Star Tools selection for multi-band filters

[Lawrence]

Until recently I was using my M25C OSC camera for imaging, using a standard light-pollution filter. ST processing was therefore 'normal'. I have obtained a multi-band (MB) filter which comprises Ha+Hb+S+O. I have noticed that there is an option to selection Narrow-band imaging during the early stages of ST processing. However, when I selected that to see how the colour was affected, it seems to be the reverse of what I might expect. I have taken a few images of M57 and expected the outer red and inner core blue, but in fact the result appears to be the reverse. Should I perhaps process the MB as for a normal OSC camera and adjust as required?

All advice welcomed.

Lawrence Harris

[Lawrence]

I am hoping that including a sample image from the processing might stimulate some kind soul to help with advice on this query. My puzzle: what colour selections should I do to get nearer true colour when using a multi-band filter with my M25C OSC camera?

TIA

Lawrence Harris

[happy-kat]

There is a recent post where Ivo talks about tri band filters
viewtopic.php?f=10&t=1696&p=7332&hilit=tri#p7332

How are you loading your image?

[Lawrence]

There is a recent post where Ivo talks about tri band filters
viewtopic.php?f=10&t=1696&p=7332&hilit=tri#p7332

How are you loading your image?

The usual way: load+OSC (option 2).

I have had to pause for the moment due to the 'compose' suddenly sticking on.

Lawrence

[happy-kat]

Are you using Open to load a single fits file, or are you using the compose module to load the three channels from your fits file?

Note sure if what you are trying to say is that the compose module has frozen locked and you can't cancel or just continue using it.

[Lawrence]

Are you using Open to load a single fits file, or are you using the compose module to load the three channels from your fits file?

Note sure if what you are trying to say is that the compose module has frozen locked and you can't cancel or just continue using it.

The situation is this: until a day or three ago I would load the FITS file via open+option 2. The M25C is behind a quadband filter so effectively 'believes' that everything is normal. I have been taking pics this way for several nights while we had clear skies. I have mostly captured galaxies except for 1 nebula vdB111. I have then processed as before with ST, using open+option 2. Option 2 lists OSC so that seemed the most reasonable choice. A few days back I updated ST to the latest alpha version. It was apparently from here that it all changed and 'compose' switched itself on - speaking from memory because I didn't need to notice exactly when things went wrong.

My concern happened when I did a series of MB + M25C on M27 and then processed it. To my surprise the colours were essentially reversed with the outer layer blue (instead of red) and the inner core a different hue - see picture. This is when I stopped. But then I found myself in 'compose' mode with no colour and not able to get out of it. Yesterday I discovered that if I load the MD FITS image using the first option instead of option2, I get my colour back - but still 'reversed'. There must be a reason for the wrong colour so here I stop until I know how to fix it. Option 1 says unstretched etc and that also qualifies as a valid 'load/open' option.

My ST processing is at a basic level and I have yet to routinely use the various 'quality' options. I only use bin-crop-wipe-a/d so far. The compose option has me baffled doing processing in b/w with a known colour image. That's a new learning curve.

Thanks for any help.

Lawrence

[admin]

Hi Lawrence,

It appears the chosen Bayer Matrix ("RGGB" according to the filename you uploaded) is incorrect, as the channels indeed appear to be swapped. Try BGGR instead.

There are two ways you can process duo/tri/quadband filtered data;

1. You can import as normal and treat it like any other DSLR dataset.
2. You can use the Compose module to create and import the necessary datasets to process a "pure" HOO bi-color, which can be remapped to other popular bi-color schemes in the Color module.

It's probably best to keep things simple for now and keep with #1.

[Lawrence]

https://www.dropbox.com/s/6okjjygtxa5em ... 9.fit?dl=0

https://www.dropbox.com/s/a5cr1358qzovx ... r.fit?dl=0

I have investigated the collection of stacked files with various labels such as rggb and bggr. The picture that I posted was in fact rggb and not bggr as I typed - sorry Ivo. I have reprocessed the masters using both registrations and sadly now they give identical and wrong colours. Frankly I do not know what I did to get the correct one! I have still got the master stacked image that shows the correct colour - but I did not put any specific label on it and the FITS data is unhelpful I have uploaded the raw stacked image obtained using bggr which is the recommended registration for the M25C OSC. The images were obtained using the OSC fitted with a quad-band filter. I had upgraded DSS and found that it reset all the previous settings (!) so I have been through them all and re-selected bggr for the M25C. Once I can reliably re-create the M57 image with outer red area as has always appeared on my pre-filter images I will then be re-assured that the colour settings are correct. I am also uploading the M25C stacked image that shows the correct colour - M57 4 MB (MB means multi-band).

Meanwhile if anyone can analyse the dropbox file I would appreciate it. I would be happy to re-run the DSS program with any new suggested settings.

