OSC vs mono+RGB for near-live viewing of open clusters

[Martin Meredith]

Since obtaining RGB filters I've been waiting for an opportunity to compare the mono Lodestar X2 + filters with the colour Lodestar. Last night was the first opportunity to do so. Caveats: (i) these should to be treated as a pilot and I'll redo them when the moon is out of the way; (ii)  the colour Lodestar I have is the first generation model so is less sensitive.

My aim was to check out star colour rendition (and the ease of adjusting it using LodestarLive) for the two manners of acquisition. Open clusters with a mix of young blue and older yellow/orange stars seem appropriate so I chose a couple of clusters in Cassiopeia: M103 and NGC663.

I used 15s subs throughout. For the mono, I stacked 4x15s for each of RGB (i.e 12x15 in all) while for the OSC I stacked 12x15. For each camera I collected 21 x 15s darks. I adjusted the image 'live' to get the closest colour rendition as I could using high quality images in SkySafari as a reference.

First, M103. The left part is from the mono+RGB, and the right from the OSC.

Next, NGC 663. Again, left mono+RGB, right OSC.

As you can see, there is some colour in the mono+RGB, but no matter what I did I couldn't extract much if any blue. For the OSC I could barely extract any colour at all, and believe me, I tried! The mono+RGB is capable of reproducing decent star colours in LL as this image of M7 I took a while back shows:

The near-full moon in the same quadrant of sky won't have helped, so the comparison is not very meaningful, although it is interesting that the mono+RGB did at least manage to extract some (correct) colour. I'm new to this, but is it the case that moonlight, as reflected sunlight, boosts the yellow and masks the blue? Any comments on possible differential effects of strong moonlight on blue rendition, or on the methodology, would be much appreciated.

To be continued in a week or so...

Martin

[Dom543]

Martin,

Very interesting experiment.

The first question that comes to mind is if there are any young blue stars in M103 and NGC663?

The second observation is that I see two distinctly different star colors among the faintest small stars: Bluish grey and yellowish white. This is very noticeable on the OSC images but I am not so certain about it on the RGB image. These may be only spurious colors due to star sizes getting close to the pixel scale. I am curious to hear, if others also see this variation or if it's just that my vision gets biased by my mind.

Clear Skies!

--Dom

P.s. Here are examples of colors of faint stars that I believe to see. The crop is from the right side of Martin's image. Please check the colors on the original image as my as my red pointers may interfere. --Dom

[Zakalwe]

Don't forget that an OSC has double the pixels of the red or blue due to the Bayer matrix. An RGB combination will assign equal values to R, G and B (I'm ignoring any sensitivity difference to the various frequencies of the underlying sensor).

[Martin Meredith]

Dom, good questions. I looked at a number of images from well-known astroimagers such as Robert Gendler, as well as the coloured DSS images (via ALADIN) and these suggest that for both clusters the bulk of the brighter members (20+) are what I would describe as electric blue. I'm assuming that the DSS images at least are colour-calibrated based on G2V stars (which is one technique). 

I think some of the weird colours in the OSC case (mauve!) are just incorrect and result from me trying to set the main stars to something reasonable (and failing). That there appear to be fewer such cases in the RGB image suggests it is doing a better job.

One thing I should have done is use the L filter with the OSC camera so that at least the overall passband matched the RGB case. I actually shot the OSC through the C filter (just to avoid removing the filter wheel). Some of the difficulty could stem from excess light pollution outside the bandwidth of the L filter. It is a little unfair to the OSC to have to deal with this. 

I'm looking forward to repeating the test without the moon. I only went ahead because I read somewhere that moonlit nights are still OK for collecting colour data...

Martin

[Zakalwe]

I'm looking forward to repeating the test without the moon. I only went ahead because I read somewhere that moonlit nights are still OK for collecting colour data...

Martin

Moonlight will wash out colour data. The only thing that, in my limited experience, is worth collecting when the Moon is bright is hydrogen-alpha. Even that will get washed out unless you are imaging in a different quadrant of the sky.

[Martin Meredith]

Thanks Zakalwe. I was hoping someone in the know would confirm that.

