Lum data in HaRGB?

[jtheios]

I've been having fun using Silvercup's new HaRGB Combination script, as well as studying Vincent's tutorial and previous posts (and Harry's video) on the subject. So far so good. But my question is what to do with the Lum data I've gathered? My options seem to be: 1) discard those images (seems like a waste), 2) create a non-linear LRGB image and combine that with the Ha, or 3) or combine the linear HaRGB image with my Lum data. Preliminary testing doesn't seem to favor options #2 and #3. My best results seem to be just sticking to R, G, B and Ha frames. Going forward, I can change my acquisition strategy for those object I plan to merge Ha data into-- but what should I do about the Lum data I've already gathered?

Any thoughts would be greatly appreciated.

-- John

[Terry Danks]

Hi John:

I have no advice to offer, just a couple of questions.

I am pretty new to PI . . . and, for that matter, pretty new to Ha imaging.

What is this Silvercup script you refer to and where do I find it?

I have amassed 9 hours integration time on IC434, seven of that is straight Ha. I have just stacked ALL the frames, RGB and Ha into DSS. It seems to turn out well but perhaps there is a better way. I have tried splitting the channels as follows.

Split the Ha into RGB and discard the G and the B.
Split the RGB. Save the G and the B.
Combine the R channel from the RGB data with the R channel from the Ha data and then . . .
Recombine to get an Ha+R, G and B image.

That seemed inconsistent . . . sometimes working and sometimes giving me a rather strange cast. In no case (I've only tried a few and could be making any number of mistakes) was the result in any way BETTER than just stacking all the light frames, RGB and Ha in DSS.

Terry Danks

[jtheios]

Terry,

The script can be found here:

http://pixinsight.com/forum/index.php?topic=3446.0

and is based on Vincent Peris's work here:

http://www.pixinsight.com/tutorials/narrowband/theory/en.html

Harry Page also has a video explaining how to implement the approach with PixelMath-- which Silvercup's script implements in a more automated fashion.

I think this will really help you.

John

[jeffweiss9]

     I have the same question as John above. I've always collected HaLRGB with the RGB at 2x2 binning to take advantage of that fundamental perception trick that the eye gets it's resolution information almost entirely from the L, not the color.   Vicent's method, on other hand, seems directed at HaRGB only where all subs, including RGB, are collected at 1x1 binning.  That obviously works very well and I'm tempted to change my Ha collection habits but it seems to me that is a more inefficient method of collecting the data by basically getting the luminosity from the 1x1 binned RGB.
     The only thing I still do outsided of PI (except print and make jpgs) is HaLRGB combination. For that I use Tony Hallas' HaLRGB combination method which is really (HaL)(HaR)GB.  It uses the Ha both as luminosity (primarily for pulling out the faint structure not seen in Lum) and strengthening the red channel.   Unfortunately the technique uses layers and requires PS but has the virtue that you do your best LRGB image and then put the layers together with the Ha in a final mesh that is greatly adjustable until you get it the way you like.  So I also wonder if anyone is working on a script or tutorial, or just can offer advice, for doing HaLRGB combination in something like this style within PixInsight?
Thanks,
-Jeff

[jtheios]

Anyone have thoughts on this?

[pfile]

i have not read vicent's method nor silvercup's script, so i cant comment on the strengths of that method.

i guess the most traditional thing to do is merge the Ha into the red channel in some proportion (i guess according to taste) and then do an LRGB combine.

regarding photoshop vs. pixinsight... having never used photoshop for astro work, what i've done is simply use pixelmath to mix the R and Ha, and then use the LRGB process to merge it all together. when i have done this with "real" telescopes i've done bin 1x1 for the L and bin 2x2 for everything else.

[Terry Danks]

Terry,

The script can be found here:

http://pixinsight.com/forum/index.php?topic=3446.0

I think this will really help you.

John

Thanks very much, John.

I tried it out today and it is marvelous! Far less work for me than "rolling my own" in PI.
A big thanks to Silvercup for his work!

[Terry Danks]

Since using this script, I've encountered some difficulties. They are illustrated by a montage (warning: 750KB!) at http://danks.netfirms.com/PIHaRGB.jpg

1/ Note that all the HaRGB images contain very unnatural, totally red stars that are cosmetically unacceptable. The pure Ha   data alone does not produce these unsightly stars.
2/ None of the HaRGB combos are as pleasing as the Ha data alone or the RGB data alone.
    These are both displayed at http://danks.netfirms.com/ngc2244.htm
3/ It seems the Ha multiplication factor is quite insensitive? The Ha=0.5 seems as heavily weighted to Ha as is the Ha=5.0?

Am I doing something wrong to get these results?

BTW, I have tried to be consistent in processing these. There has been no noise reduction applied at any stage . . . no attempt to incorporate darks or flats . . . that's for another post.

[Silvercup]

Hi:

It seems that you don't have aligned the images previously.

Silvercup

[Terry Danks]

Hi Silvercup.

Routine was:

1/ Align all Ha images with each other. Then integrate them to produce a single
"master" Ha image.
2/ The same as #1 for all the RGB images.
3/ Align the two resulting images with each other, i.e., the "master" Ha image with the "master" RGB image.
4/ Run the HaRGB script.

So everything was aligned.

And many thanks for your assistance.

Terry

[zvrastil]

Hi Terry,

I agree with Silvercup that what you have in your examples is clearly result of misaligned images. In your combined images, every bright star has its red Ha counterpart shifted to the bottom-left, just look at the patterns. I'd suggest to re-check image alignment (for example by placing one image window on top of the other and switching between them with Ctrl+PgDn) and try to re-align them with StarAlignment. Maybe, you just referenced wrong image from the script?

cheers, Zbynek

[Terry Danks]

Hi Zbynek:

I spent a good part of the day trying to see where I am going wrong.

I realigned (Star Alignment Tool) and re-stacked (Integrate Tool) the Ha subs and then the RGB subs.
So I had two images, an integrated RGB image and an integrated Ha image.

I then ran the Star Alignment Tool on just these two images.

Finally I ran the HaRGB script, being careful to use the "-r" registered Ha and RGB files.

Same result.

In desperation I took two jpg images produced from DSS, one for Ha and one for RGB.
I aligned them in PI using the Star alignment tool.
Then Ran the HaRGB script on the two "-r" registered files. . .

Success!

But I don't yet know what is going wrong in the alignment or the integration steps in PI when I try to do the whole process in PI.

Terry