TIA

Lawrence Harris

[admin]

Hi Lawrence,

Thank you for sharing the datasets. The one thing that I find a little bit odd, is some very pronounced mottling. Something is not quite right here. Did you dither between frames? I can also spot denabering ("zipper") artifacts, pointing to the use of an unsuitable debayering algorithm (did you select AHD by accident?).

Have a look here for recommended DSS settings and here for tips/do's and dont's with regards to acquiring good datasets. For example, I note an apparent lack of flats with vignetting and dust donut readily visible. Flats are really not optional...

In 1.6, you can remap the channels on-the-fly using the Matrix option, which I did. Mapping RGB to BRG seems to do the trick;

M57 5m x4 bggr (2).jpg (16.7 KiB) Viewed 6461 times

Simple process flow as follows;
--- Auto Develop
To see what we got.
--- Crop
Parameter [X1] set to [1270 pixels]
Parameter [Y1] set to [881 pixels]
Parameter [X2] set to [1709 pixels (-1323)]
Parameter [Y2] set to [1198 pixels (-818)]
Image size is 439 x 317
--- Auto Develop
Parameter [Ignore Fine Detail <] set to [2.8 pixels]
Parameter [RoI X1] set to [170 pixels]
Parameter [RoI Y1] set to [119 pixels]
Parameter [RoI X2] set to [291 pixels (-148)]
Parameter [RoI Y2] set to [213 pixels (-104)]
--- HDR
Parameter [Dark/Bright Response] set to [Full]
Parameter [Detail Size Range] set to [1000 pixels]
Parameter [Strength] set to [1.2]
--- Deconvolution
Parameter [Primary PSF] set to [Moffat Beta=4.765 (Trujillo)]
Parameter [Tracking Propagation] set to [During Regularization (Quality)]
Parameter [Primary Radius] set to [6.5 pixels]
Parameter [Iterations] set to [20]
Mask used (BASE64 PNG encoded) (masked out the "fat" star at 3 o'clock)

iVBORw0KGgoAAAANSUhEUgAAAbcAAAE9AQAAAACgxxH4AAAAdUlEQVR4Ae3SMQ2AQBQFwZ8gBClIwxpSkHAlBeEQQAOvILlktp9uq0ddHMdxHMdxHMdxHMdxHMdxHMdxHMdxHMdx3JvOWiLXQrfXHLmN4z649LNW66/umHrkHnEcx3Ecx3Ecx3Ecx3Ecx3Ecx3Ecx3EcN6C7AdUuuL+OqbvrAAAAAElFTkSuQmCC

--- Color
Defaults.
Parameter [Matrix] set to [RGB:BGR]
--- Wavelet De-Noise (switch Tracking off)
Parameter [Grain Size] set to [7.1 pixels]
--- Wavelet De-Noise
Parameter [Smoothness] set to [73 %]

[Lawrence]

Hello Ivo

My reply is a little complex:

Firstly, I assume that you have used the bggr stacked image - as indicated in the caption. To answer initial questions: yes I used AHD in earlier stacked versions then reverted to the other one. I had previously been advised to use AHD.

I invariably use darks and bias, and mostly flats. On this occasion I did not have 100% flats, just a short run (due to earlier cloud preventing twilight flats). My view was that the problem of reversed colours would not be affected by any oversights. Dithering: I normally dither but for an unidentified reason, the SGP new version failed to activate PHD2 so I could only take pictures using guiding + imaging independently and therefore no dither. I have since fixed this problem.

'Mapping matrix on the fly ---- sorry, not seen any such option.

autodevelop - yes.
crop - yes - - - - - but you advised bin first. I did bin next.
no mention of wipe??????????? I did wipe anyway.
autodevelop - yes (all parameters set although very difficult to read due to poorly set colours (dark on darker).

HDR (note, the guide says contrast first.
settings done

deconvolution: this should have come before - as per instructions?
primary PFS - sorry, no such option
tracking etc - sorry, no such option
assume primary radius = radius - OK.
iterations - OK
mask - I masked out the bright star at 3

colours
parameter matrix - and this is where it all ended. There is no such option. LRGB options do not include RGB BGR or anything similar.

Summary

I can see the amazing capabilities of Star Tools but when options described do not appear in front of me then I know that I have done something wrong. Going back to basics, Using the original BGGR stacked image with (or without) calibration images gives me a blue outer ring where it should be red. Somehow, on one occasion I must have done the right thing and produced the valid stacked image included on the dropbox collection - but I do not know how I got it. Until I can reproduce this, I cannot proceed further with processing.

My thanks for everyone's trouble. I would still be very grateful if anyone can clarify any of the stages where I have gone wrong. I tried to find a video tutorial on YouTube but they are all using older versions and have not shown any using the new version. One shows DSLR processing but I am using a OSC.

Lawrence Harris