Trialling CCDSTack - is it better than DSS for a DSLR user?

[g__day (Matthew)]

I'll be home bound for the next few months - major operation on my jaw next week (requiring bone graft from leg) - yuck. So whilst I'm recovering I thought amongst other things I'll brush up on my image processing skills.

It took me a long, long time to get Deep Sky Stacker working correctly to get great looking images for a Canon DSLR. I would often combine 10, 20 and 30 minute subs. My biggest three mistakes were 1) having goofed the BIAS shots (1/4000th of a second yes - but the camera lens was off - not on - my very, very bad), 2) not pushing the saturation up far enough so all my shots looked faded (about +21% is perfect for me) and 3) not colour aligning and setting the strech points well. Once I mastered this - shots looked great!

So I thought I'd download the 30 day trial of CCDStacker and compare it against DSS. Luckily there are some how to videos (designed for mono cameras) and the instruction manual follows a logical work flow. My first attempt at following all the steps on the video failed! CCDSTacker my subs as having a few hundred more pixels than DSS did - so it didn't like my master flats, bias and darks. So I had to use CCDSStacker to recreate BIAS, Flat and Dark masters so all images would have the same dimensions. After this I could readily complete the workflow - then guess how to add colour.

The last step didn't give me an image anywhere close to DSS quality. I had tried to process 5 * 30 minute subs of Eta Carina taken in April. The workflow was Calibrate (adding dark and flat frames), Auto scale, Debloom, register, Normalise, and Sum the images. By this stage everything looked a faded red colour with strange dark / light zone borders. Then I tried to add colour.

So all up my first go with CCDStacker was just a little better than my first go with DSS! It certainly appears that CCDStacker has alot more advanced functions than DSS; but it strikes me it was aimed at mono camera users.

I ponder has anyone before me done a solid comparison and review of DSS versus CCDStacker to highlight their respective strengths and weaknesses?

At my disposal now I have DSS, CCDStacker, MaximDL and CS4 - and free time! I am trying to figure out the best workflow I can configure with all this software to process DSLR images. I ponder before I start this has anyone else made excellent headway down this path? I believe many of these software versions have some overlapping workflow capabilities - I am trying to judge where to switch from one products capabilities to anothers and most importantly why!

Lets see what I find out!

[multiweb (Marc)]

Sorry but comparing CCD Stack and DSS would be like racing a donkey against a horse. Don't get me wrong DSS does the job for a lot of people but when you get a deeper undestanding of calibration, data rejection and stacking then CCD Stack is in a league of its own. That's all I've been using since day one to process my OSC shots. Maxim is very good too but I'm more comfortable with CCD Stack. I use both.

[g__day (Matthew)]

Marc - doesn't surprise me! As I'm still a beginner on image processing (year two for me). I'd really be delighted to hear exactly where and how you see the software you use in your workflow pull ahead. Kind of like a workflow based analysis of strengths and weakness, e.g. register step - now both X and Y can do this but X is better because ...

For instance you are familiar with both CCD Stack and Maxim (which I have barely touched). What do you prefer in CCD Stack over MaximDL?

Thanks,

Matt

[multiweb (Marc)]

For a start Maxim does everything . And very well I might add. I use it for my AO guiding. But for a beginner its interface is overwhelming. Coming from DSS it scared the hell out of me. I didn't know where to start.

That's why 4 years ago I bought a Licence of CCD Stack instead. I was looking for the next step in processing being a DSS user as all noobs start I guess. Monte Wilson, Rob Greaves already used CCD Stack so I was looking over their shoulder at Ilford and it was a natural progression for me having been exposed to it by these guys.

CCD Stack is a sacking program. That's all it does. IMHO it is THE stacking program out there once you truly understand what data rejection is really all about and don't cut corners any more. I'd say 99% of people out there capture very good data but a good 70% butcher it before it even makes it to the final.

