Image Processing - Order of steps?

[graemlourens]

Hi guys.

The last 2 months i have been reading dozens of hours of tutorials as well as just trying everything out. 

It has been great discovering all the different techniques, and i have been able to handle the most basic parts of stacking, enhancing and so on.

My big question is the order of these different steps.

For now i'm focusing on stars & nebulas, not planetary imaging.

I use

- Deepskystacker

- Lightroom 5.6

- Photoshop CS6

Whats the most logical order?

Stacking the untouched raw images, then continuing on the TIFF in Lightroom / Photoshop?

Or first slightly enhancing the raws in Lightroom (all together) and then Stacking?

I guess i have tried so many different things, that i now am not sure whats the correct oder.

Furthermore i always have the feeling i have to stay on raw editing as long as i can, as with stacking i'm left with a TIFF, that has not got all the information anymore (or is this irrelevant?)

Any ideas well appreciated.

Kind regards, Graem

[swag72]

Hi Graem, as I understand it TIFF's are not like JPEGS and they will not lose info as you save etc.

I stack, then convert to TIFF - Then I work on the TIFFS in Photoshop. 

[graemlourens]

Hey Swag.

Checked out your pictures on flickr, i guess you know what you're talking about Great pictures!

Is there any reason in editing the Raws (contrast, exposure) before stacking or is that contraproductive or irrelevant?

Kind regards, Graem

[swag72]

I'm not sure about raws, I've not used a DSLR for ages.

Prior to stacking don't do any form of processing at all.

That's my way of doing stuff.... Not necessarily right for all, but works for me.

[IanL]

You should calibrate the raw files (apply bias and dark subtraction, then flats) and then stack them.  If you're using DSS and load the bias, darks and flats in to the appropriate places it will automate the whole process in the correct order.

You absolutely should not apply any other adjustments (such as contrast / histogram adjustments)  prior to stacking as (put simply) it will mess up the maths used by the stacking process and you will end up with a worse result.

[graemlourens]

Ok thx a lot, you answered my question!

Just did not want to start getting used to processes that are totally wrong.

Last question: The Edit-Possibilities of Lightroom (to TIFF or RAW) are no match for Photoshop is that right? I guess Lightroom is more an Organizer & Catalogsystem, and Photoshop has the real editing power, so i should rather learn to do all the editing directly in photoshop?

Regards, Graem

[opticalpath]

As IanL said, calibrate the raw images .... but no tweaking before stacking! The purpose of stacking multiple exposures is to increase signal to noise ratio. Any stretching or other similar processing adds noise so you definitely don't want to do that to the individual raws.

Regarding Photoshop etc, one thing that happens when you transfer your stacked image to Photoshop is that you immediately lose some bit depth. Not sure how/if DSS does this but stacking programs usually can save the summed image in floating point or 32 bit format FIT. If you can, it may be beneficial to do an intial moderate stretch on the high bit depth summed image before saving it as a 16 bit TIFF and passing to PS for the remaining processing.

Adrian

[ollypenrice]

I agree with the others but Pixinsight lies in my workflow (certainly for the RGB, which is what you have from your camera) immediately after calibrating and stacking. It isn't free but it's good value and absolutely essential for me even from a dark site. The two PI routines I run on the linear image are DBE and SCNR green. This performs a first class colour calibration and, above all, removes colour gradients which somehow appear from nowhere very obvious! 

After this I'm more comfortable in Photoshop and do my Levels and Curves, layers and layer masking there.

Olly

http://ollypenrice.smugmug.com/Other/Best-of-Les-Granges/22435624_WLMPTM#!i=2266922474&k=Sc3kgzc

[MattJenko]

As a DSLR man I have an extra step in that I find DSS struggles to correctly process the raw camera files (both from Pentax and Canon camera), so I import the raw files into Lightroom and then export them as TIFF files for DSS to use. Once stacked, it generates a 32bit tiff which I then process in Photoshop, but as mentioned above, I have to convert that to a 16bit image at some point as there is a lot of PS functionality that will not work with 32bit images.

[laser_jock99]

I'm not sure about raws, I've not used a DSLR for ages.

