true colour in globular clusters

[Mike in Rancho]

Here are my stacks for Tucanae 47.

https://drive.google.com/drive/folders/ ... drive_link

Thanks for all your help!

Hey Richard, cool. But, it's wanting me to "log in" to Google, so I'm guessing the share settings aren't public to allow download. Anyone-with-link is the standard Drive method of granting that permission.

[Rkonrad]

Try this one. https://drive.google.com/drive/folders/ ... drive_link
I dont really understand this. If this doesn't work I'll try dropbox.

[Mike in Rancho]

No change, Richard.

Drive should work fine though. How are you setting the share permissions?

Whatever you want to share out, file or folder, and from the main window or the tree on the left panel, right click it. Go to Share in the menu, and then select Share again (rather than copy link).

You want it to be like this:

Drive Share.jpg (62.57 KiB) Viewed 63971 times

Otherwise when restricted you would just be able to share with named users/groups within Google, if you have those ID's.

Hopefully that works!

[Rkonrad]

Sorry its taken me awhile to fix this - the correct link to the stacks is here
https://drive.google.com/file/d/1D1xdcE ... sp=sharing, https://drive.google.com/file/d/1Fsjmer ... sp=sharing, https://drive.google.com/file/d/1JKVzxJ ... sp=sharing, https://drive.google.com/file/d/1hdsqQq ... sp=sharing

[Mike in Rancho]

Hi Richard,

Ah I remember this mystery. And no worries, it's a hobby and everyone gets busy. For future reference, all files can be uploaded (or moved) into a single Drive folder which then gets shared out, but all four of your links worked just fine.

The colors still don't quite end up where I think they should either.

Because of the narrow field, star sampling seemed unlikely to provide a proper reference. So I just looked at the background away from the glob as an estimate of neutral balance. So manual adjustments and frequent checking of MaxRGB, as well as looking up some comparison images. For such a nice glob, surprisingly few out there. I believe it's often shown as a fairly golden/yellow target, with a couple medium blue stars in the outer reaches (along with several red/orange). After checking Stellarium though, I question those blue stars, since their B-V values make them rather Sun-like, and thus perhaps ought to be white.

I will say that you captured an astounding amount of small, yet nicely round and defined, faint stars in this glob.

There's some field shifting going on here and I think I even missed some of it the first time around, and cropped insufficiently. That forced me to finally use UnCal1 in Wipe. Later I tried a more severe crop, and that Wiped better, but ultimately the underlying color balance issue remained. There's also a fair bit of either misalignment or CA.

I may try to re-register these files to see if that helps. Otherwise, a lot of Highlight Repair in color, as well as some lowering of the bright saturation, aided in reducing some of the oddball colors, along with red bias decrease. But even so, the central background stars lean towards a light blue.

I'm wondering if there were calibration issues here, particularly with flats, that raise some color problems.

Note that even beyond the field shifting, there's corner overcorrection and a vignetting profile across some channels. Could that lead to different color balances in different parts of the combined image, throwing things off? I'll have to think about this and play around wit the files some more.

RK T47 LRGB auto.jpg (575.92 KiB) Viewed 55225 times

[Rkonrad]

Thanks for all your hard work Mike - really amazing. A few questions:

After checking Stellarium though, I question those blue stars, since their B-V values make them rather Sun-like, and thus perhaps ought to be white.

Stellarium provides you with such information? I never knew (how?)

There's also a fair bit of either misalignment or CA.

I'm not sure what you mean here.
For some reason, some of the frames were off centre even though these are the same coordinates. The calibration frames are all taken care of by ITelescope. I could get the flats and inspect them (or you). FYI the stacking programme is Astropixelprocesser - I pretty much used defaults. I could try again with ASAP.

Anyway - thanks again. Let me know if you need anything.

Richard

ps i just remembered - the white corners - sort of vignetting is because the sensor is too large to accommodate the filter and filter wheel. This the main reason this is a 'free scope" for members ( a great deal!)

[Mike in Rancho]

Hi Richard,

Ah, interesting rig. I think I found that, here? https://support.itelescope.net/support/ ... lescope-33

Well, big scope, big mount, and big old school CCD camera. With big pixels. But, not so big of an image circle?

I wonder where the vignette is happening. The farther from the sensor, the fuzzier I think the falloff should be. But if closer, then likely sharper. I imagine that could still be flat-fielded with the right calibration, but I wonder if it requires a little extra work. If there's too much falloff, perhaps the overall scaling isn't working out properly.

APP is known as a fine stacker, but I don't know what its defaults are. What did you feed in for the full calibration, if I may ask? Off the top of my head, CCD may require dark flats, whereas newer CMOS can use bias on the flats side of the equation.