Adam Block's tutorials are the way to get a foot in the door. My basic process (in order) is:

Bad Pixel/Default columns map applied. (sensor specific originally built from a darks stack)
Calibration (dark subtraction, flat fielding, bias)
Debayering (if needed) to Red, Green and Blue Channels

Then for each channel:

Reject Range and interpolate pixels 3 iterations (gets rid of hot pixels previously painted)
Registration to the sub with the lowest FWHM (measured in CCDIS)
Alignment algorithm is CCDIS star pattern then Quadratic B-Spline
Normalization to the sub with the best contrast (measured in CCDIS)
Select sub with best FWHM again.
Reject Hot pixels Algorithm apllied to whole stack
STD Sigma Reject 2 as a factor on the whole stack (cumulative to previous data rejection)
Combine Mean.

This is repeated for each channel that you then recombine to create your color or keep mono if you shoot Ha, etc...

This is the basic approach. Note how its flow is sequential. One step after the other, again and again. And you are in total control of (and understanding clearly) the parameters in each step.

[g__day (Matthew)]

Okay - put simply WoW and thanks! At least i have started the journey!

[marki]

Yep agree with Marc, CCDstack is much more user friendly then Maxim when it comes to processing yet still very powerful. A big stepup from DSS.

Mark

[Tandum (Robin)]

If you are about to spend money, I'd suggest you check out what PixInsight can do.

[g__day (Matthew)]

Taa, I guess this will evolve into a race between MaxImDL vs CCD Stack. Mind you both have capabilities far beyond my current skill level - so a choice between them will be challenging.

I played with MaximDL and its plate solving capabilies on some old frame - and was fairly impressed. Mind you I haven't figured how you'd use that yet to say frame a shot exactly the same way over a series of nights without have to use the mounts hand controller and taking repeat images until the shot is exactly framed the way say your first evenings shot was. I'd love to do that in a more automate manner!

For instance say I want to image M20. I frame the shot roughly with some 30 - 120 second shots until I am happy with the image positioning and confirm I have a good guide star (in PHD). Then I might shoot several hours of 20 minute subs, then move to my next target and repeat this process. I might want to do exactly the same every night for the next 10 days - then stack my best shots. I'd like a more automated way of exactly framing each subsequent nights shots to where the first days shooting was positioned. Can Maxim DL Pro Suite 5.15 with Pinpoint LE do this? Can CCD Stack? Or does more advanced capabilities like this require full Pinpoint and or CCD Navigator or other more powerful software?

Thanks for your knowledge all!

Matt

[Tandum (Robin)]

Matt,
CCD stack does not do camera control as far as I know. CCD Soft is the camera control software you normally get with an sbig camera, or at least I did. I thought you where after stacking and processing software.

I use maxim to capture and the sky to control the scope and dome. I am doing what you want to do with maxim. I connect TheSky to POTH which then starts the mount and the dome. I slew to target using TheSky, but I guess any astro thing will do. While it is slewing I start maxim, connect the imaging camera and the guide camera and turn on the coolers. I then connect maxim to POTH.

Next I do darks for the guide camera and take a shot of the target, 5 seconds binned x 2. I then pinpoint that image. Once it is resolved I sync the telescope in maxim to the resolved position which passes the new co-ordinates to thesky via the mount. I then get thesky to re-slew to target. I am now on target. I now get a guide star and start guiding.

Tomorrow night I can do the exact same thing and be in the exact same spot. If I opt to move the scope a bit to get a better framing of the target, I can open that saved image on another night, pinpoint resolve that image, and slew to that location and repeat capture.

This does not require the bought version of pinpoint, the version that comes with maxim works fine although I've found the fewer stars in the captured image, the quicker it resolves a location. You do need the GSC corrected, which you can get from the pinpoint web site for free.
robin

[g__day (Matthew)]

Robin, et al,

Huge thanks! I got off target there - mixing post processing with my image capture wish list!

