First try of RGB processing

[mstriebeck]

Hi,

I am imaging from near downtown San Jose (CA). So far, mostly narrowband imaging because of the light pollution. But now I tried RGB imaging and wanted to try out Pixinsight's gradient removal and such techniques.

I found a great tutorial in this thread: http://pixinsight.com/forum/index.php?topic=4385.0. I followed the steps in there to the ColorSaturation step and ended up with this image: http://www.astrobin.com/91481/0/. While I'm pretty happy with the details and background, I am not able to get any color out (I have seen color in images from others).

Also, I took some Ha data too. What is the best point to integrate the Ha data with the RGB data? Right at the beginning? Or do you process the RGB image and the Ha image separately to a certain point and then integrate?

Thanks  for any help!

     MarkS

[Alejandro Tombolini]

Hi Mark, It seem that is needed more saturation. Maybe you should had increased a little more Saturation or if you performed a chrominance noise reduction, it could have been too agressive.
Which were your processing steps?

Saludos, Alejandro

[mstriebeck]

Hi Alejandro,

Thanks for the reply. I tried the whole workflow now 3 times and always with (pretty much) the same results. I posted my steps and screenshots here: http://mstriebeck-astrophotography.blogspot.com/2014/04/galaxy-imaging-from-our-backyard-using.html

There are 3 steps where I'm not sure if I'm doing something wrong (in the blog post, I marked these with red comments):

1. DBE: Although, I am getting the gradients very much down, there are still some gradients left. But I couldn't figure out how to get rid of them.

2. BackgroundNeutralization: I don't see a big difference before and after - the backround stays in its red/green patterns

3. ColorSaturation: Although, I'm applying a quite extreme saturation, no color comes out

Any ideas or pointer is appreciated!

     MarkS

[RickS]

Mark,

Just a few points after looking at your blog post:

• Probably no point in using both ABE and DBE.  I'd just hit the image with DBE, maybe more than once
• It's often better to use a smaller number of samples with DBE rather than covering the image in spots
• Background neutralization will give you an average neutral background by scaling the overall R, G and B.  It won't remove chrominance noise
• Try boosting saturation with the "S" curve in CurvesTransformation.  Just pull the curve upwards.  You may need to apply a luminance mask to limit this effect to the brighter areas of the image.

Cheers,
Rick.

[Alejandro Tombolini]

Hi Mark,

1.
I think that it is very important to eleiminate any gradient before continue as NGC 3628 and M66 have fine nebulosity that could be highlighted.
Do not take the process in the previous example as a recipe as it may be only situable for that image.
As Rick said I would start with DBE placing the samples conveniently to remove the gradiente. I made a try over the stretched .jpg and seems to work. After that CanonBandingReduction. See attached processes.
It is a good idea to use different DBE configuration for each channel and you may obtain a better result in this image.

There aren't any gradients left, but still a lot of background noise.

DBE will not remove the noise, then it is ok to have noise at this step.

2.

I am not sure if I'm doing something wrong here. There is virtually no difference before and after BackgroundNeutralization...

Applying DynamicBackgroundExtration successfully will let you have a quite netural background, then BackgroundNeutralization you will not have to do a big job, that why you do not see so much difference.

3.
Deconvolution. It seems that the generated PSF include only three stars, remember that you must highlight all the stars before exporting the PSF.

4.

You can see how much I increase saturation, but there isn't much color coming out :-(

You can increase Saturation further if you change the "Range" parameter. Set it in a high value and increase Saturation a little more. This will generate a big chrominance noise, but it is easy to remove.

Saludos,
Alejandro.

[pfile]

after BackgroundNeutralization and ColorCalibration, make sure your STF is set to linked channels and re-compute it. otherwise STF is "re-doing" the color balance to what it thinks the color balance should be, on top of the correct color balance.

of course with well balanced data the STF should not be terribly wrong with the channels unlinked, but the proper thing to do is lock the channels after CC.

rob

[pfile]

oh with regards sampling in DBE, if your gradients are smooth and well-behaved, then fewer points are definitely better.

but in situations where there is a lot of light pollution, and you track a target all night, it's easy to end up with U-shaped LP gradients, and worse. in this case you have to place enough samples to be able to capture this complex shape. also with LP you may need to set the tolerance quite high to capture the brighter parts of the background. this will cause DBE to pick up on halos of stars in some samples so you need to take care to try to move all of your samples off of any bright features that are not background. otherwise then the computed background will have false bright areas.

bottom line: check all your automatically generated samples. if you don't want to step thru them one by one you can tell DBE to use a smoothness of 0, then just compute the background, stretch it with STF and see if it looks too lumpy. you can then hone in on those sample that are probably 'bad' and need to be moved.

rob

[mstriebeck]

Thanks everybody - yes, I am tinkering a lot with the sampling points for DBE.

Would it actually be easier to remove as much as possible gradients from the individual color images before combining them? Or is that a wash?

[pfile]

i have done that. however, DBE (in it's current form - juan is rewriting it so this may change) can not be used with ImageContainer. so you have to use ABE for this purpose if you have a lot of subs and you want to do the background subtraction automatically using ImageContainer.

for this image: http://www.astrobin.com/68802/

i employed that technique, but i ended up following up with DBE on the integrated results. however, the DBE was a lot easier to accomplish with the pre-processed subs.

rob