify Photoshop over PixInsight, particularly considering the former is more expensive (considerably so!).

[NickK]

Both tools can replicate the other..

For example - deconvolution here: http://stargazerslou...us#entry1589751

Post #3 is PI and post #4 is PS - resulting in very similar results for M57.

The layering is effectively manually recreating what PI will do - however PI does it with more accuracy and precision in my opinion. In this example - the PS is stretched to a different luminosity but if you look at the detail, there's more in the PI than the PS - due to the approach.

PS is used by professionals for a reason.. although in the case of AP.. I think that PI's specialisation makes it a rival.

[earth titan]

I have to say I like the histogram stretching in PI over the PS. In PS it takes many iterations of curves and levels to get the same effect, where PI does it all in one go. Seems to end up about the same in the end, but PI does seem to keep colours under control more.

The more I play with PI, the more I like it.

I think they end up the same in the end,

Typed by me on my fone, using fumms... Excuse eny speling errurs.

[fwm891]

The biggest problem regardless of the processing tools is the control of the millions to one contrast range created in nature, we are trying to reproduce at best on a monitor or worse on a printed page. Something has to give. Compressing all natures range into the restricted display ranges when producing an 'image' will never be 'scientific'. That's why most science is in the form of spectral and graphical outputs - not very pretty but more useful.

Whether PI or PS or AA5 or... is better is subjective, none are scientific, The CCD's we commonly use even with calibration are limited in what they can output.

So I don't do science with my kit, and I'm still trying to make a pretty picture....

[Gib007]

The biggest problem regardless of the processing tools is the control of the millions to one contrast range created in nature, we are trying to reproduce at best on a monitor or worse on a printed page. Something has to give. Compressing all natures range into the restricted display ranges when producing an 'image' will never be 'scientific'. That's why most science is in the form of spectral and graphical outputs - not very pretty but more useful.

Whether PI or PS or AA5 or... is better is subjective, none are scientific, The CCD's we commonly use even with calibration are limited in what they can output.

So I don't do science with my kit, and I'm still trying to make a pretty picture....

Very true. I have to say that I absolutely love my monitor for this. No other screen I've seen (or print) can reproduce the sheer contrast that my monitor does.