I control my gear with The Sky 6 PE -> Maxpoint -> Vixen SS2K drivers under ASCOM 6. Maxpoint acts as a hub, so I should be able to do what you suggest above. I didn't realise that MaximDL -> Pinpoint LE would also update Maxpoint and or The Sky 6's pointing model in an automated fashion - that is way cool!

I'll try all this out then get back to image procesisng!

Many, many cheers to you guys!

Matt

[Tandum (Robin)]

Matt,
I found maxpoint unnecessary and dropped it from my setup. Even with the eq6 mount I have no problem getting on target this way. If it gets lost, I just slew/sync to a known star and pinpoint it to get back on track.

I'll admit the eq6 is a pig after a flip, I've wasted a lot of time getting it back on track as it gets lost, but the em-200 doesn't it's gold there.
robin

[g__day (Matthew)]

Robin,

I found Maxpoint with an 80 star model brings my pointing accuracy from about 3 arc minutes to a bit less than 1 arc minute. Before plate solving a 55 arc second pointing error (when my Canon images a 20 * 30 arc minute slide of sky was desirable). Very helpful for dim targets!

Matt

[gbeal]

Matthew,
I have used CCDStack for quite a while, and liked it immensely. I have however changed my processing software, and am more happy now.
CCDStack, and FITs files from a mono camera work well. Where it didn't do what I wanted was when I tried to feed it many (10 - 12 plus) frames from my Starlight M25c, so effectively an average DSLR sized chip. It choked on my aging PC.
OK, I can't blame entirely CCDStack, but other software runs better. I did ask on the forum (there is a great forum for CCDStack users), and the consensus was "upgrade the PC". The newer version I thought might help, but evidently it won't, and is more designed for the multi-threaded processors, something I don't have.
So, where does this leave you and your original query? For my money, any DSLR user will have difficulty with CCDStack if you use more than about a dozen frames, AND if you have an older PC like me. It just doesn't work. I ended up batching the frames and then combining later.
There is a workflow suggestion in the "user Manual" somewhere, but from memory I opened up the frames, calibrated (made a master bias/flat/dark first), then saved those calibrated frames. Once saved I debayered into the red, and registered, normalized, data-rejected, combined, saved. Then did the green as above, then the blue.
This gave me three frames, R, G, and B. Opened up Colour and "made the tri-colour" image. Bear with me, this is all from memory, but something along those lines.
Worked, and like I said worked exceptionally well, just not with the size of frames and age of my PC.
As has been pointed out, no camera control. Louie has been euphoric about AstroArt 5, maybe rub shoulders with him, if you believe half of what he says LOL, it is the bee's knee's.
Gary

[Octane (Humayun)]

Here's a tip to get your scope pointing at the same location night after night.

Slew to target and solve it and sync to solved position.

Click Calibrate and set to auto and use PinPoint.

What calibrate does (after solving) is tells PinPoint precisely what arcseconds per pixel your resolution is. So, that, when you solve the previous night's reference frame, and hit Goto, it knows exactly how much to slew the scope to line up tonight's object with the previous night's reference frame.

Also, in the autosave window, where you set up your exposure slots, enable astrometry (solving only) option. This then embeds the precise RA/DEC coordinates in the FITS header (assuming you're using a CCD camera). When it comes time to align your stack, you choose the astrometic correction option, which is supposed to be the most precise option of the lot.

Now, load last night's reference frame and solve it.

You will now have the error between what you're pointing at tonight and what you took last night. Once solved, simply click Goto (which contains the proper RA/DEC coordinates of the area you wanted to go to in the first place) and you'll be bang on.

Also, I find myself using MaxIm DL's observatory control stuff to slew to targets instead of my usual Starry Night Pro Plus 6 shenanigans. I have both bits of software connected to the port, though, so when I send slews via MaxIm DL, it also shows where Gemini is in Starry Night Pro Plus.

Any questions, feel free to ask.

H

[RickS (Rick)]

Something I find really useful and cool in Maxim is the ability to overlay an image on the observatory zoom window (see the Options list).