Re Halpha, I came to the same conclusion last night to end the session -- lots of Ha and OIII on M33 (even nearer the moon, about 30 degrees away). This animation shows the very poor luminance signal and the Ha/OIII superimposed. Ha mapped to R, OIII mapped to G and B, then a bit of L mapped to G and B too. The multispectral LodestarLive does allow for some great tricks. This animation simulates precisely the live effect of checking and unchecking the 'DisplaySelected' checkbox -- a great tool for outreach.

Martin

[nightvision]

Interesting comparison, I see more colours in the OSC images.

[nytecam]

Interesting experiment Martin and great results:-) I assumed the L filter is an IR Block - otherwise the extended IR sensitivity of the Lodestar will bias results especially with reddish stars recording from experience with the opposite cyan bias!

The reproduced star colour you show is very subtle and perhaps better shown via enlarged extracts. It could be that it's all too small seen on my tablet! Keep up the good work but I'm pleased the colour Lodestar fairs quite well in your comparison shots. The short coming with OSC is under sampling with stars falling onto too few pixels in the CYMG matrix :-)

Nytecam

[Martin Meredith]

Interesting experiment Martin and great results:-) I assumed the L filter is an IR Block - otherwise the extended IR sensitivity of the Lodestar will bias results especially with reddish stars recording from experience with the opposite cyan bias!

The reproduced star colour you show is very subtle and perhaps better shown via enlarged extracts. It could be that it's all too small seen on my tablet! Keep up the good work but I'm pleased the colour Lodestar fairs quite well in your comparison shots. The short coming with OSC is under sampling with stars falling onto too few pixels in the CYMG matrix :-)

Nytecam

Thanks. Yes, the L filter is effectively IR/UV blocker. 

Hope to repeat this tonight before the moon comes up.

Martin

[RobertI]

This animation shows the very poor luminance signal and the Ha/OIII superimposed. Ha mapped to R, OIII mapped to G and B, then a bit of L mapped to G and B too. The multispectral LodestarLive does allow for some great tricks. This animation simulates precisely the live effect of checking and unchecking the 'DisplaySelected' checkbox -- a great tool for outreach.

animation.gif

Martin

That's a really nice little demonstration of Ha over luminance. i have always wondered what an Ha only image of a galaxy looks like and this (almost) shows it!

Rob

[Martin Meredith]

OK, to continue. I collected more data on M103 last night prior to moonrise, with SQM readings hovering around 20 (not bad). I used the same approach and exposures as earlier. One difference is that I used an L filter (which restricts the passband to the same as the combined RGB, reducing any IR/UV influence) for the OSC Lodestar. I manipulated the images live to get the best colour rendition for the red giant in the centre of the cluster (the one labelled B in the later images).

The above shot shows the data from 2 nights ago under a near full moon (top row) and from last night. The mono camera results are in the left column. (Spot the one that requires some mental gymnastics…)

Here's the same zoomed in and rotated where necessary, but otherwise untouched.

I've labelled some of the stars to make it easier to discuss their colours. Some of the information (and the best photo) comes from p337 of the Atlas of the Messier Objects by Stoyan et al (a brilliant book).

A is the double star TDS 1912, mag 11.1 type M6, secondary mag 13.6 but only 1.8" distant (so not visible here). So A should be reddish.

B is the primary of the variable double S.W. Burnham 22; D is the other component. B is a red giant of type M1.

C is the double Struve 131 noted in BCH by Smyth as having colours "straw and dusky blue";  brighter member is type B3. More recent works suggest it the brighter member of mag 7.2 is a white blue giant of type B5, with a blue companion (in direction of and a Be companion twice as far in the same direction.

STI 232 is mag 10.8, type G8, with mag 13.2 companion at 11.2" separation; G8 is a yellowish colour.

The renditions from last night are much better than those captured under strong moonlight. I rather like the mono+RGB version. Using the controls available in LodestarLive (essentially hue and saturation once histograms for the three channels were lined up), it was far easier to manipulate the mono+RGB image. But I think the OSC is also picking up some of the colour distinctions, though without much saturation. I had the turn the saturation control down for the mono+RGB case to avoid the lurid results you can see from two nights back (which perhaps also explains the overall yellow cast to the background).