Prior to stacking don't do any form of processing at all.

That's my way of doing stuff.... Not necessarily right for all, but works for me.

DSS never seems stack my RAW files (at least mine anyway!) well at all. I get far better results by changeing the RAW files to TIFF then stacking. I even do some basic processing to the TIFFs prior to stacking. Could just be my 'unusual' Fuji RAW files though, it might be different for Nikon/Canon/Sony RAW images?

[AstroAndy]

Here's my method

- Image aquiring

- calibrate (darks flats and bias) and stack in DSS, or native software (1 step)

- With CCD (one shot color) color convert (debayer), in native software, in DSS, the settings will take care of the color conversion; with DSLR, not needed)

- then process the resulting tif in Photoshop

Sometimes, I will precalibrate and color convert my imgs. in the native software, and export them as tifs. These tifs then go into DSS. The reason I do this is because sometimes, with my CCD imgs, with DSS I end up with a well stacked, horrible img (additionally, I don't know why, DSS doesn't take my perfectly good flats for some reason). At the same time, my native software doesn't stack as well as DSS, so I find a median way, so to speak.

Andy

[graemlourens]

thx for your answers.

Have been playing around now for dozens of hours.

Good tipp with the TIFF instead of RAW for DSS, will try that tonight and see if i can spot any difference!

Its processing my Nikon Raw's pretty well (i think), but will compare.

Graem

[graemlourens]

I quickly first exported to TIFF (proPhoto RGB), then stacked. The histogram looks totally different, and for me personally much more difficult to get smth out of it!

Attached are 2 photos (menkalinan, one of my testing stars) , one from raw processing, one from tiff.

Am i doing smth wrong? (i don't mean that its white, without changes) is the initial histogram of the tiff correct or looks strange?

Regards, Graem

[graemlourens]

I revert my comment 'more difficult to get smth out of it'.

It seems that i've got much more color information with the TIFF's with DSS, than with the RAW.

whats you're guys opinion on it? rather RAW or TIFF's to DSS?

Whats the desireable histogram one wants to see in DSS before going onto editing in PS?

[RCompassi]

For some reason, if I stack my raw files, i get a VERY drk picture in DSS tha is nearly impossible to process in PS.

I do this:

Open each RAW file in Adobe Camera Raw (wich automatically erases the hot pixels), and save as Tiff files, then I stack them in DSS, then use PS.

Cheers

[ollypenrice]

I read frequently of struggles with DSS and wonder if it's worth it? AstroArt 5.0 is a demon programme with excellent stacking/calbrating routines which are intuitive, flexible, transparent and fast. If you want to compare Sigma with Median with Average, for instance, it takes no time at all. It will repair columns and rows and has an excellent hot pixel routine. The Sigma clip will deal with multiple sat trails as well.

Olly

[kalasinman]

I quickly first exported to TIFF (proPhoto RGB), then stacked. The histogram looks totally different, and for me personally much more difficult to get smth out of it!

Attached are 2 photos (menkalinan, one of my testing stars) , one from raw processing, one from tiff.

Am i doing smth wrong? (i don't mean that its white, without changes) is the initial histogram of the tiff correct or looks strange?

Regards, Graem

The first is well exposed the second is way over. There is a setting in DSS stacking settings to auto align color channels. Looks like there is dark point clipping. Maybe look at Nikonhacker web site to see if their Nikon firmware would be better, with you Nikon. My d5100 likes it

[kalasinman]

Add NEF (RAW) files, and calibration masters to DSS . register, align and stack, one process. After, click register, then cancel and check scores of and look at individual subs and uncheck low scoring frames. Then click stack again. Open resulting autosave in LR and just save as TIFF. No changes, Then open with PS for levels and curves, save and close PS. Open with LR for finish adjustments.

[dph1nm]

As other have said you should calibrate and stack using RAW not tiff. Tiffs will not work correctly as they have been debayered (and have almost certainly  had non-linear scaling applied). Having said that, if they all have the same scaling applied then the stacking will probably work fine - calibration (darks/flats) probably won't.

