Please can I validate my RGB + Ha workflow?

[Phil Leigh]

Hi,

I'm having great success capturing Ha with my heavily modified DSLR (CDS600D) and Baader 7nm Ha filter.

I've been working on the Pelican Nebula (IC5070) which as you know is very Ha-rich.

This is my workflow - I've never done this before and would like to check I am doing things properly:

Starting with a large set of CR2 (RAW) files...

1) use Batch PP script as normal to create a master RGB image from the unfiltered images
2) use Batch PP script to calibrate, debayer, then align the Ha filtered images but do not integrate them
3) use the Batch Channel extraction to extract the red channel only from each aligned frame from step 2
4) Integrate the "red" images from step 3 to create a grayscale "Ha" master.
5) register the Ha and RGB masters to each other
6) use the NBRGBCombination script to blend the Ha data with the red channel of the RGB data


Is this the correct way to work?

many thanks
Phil

[Phil Leigh]

Anyone?
I'm having severe difficulty getting a really good result because the narrowband (Ha) stars are too different from the RGB versions.
I've tried different ways of doing this (all based on Vicent's process - or at least the part of it that is demonstrated in Harry Page's video and encapsulated in the NBrgb script... the Ha nebulosity looks great, but those stars... ouch!

All but the smallest stars are either white or have a red centre/edge to them.

Vicent mentions in this in "part 2" of his tutorial but due to my inexperience I can't really translate his words into a set of Pi actions :-(

[topboxman]

There is another thread asking similar questions about how to get nice stars when mixing NB (Ha in your case) images with RGB using PixInsight. So far no one has responded.

http://pixinsight.com/forum/index.php?topic=6041.0

Peter

[Phil Leigh]

Yeah - maybe Vicent or Juan or someone with the answer will stop by... :-)

[jerryyyyy]

Have you tried the script by Harry on the integration.... there is also a second one by the French group under multichannel synthesis scripts.  Harry's I have to execute but it is there under HaRGBCombinationHarry.  I guess you could do it your way, but there are a lot of controls in these scripts.  If I use Harry's I only add 10% Ha. 

BTW, after I use the script I have then done a ColorCorrection and that helps with the pink cast. 



Please realize I am a novice here with maybe 6 months experience and a limited scope. 

This one was processed without the final color correction:



I have to do another post as I am trying to fine tune my workflow and not sure when I should convert from linear to non-linear.  Right now my flow is:

Preprocessing Scripts with as many subs as I can collect
LRGB Combo with noise reduction at 3.... 4 produces red circles around stars
DBE making sure collection points are not over stars

CONVERSION by default parameters of STF

MaskedStretch... with some manual tweeking
TGVDenoise if needed
Curves to adjust color saturation

Hope this helps.... hang in there.... and anyone else please comment on the workflow.... Thanks

[topboxman]

Hi Jerry,

What was the multiplication factor when using HaRGB script? You might need to raise it to maybe about a factor of between 3 to 5 or even higher for galaxies. There is a latest HaRGB script renamed to NbRGB script to include all three NB filters for all three RGB bands. To run NbRGB using only Ha, it's identical to HaRGB.

Latest script:

http://pixinsight.com/forum/index.php?topic=3446.msg40130#msg40130

The previews for latest script appear to be better.

Peter

[jerryyyyy]

Hi Jerry,

What was the multiplication factor when using HaRGB script? You might need to raise it to maybe about a factor of between 3 to 5 or even higher for galaxies. There is a latest HaRGB script renamed to NbRGB script to include all three NB filters for all three RGB bands. To run NbRGB using only Ha, it's identical to HaRGB.

Latest script:

http://pixinsight.com/forum/index.php?topic=3446.msg40130#msg40130

The previews for latest script appear to be better.

Peter

Thanks.  I have used multipliers < 1.0 because the stars and background turn pink/magenta.  Just juices the red up a little.

I cannot find that new script?  NbRGB?  Does not appear in menus or folder with other executables.  Once I find it, I assume all is run in linear? 

PS  I also see in your workflow you do DBE on all 3 chanels separately... is it not the same to combine to an RGB image and then do the DBE on that image?  Does it not do each channel separately?  Less work....

[topboxman]

I cannot find that new script?  NbRGB?  Does not appear in menus or folder with other executables.  Once I find it, I assume all is run in linear?

Look for "NBRGBCombinationV1.5.zip" at bottom of message and click on it. Yes, it runs in linear.

Peter

[pfile]

PS  I also see in your workflow you do DBE on all 3 chanels separately... is it not the same to combine to an RGB image and then do the DBE on that image?  Does it not do each channel separately?  Less work....

if for some reason the gradients in each channel differ wildly from the others, the single set of parameters for DBE might not capture each channel's gradient as well as could be done on each channel individually...

rob

[MikeOates]

Rob,

Until this thread, I had not realised the need to treat each filtered subs separately using DBE. I can now see how easily the gradients may differ if the different filtered subs are taken at different times of the night, or on different nights, when the target could be at varying altitudes in the sky, resulting in different gradients.

Mike

[Phil Leigh]

Yes that's really useful - thanks!
Will do DBE on my RGB and Ha separately  before using the NBRGB script as they will always be separated by hours or days...

[topboxman]

Is it better to use same DBE samples for individual channels including narrow band or manually do DBE for each channel? If imaged from exactly the same location and approximately same dates, it may not make much difference.

I've always done DBE for Luminance and combined HaRGB but never for each channel. Ahhhhh, more work!!!

Peter

[pfile]

well, the proof is in the pudding. it's probably okay to try it on the RGB and see if the result makes sense. since i'm using a mono camera i just do DBE on each stack separately. if i were still using OSC then i'd probably try it both ways. with some of my setups the CA was so bad that i had to re-register the R and B to the G, so i had to split the RGB into separate channels anyway.

if your gradients are really, really bad, they can compromise the statistical analysis that ImageIntegration has to do. there's at least one thread here describing using ImageContainer and ABE to pre-background extract images before registration and integration. of course that requires extra care since you are applying ABE to a bunch of images so you have to take care to come up with good settings for ABE that work well across all the images you are trying to process in one go.

rob

[topboxman]

It looks like a personal preference. PixInsight's video tutorial showed DBE on combined RGB.

http://www.pixinsight.com/videos/NGC1808LRGB-vperis/en.html

I know it's an old tutorial and it's possible that PixInsight group may prefer DBE on separate channels now.

Isn't easier to see gradients on combined RGB image than individual grey channels? Looking at color image seems easier to differentiate between gradients and others. Maybe RGB combine first, make DBE samples, save DBE process icon and use same DBE samples for individual channels?

Peter

[jerryyyyy]

Yes, I got the habit from the old tutorial on DBE.  I have some new data on a weird new object NGC772.  The Green gradient runs in the opposite direction from the Red gradient so the composite picture is like a popsickle.  Not easy to shoot RGB near San Francisco foggy skys.  I'll run each DBE separately this time. 

Two questions though,

would you run a Linear Fit on the 3 images before the integration... on the one hand makes sense to normalize on the other hand maybe I am squashing a signal;

and LRGBCombo or ChannelCombination... which to use to combine... I have been using LRGB to benefit from the noise reduction. 

I did download the new script and will try it on this object as I have some H-alpha.