Cheers,
Rick

[g__day (Matthew)]

Humayan,

Can I just confirm a few things to understand if there are any unexpected side effects.

On my Skysensor2000-PC - I set up a 3 star alignment and enter the permanently set up mount into polar aligned mode. I believe I'm well within an half arc minute of the SCP, so the SS2K compensates for DEC drift periodically quite well. I believe the SS2K loses about a second a day, so I reset its clock at the start of each session. I also check the 3 star alignment every quarter or so.

Each quarter - after I have my 3 star alignment for the mount's hand controller I build an 80 star sky model in Maxpoint. Mmaxpoint confirms I am about 20 to 40 arc seconds off the SCP and generally improves my pointing from abour 3 arc minutes to say 55 arc seconds.

So generally each night I image in a quarter I just check the focus is correct with a Bhatinov mask - then get straight into framing my first target. Let's say its M20 - I goto to it and calibrate my autoguider - PHD - usally on 3 second intervals. Then I switch off the guiding and just leave it in passive mode until I have M20 framed exactly where I want it in say 60 - 120 second test shots. This is trial and error a bit and might take me around 5 - 10 minutes to do satisfactorily!

From there I switch the autoguider on, let it stabilise for a minute then leap into imaging runs over a course of nights.

PS My arc seconds per pixel is 0.44.

* * * * * * *

With your method - I'd do the same - up to the slew to M20, in say the Sky6, but then I take shots using MaximDL (rather than DSLR Shutter -> DSUB -> Canon EOS Utilities) so that it rather than Canon's Zoombrowser gets the image. Next under MaximDL I'd plate sove - to get exact coordinates and issue a sync. I presume the sync only updates Maxpoint or does it change the SS2K controller too?

Qu 1. What does the sync do to both MaxPoint and SS2K's sky models? Do I have to worry that this will stuff things up (i.e. like change one of the SS2K's 3 alignment stars and reset its drift compensation algorithm's incomprehensible parameters incorrectly or wierdly)? Or does the sync only adjust MaxPoint's pointing parameters and leave teh SS2K sky models calibration stars unaffected. I remember playing with sync once under CDC or maybe the Sky6 on a nebulae and had pretty wierd pointing behavour until I relaised the mount's 3 star model was confused!

So I solve the current test shot and sync - then I reload night one's correctly framed image and solve it and issue a goto to it to eliminate the discrepancy - which sounds really clever. Any residual pointing error should only be a backlash factor - so very small indeed.

Many thanks if I have understood this correctly.

Matt

[peter_4059 (Peter)]

I've been looking at CCDstack however having read Gary's comments I'm now wondering if there is going to be a problem stacking multiple colour images from the QHY8?

Also wondering if the CCDIS plugin offers the same functionality as the standalone CCDIS when it comes to measuring curvature/tilt etc?

[multiweb (Marc)]

Originally Posted by peter_4059

I've been looking at CCDstack however having read Gary's comments I'm now wondering if there is going to be a problem stacking multiple colour images from the QHY8?

Also wondering if the CCDIS plugin offers the same functionality as the standalone CCDIS when it comes to measuring curvature/tilt etc?

On my old WinXP machine [2GB ram] I used to stack in batches of 10. You need a fast machine to run it. On a 64bit OS it just flies. Nothing compares. The CCDIS plugin is only the star pattern matching algo in CCD Stack. CCDIS is a separate product that does a lot more.

[peter_4059 (Peter)]

Thanks Marc.

Sounds like I need to invest in a new PC before contemplating this. Also, do you need to purchase the CCDIS plugin and the app or do you get the plugin if you purchase the app?

Pete.

[multiweb (Marc)]

Originally Posted by peter_4059

Thanks Marc.

Sounds like I need to invest in a new PC before contemplating this. Also, do you need to purchase the CCDIS plugin and the app or do you get the plugin if you purchase the app?

Pete.

You're best buying CCDIS as it installs the plugin in CCD Stack when you set it up. CCDIS is great to diagnose problems and fix your imaging train so worth the money.