Still not much blue coming through from the younger cluster members…. but the quest continues.

Martin

[Astrojedi]

Martin,

You could be losing some dynamic range due to the combination of higher read noise (~10-15e) and an obstructive color matrix for the original Lodestar C which may explain difference in color between the OSC and Mono RGB. Ultimately you cannot stack what you don't capture.

One way to test this is to try slightly longer sub exposures say 25-30s with the OSC if your mount can handle it (keeping the total integration time the same) and see if you get better color.

Hiten

[HiloDon]

Martin,

You could be losing some dynamic range due to the combination of higher read noise (~10-15e) and an obstructive color matrix for the original Lodestar C which may explain difference in color between the OSC and Mono RGB. Ultimately you cannot stack what you don't capture.

One way to test this is to try slightly longer sub exposures say 25-30s with the OSC if your mount can handle it (keeping the total integration time the same) and see if you get better color.

Hiten

Hiten and Martin,

I think Hiten's on to something about the dynamic range, but it could be that Martin's subs are too long and he is getting saturation of the bright stars. I know I have posted this reference before, but Keith Wiley does a good job of explaining dynamic range and proper sub exposure in this reference. 15 seconds seems long to me for a star cluster.

http://keithwiley.com/astroPhotography/imageStacking.shtml

Don

[Martin Meredith]

Don, you might be right. I checked the FITS for one of the OSC subs and the bright stars are definitely reaching saturation. The colour that is coming through is on the fringes where the PSF isn't saturated. Looking at the fainter non-saturated stars I do indeed see more colour (in the sense of say B values > R/G). So, shorter subs. You're looking at the right person ;-) OK, when's the next clear night around here?

Hiten, thanks for the suggestion too. My mount is good to about 30s, or 45 if I'm pushing it, in alt-az. At some point I will learn to polar align but the need hasn't yet arisen.

Once this is sorted, the next step is to find a quick way to calibrate (e.g., by clicking on a star with known spectral type at roughly the same elevation). I'm a great believer in getting as close to 'true colour' as possible, although I know that is open to a lot of interpretation. It is fascinating to look at various renditions of M103 and see the numerous different star colour assignments...

We don't see many shots of OCs in EAA because I think the eyepiece view is often better, but my aim here is to see if we can bring open clusters to life and examine some of their varied forms and compositions, so if I can get a good yellow/orange, blue, red, white discrimination I'll be happy for now.

Martin

[Astrojedi]

Martin,

I have never had an issue of loss of color due to longer exposures, only shorter exposures but Don does present an interesting perspective. Lodestar C has a 50k full well so I am surprised it is getting saturated so quickly.

Only way to find out is to experiment (isn't that the fun of it ).

I love seeing the colors in the stars in open clusters... attached is a live view image of C14 double cluster I took over the summer.

Hiten

[Astrojedi]

Forgot to mention (cannot edit a post on SGL)... this is 5x5s exposures with ASI224 using a Orion 50mm f3.2 guidescope (actually I was inspired by Don's captures to use the 50mm for live viewing...).

[Martin Meredith]

OK, call me obsessive, but I've assembled some decent colour data on M103 members. Fortunately, this is a well-studied cluster. Using the data here [1] from paper [2], I've taken the BV colour values and converted to RGB using [3] and plotted them, with magnitudes scaled. The plot is down to mag 19. Note that the plot only shows confirmed cluster members, so some of the brighter members that are in my shot are field stars and do not appear in the plot.

Here's a portion of the Lodestar Mono+RGB shot and the corresponding BV colour data from [2]. I've chosen fainter, non-saturated stars. The colour correspondence is quite reasonable, I think, but could be improved further by calibration. 

If anyone enjoys looking for correlations the complete plot is here

M103.pdf

and should be compared with this image

The horizontal lines mark more or less the extent of the image. It took me a little while to mentally align the two, largely because of the missing field stars (for instance, the bright red giant in the centre, and the bright blue/white star to its NW are field stars). Anyway, I hope this plot will be useful to anyone wishing to check their own shots of M103.