In any event, though the edges can be cropped, query if a calibration problem with the flats is causing that strong blue cast in the center. There really was no shaking it no matter what I tried. Ultimately I even (after doing a dynamic crop in PI and re-star-aligning the files) ran Wipe on each filter individually at UnCal1, before recompositing, but I'm not sure that accomplished much - or even more than just UnCal1 after composition would have.

I also started inspecting each channel one by one. Siril turned out to be the easiest for that, with RGB composition, and I just left the L out. This was to try to figure out why the stars were multicolored, as if the channels weren't aligning right or had chromatic aberration (CA). Here's where things became intriguing. The green channel looks best. Very nice. But the blue and the red seem to be just out of focus, probably with red the worst. So, spatially, those stars were bloated in red and blue as compared to the green data. Or maybe seeing was worse during those filters, or something stacked wrong?

The stars in the red channel also looked "eggy," as if there were some tracking drift or guiding problems. The unfortunate result of all this is some bright red ringing around the stars, and even red wings to the sides of many. I wonder if blinking through the subs (autostretched) would provide any answers.

But, I tried to do the best I could with all that in mind. In order to not have especially the oval red stars influence the synthetic L, I composited as L, RGB, declining any synthetic contribution. Prior to that I tried using only green as the luminance data, but that didn't really pan out as I hoped. The flaws in the chrominance still showed up when the luminance data (green at that time) was stretched.

I used OptiDev with a few ticks of IFD and then drew a ROI, keeping the global stretch tame, and followed that up with a mild HDR. Have to keep some power in the core! Then SVD, Color, Denoise. That's it.

Color was of course the trickiest, and I played around with the saturations and the highlight repair (max) in order to try to achieve some nice coloration without revealing the aforementioned flaws too much. Basically I tried to hide them. As such, the colors here are somewhat dull as you can see.

For the "true color" balance, I just used the sliders manually with frequent checks of MaxRGB, and aimed myself at some landmark stars I could find the B-V values for in Stellarium (listed right below magnitude when you click on a star).

How accurate are B-V values? Dunno. And it's not the only color index. I've seen somewhat different results, color wise, for the same star, if measured in B-V versus U-B or the others. https://en.wikipedia.org/wiki/Color_index

This guy came up with a more complete chart, which I consulted for better gradation in between the rows of the Wikipedia chart. http://www.vendian.org/mncharity/dir3/s ... tails.html

Of the handful of decent Tuc 47 images out there, I think this APOD seemed to have a pretty good balance that concurred with the B-V values, and didn't blow out the saturation or turn the white stars blue. https://apod.nasa.gov/apod/ap110116.html

I suppose my lightly saturated rendition is kind of boring, since Tuc 47 is supposed to be a jewel box of the southern skies, right? Maybe one of our other regulars can figure this one out better.

RKonrad Tuc47 test ST9 2A.jpg (666.94 KiB) Viewed 54982 times

[Rkonrad]

Thanks for this! ill need a bit of time to digest what you've written. Meanwhile I could at least re shoot the red and blue frames and then crop the image even more aggressively.

Generally this system produces good colours with other targets.

Thanks Richard

[decay]

Hi Richard & Mike,

I tried to process this shortly after Richard posted the links. But the result was really hideous and so I thought posting it would be of no use.

I also noticed this obvious vignetting but decided to focus on the main problem and so I went on with cropping and aggressive wiping.

But whatever I tried to do to get a decent colouring, nothing helped.

So yes, I too would say this is a problem specific to this data set. You did a lot of research Mike, and I’m sure you brought out the best. Maybe your work can help to find out what went wrong with image acquisition?

I too would try to re-shoot this and not spend more time and effort to process this problematic data set. But of course, it would be helpful to know what went wrong before starting the next session.

Let us know, how you get on. Good luck!

Best regards, Dietmar.

[Mike in Rancho]

Glad you took a look also, Dietmar.

In an attempt to get a passable but more saturated version, I tried hammering at it some more, but without any success that I am happy with. I thought that a Color Tame Dering in Shrink, along with a slight Layer blur on the same mask, would help. And it sort of did, but not fully, and it starts getting to the point where, perhaps on top of the heavy highlight repair already applied in Color, a disconnect just really builds between the luminance and chrominance.

I'll try to look at a few other things to see if I can discern how much is baked into the datasets versus my crummy processing.

Richard do you have any other data from this rig, including some that you think worked well? I wouldn't mind examining that. Not so much for the target, but to see how the star stretching and star color reacted in a different acquisition session.