TMB and FSQ/TMB hybrid image of IC443

[neilfleming]

Finally, some data to show! It has been a while. I've been off and "dialing in" my TMB 203 F/7. Here is an H-Alpha shot of IC443, consisting of both the FSQ and the TMB data:

Exposures:
- FSQ: 12x30 minute subs plus 14x5 minute subs, totaling 7 hours, 20 minutes
- TMB: 16x30 minute subs plus 29x5 minute subs, totaling 10 hours, 25 minutes
- Overall: 17 hours and 45 minutes

Processing:
This image is a composite or hybrid of the data from the FSQ 106-NSV F/5 (530mm) and the TMB 203 F/7 (1421mm).

As you may recall, I am in the process of "dialing in" a longer focal length scope. (The TMB 203 F/7. For this image, I did not as yet have the flattener, although I did put it into play recently.) Along the way, I have captured data of IC443 with both that scope and my FSQ, and am experimenting with optimal ways to combine the data. (A.k.a. Rob Gendler?s ?hybrid? approach.) I wanted to combine the detail from the TMB with the broader field of view from the TAK FSQ.

The thought was that by upsizing the FSQ data, which at 3.51 arc seconds per pixel (asp) with the 6303-based chip was very much undersampled, BEFORE subexposure registration would provide a better result in terms of assembling the FSQ image. There is a gain in anti-aliasing as the result of upsizing the initial data. Upsizing at 200% gave 4 pixels for each original. This improvement not only was apparent when presented at 100% original size, but it also did allow for the presenting of the data at a 150% size for better layering in of longer focal length data, as it was acquired. So, the final result included the FSQ widefield data being presented at 150% of the original size.

Another interesting attribute of this series of images is that virtually all of the bloom removal was done in CCDStack. (The FSQ 30-minute data had blooms stretching 2/3rds the way across the image.) This was done by processing the 5-minute FSQ subs ahead of the 30-minute subs. I first produced a "5-minute sub master", via a SUM combine. Then, while processing the 30-minute subs, I made the blooms "transparent", then incorporated the 5-minute sub master in the stack to allow for the real data to poke through.

I broke this into three main sub-projects:
A. Processing of the FSQ data at 150%
- Processing of the 5-minute data came first
- The 30-minute data was processed second
B. Processing of the TMB data at 100%
- Processing of the 5-minute data came first
- The 30-minute data was processed second
C. Downsizing of the TMB data to 56.84%, and layering in that data into the FSQ result

A) Workflow for the FSQ data:
CCDStack:
Process the short, 5-minute, Ha data:
1) Calibration with darks, flats, and bias frames
2) Bloom rejection
3) Impute (minor) bloomed pixels
4) Rotate all subs to the same object orientation
5) Upsize to 200%
6) Image registration (bilinear)
7) Normalization (Auto)
8) Data rejection (Poisson sigma)
9) SUM combine. The SUM combine is required so that this 'mini-master' can be used to present data underlying the blooms from the longer-duration subs later in the process

Process the longer, 30-minute, Ha data:
1) Calibration with darks, flats, and bias frames
2) Bloom rejection
3) Set bloomed pixels to, "MissingVal", so that they are, in effect, 'transparent'
4) Rotate all subs to the same object orientation
5) Upsized to 200%
6) Load in the prior 'mini-master' of the SUM combine of the 5-minute data
7) Image registration, to the "mini-master" as the reference image (bilinear)
8) Normalization (Auto)
9) Data rejection (Poisson sigma)
10) Mean combine the achieve the final result
11) Downsize to 150%, in order to allow the deconvolution to happen without running out of memory
12) Deconvolution, Positive Constraint, 20 iterations
13) Save as a scaled TIF for importing to Photoshop

B) Workflow for the TMB data:
- Same as the FSQ workflow, excepting all processing was done at 100%
- The final TIF was registered to the FSQ data in Registar

Photoshop CS2:
- Noise removal (NeatImage)
- Shadow-highlight to bring up the background data, tone down the highlights
- Noel Carboni's Photoshop action for "local contrast enhancement"
- Overall 'contrast curve'
- Multi-layer High Pass filtering at differing pixel radii
- Sharpening
- Layer in the TMB data
- Use levels, curves, and appropriate layer mask to blend the two together

My main web page for this object has all of the various different sizes and flavors, including smaller sized versions for faster downloads:
http://www.flemingastrophotography.com/ic443.html

Here is the 100% size of the FSQ/TMB data:
http://www.flemingastrophotography.c...a-6-100pct.jpg

Here is the 100% size of the TMB-only data:
http://www.flemingastrophotography.c.../IC443-TMB.jpg

As always, constructive feedback greatly appreciated!

I hope you enjoy...Neil

[budguinn]

Neil, I saw this on the SBIG site and was hoping you'd post it here. It is one of the finest images of this area that I've seen. Your narrowband stuff is really some of the very best on the internet. The blend of the two different images is very nice...totally transparent with no discernible edges.

a beautiful image.

[Bob_riggs]

Neil,

You do "keep raising the bar" with your work. You give us inspiration to shoot for more and more, better and better, all the time. Thank you...

Cheers, and CDS!

Bob

[galacticphoto]

Excellent image and well deserved thanks to the great amount of time spent. Very complete, and useful explanation too.

Robert

[neilfleming]

Many thanks Bud, Bob, and Rob! I really appreciate your kind words.

Regards...Neil

[BT Technologies]

Stunning image Neil. Have you gotten a lot of use out of that TMB 203 yet?? I heard from Kirk that you were have a few issues with it??

[neilfleming]

Originally Posted by BT Technologies

Stunning image Neil. Have you gotten a lot of use out of that TMB 203 yet?? I heard from Kirk that you were have a few issues with it??

Yup. I was initially getting some spikes on the stars. Funny, I have not seen ANY other photos on the internet with this scope!?

It was simple, turned out to be either the spacer tabs between the lens elements, or turbulence effects from around the spacers causing the little spikes. I had a pal do up a 2mm mask, and now the results are outstanding. I'm very pleased.

...Neil