RAW astro images *should* be dark - if they are not then you are probably looking at an image which has already been scaled.  My  Canon 1000D only records numbers up to 4096. If I put those straight into a 16 bit image, where the range is up to 65536, the image is going to look pretty dark. If I use a 32 bit image (up to  4294967296) it is going to be incredibly dark!  That is why DSS images saved without any scaling applied look black.

NigelM

[MattJenko]

Olly - people tend to have the view that things from the internet should be free and that software is not something you pay for, presumably as it is not physical. I tend to have these traits as well, as often the open source applications work better than equivalent commercial ones, and some (not AstroArt I notice which has a non-save, non-expiring demo available) do not have a suitable demo version to evaluate before handing over hard currency.

[graemlourens]

Thx for your replies.

The longer i have been thinking the more i guess i have been making one profound mistake.

I am fighting with nearly all stacking-programs with the size of the images. I'm shooting 24mp (around 6000x4000), but i guess this is a huge mistake?

As i do not intend to print this out on a huge piece of paper, i now realize this does not even make sense

In what sizes do you guys shoot?

Regards, Graem

[ollypenrice]

I'm in the same ball park with file sizes using an SXVH36 and a pair of atik 11000s. It isn't a problem for me.

Regarding free and paying software I have to say that for the price of AstroArt 5.0 you get a heck of a lot for your Euro. I don't want to knock DSS but the best imagers whom I know personally don't use it. Ian King advised me to start with Astro Art nearly ten years ago and I did. I still like it a lot and am not looking to change.

Olly

[graemlourens]

DSS handles my Raw NEF perfectly, also 24mill-pix.

But Registax is a different ballgame, everything over 4000x2000 and it tends to freeze.

Autostakkert seems to handle it better

[Jokehoba]

I have a Canon 1100D and started out using DSS and The GIMP. However, DSS wouldn't load RAW correctly and so I worked with TIFF. More often than not everything processed fine, though on occasion I'd end up with a washed out stacked image and it was a struggle to extract anything of value. Actually, like you, I discovered that by doing a little pre-processing of the RAW subframes in PhotoShop I could get DSS to present a usable stacked image! I felt it was cheating - and shouldn't really be necessary - but it worked (though adding yet another step to the workflow).

But as others have said, you shouldn't have to and really you shouldn't! And TIFF files are three times the size of RAW and eat up disk space.

I then discovered that there was a DSS beta version available (see the DSS wiki) which solved the issue with Canon CR2 RAW files. Not sure if there's one for Nikon...

I agree with Olly, DSS is great package and certainly a good one on which to learn the basics. But for me, I became dissatisfied with the results I was achieving and realised there were many more excellent (and more powerful) image processing packages available - but you had to pay for them. Earlier this year I tried out Crag Stark's Nebulosity, seeing a huge improvement in the quality (and colour contrast) in my images. It's not overly expensive and goes a bit further than DSS in post-processing tools. The stacking and calibration processes are logical and seem to give more control of the whole workflow.

Like Olly I now use PixInsight too. A lot cheaper than PhotoShop and built for astroimagers. Loads of very powerful tools. I've been amazed at how much detail/contrast I'd missed from last years images. Busily reprocessing them now! I thought I was going to have to invest in better equipment to improve my images but PixInsight has done that.

Most commercial imaging software can be trialed for a reasonable time. The Nebulosity trial is unlimited but artifacts are introduced to the images degrading the quality. But it's still good enough to try it to see what it can do. Similarly PixInsight give a 45-day fully-functional trial. With all the video tutorials, etc, available, it is more than enough to decide if it works for you. 

You'll learn a lot from DSS though.

John

PS. I'll post some images to show you what PI has done for my images if you like -  but I don't want to hijack your thread. 

[MattJenko]

Word of warning on generating tiff files from RAW for processing in stacking software. It made a right old mess of using my flats. The conversion to tiff does not seem to be linear, in that the generated flats are not a proper flat to the generated tiff light frames, so when run through the calibration routines, the stacking process was overcompensating for the light distortions and dust bunnies. Stacking directly using RAW files resulted in nice clean images, so if you are using calibration frames in your processing, then I would avoid converting to tiff, based on what I experienced the last few days.

Cheers

Matt