Martin

[1] http://vizier.u-strasbg.fr/viz-bin/VizieR-3?-source=J/A%2bA/349/448/table4

[2] http://arxiv.org/pdf/astro-ph/9908059v1.pdf

[3] http://stackoverflow.com/questions/21977786/star-b-v-color-index-to-apparent-rgb-color

[DonBoy]

When imaging star clusters with the Mallincam video cameras I found that color was greatly enhanced and less star bloat was present when the Contrast was set to 50% and the Brightnest was adjust around this contrast setting of 50%.  I would imagine this would hold true for the LodeStars also.   Additionally from observation the cameras using the 418/419 sensors have the best color, with the 428/429 as 2nd best and the 828/829 with lower color response.  

Don

[nytecam]

Hi Martin - I've taken the liberty of enhancing your M103 image [below] by a factor of x200 % to show the star colours more clearly.  They seem a near perfect match for your reference - well done.

I appreciate, as you state earlier, you aim for 'true' colours but what, if we can't actually see these colours as they are too faint for the eye to perceive, is truth !   You're entering the pro world with your excellent investigative work and sometime a little exageration [of colour here] gets the point across more clearly.

This is one of the finest pieces of work I've seen on the whole forum.  Keep up the good work;-)

Nytecam

[nightvision]

For a given exposure which is the best Lodestar, the mono+RGB or OSC?

[Martin Meredith]

Thanks very much Nytecam. I agree that the concept of 'true' colours is a bit of a stretch (though maybe with a much larger aperture some of these hues would be visible to the eye). Having colours of say hot blue and older yellow and the occasional red giant in the right ballpark is what I'm aiming for. I was quite surprised how close the colours for M103 turned out though and await the next clear night to go for shorter exposures with the Lodestar C.

Tony, I think it depends on your criteria for best. The OSC is a easier to use (no fiddling with filters), but I'm guessing the colour rendition for the mono+RGB will be more straightforward to adjust. Tests continue...

Martin

[Martin Meredith]

OK... probably the final images from me for M103 captured an hour or so ago, this time taking up Don's suggestion of using shorter subs. 

Moonless, SQM around 19.6 when collecting the mono+RGB, and down to 20.25 when doing the OSC. x^0.25 compression used in both cases, and separate darks collected for each camera. These are with 5s subs throughout. For the RGB I had 1 minutes worth in each filter, and for the OSC 3 minutes worth for comparability (shot through a Baader L filter to limit the bandwidth to the same as the RGB case). Focus was adjusted in both cases using a Bhatinov mask on Caph.

As before I did the best colour adjustment I could, aiming to get the central red giant about right.  

First, mono+RGB. Here, the adjustment was pretty easy, with a low saturation value.

Then the OSC. Here it was more difficult to achieve the right balance. I had saturation nearly at maximum in this case, and the 3 channels in the histogram were tricky to line up as they had different shapes, unlike in the mono case.

The shorter subs clearly work better (thanks Don!).

For myself I prefer the mono+RGB colour rendition, over the course of all 3 nights, from full moon to rising moon to moonless. Even though there is extra work involved in shifting the filter wheel, it is no more than a few seconds and I think a price worth paying for the final view, at least on open clusters, particularly since the extra time is saved at the colour adjustment stage. Mono+RGB also leaves plenty of saturation headroom unlike in the OSC case. But I can imagine others preferring the ease-of-use of the OSC. 

If I get time and enough clear nights I'll continue with other types of object.

Martin

[RobertI]

I prefer the mono+ RGB as the differences between differently coloured stars is more pronounced without looking unatural.

As an aside, I think that is the best colour shot of an open cluster I have seen with a Lodestar - it's great to know that there is still so much I can look forward to doing with my humble little camera!

Rob

[Martin Meredith]

Very kind of you to say so, Rob, though collimation could have been improved and to my eye it has a bit of green cast. Short exposures and nonlinear scaling seem to be the order of the day for open clusters. I'm looking forward to exploring more OCs -- plenty to go for and I think with colour their individual personalities shine through.

Martin

[Paul81]

Great comparison Martin, the mono+RGB certainly shows the colours much better.

Will be interesting to see how this compares to the RGGB sensor of the ultrastar-c as CYMG is a devil to work with which is why I wonder the colours aren't so vivid.

Sent from my iPhone using Tapatalk