Stuck on M33 LRGBCombination

[Rastaban]

Hi all,

I've been reprocessing the 6 hours or so of M33 LRGB data I acquired from a dark site last November and keep running into the same roadblock. I'm pretty happy with the results of my RGB image but when it comes time to add the L it all goes sideways. I've tried adjusting the channel weight of the L but I still feel like it takes away from the LRBG image - is my L too strong? Below are some examples of what I'm working with.


RGB: DBE - BN - PCC - SNCR - AutoHistogram


L: (this is my current version, I've tried this a few times with varying degrees of stretching) DBE - MLT - AutoHistogram


LRGBCombination, no adjustments.

Finally, here's a version I previously completed (not with the above files) though there is definitely a lot more data to pull out out as can be seen above.


I appreciate any advice or pointers.

[pfile]

can you post linear masters for the RGB and L? upload to dropbox or google drive or whatever.

rob

[Rastaban]

can you post linear masters for the RGB and L? upload to dropbox or google drive or whatever.

rob

Rob,

Let me know if this works: https://drive.google.com/open?id=1q_pCs-zxR0bBdx0fuefjG81kN7UnrMRC

[pfile]

ok - for starters did you do any DBE or any gradient reduction to the L? if so then it needs to be re-done more carefully. if not, you must have some bad subs - there are dark halos around some of the stars and in parts of the galaxy.

rob

[Rastaban]

ok - for starters did you do any DBE or any gradient reduction to the L? if so then it needs to be re-done more carefully. if not, you must have some bad subs - there are dark halos around some of the stars and in parts of the galaxy.

rob

No, that's a clean integration save for a DynamicCrop. I did a quick Blink of the original L lights and made sure they were consistent, which they are, so just to make sure I didn't make any mistakes I did another ImageIntegration and got the same results. Then I thought to do one without LocalNormalization and it produced a much improved image - see the folder, its the MASTER_LIGHT LUM M33-no LNorm file.

[pfile]

ok - i should have asked... LN is very useful but (and god bless him) it's inclusion of LN in Kayron's tutorials is really causing a lot of problems for people. LN needs very careful tweaking and a clean reference and i'm not sure his tutorial really emphasizes that enough. i've seen a lot of people over the last 6 months with this type of problem.

you might re-do the RGB masters without LN if you also used LN for those...

anyway i will look at the new one in a little while.

rob

[Rastaban]

ok - i should have asked... LN is very useful but (and god bless him) it's inclusion of LN in Kayron's tutorials is really causing a lot of problems for people. LN needs very careful tweaking and a clean reference and i'm not sure his tutorial really emphasizes that enough. i've seen a lot of people over the last 6 months with this type of problem.

you might re-do the RGB masters without LN if you also used LN for those...

anyway i will look at the new one in a little while.

rob

Noted, thank you, Rob.

[ngc1535]

If I might add- there could also be something a bit more fundamental (haha) going on here.
The luminance and RGB images as posted are quite bright. Managing the brightness of the permanently stretched luminance and RGB image is a critical part of getting a successful mixing. In the case below... both the chrominance (RGB) and luminance are too near the top end and adjusting the L/RGB contribution in the LRGB combination tool will not recover a desired result.

So, if the problem is a low contrast, faintly colored image (I could be wrong and mischaracterizing the original poster)- the answer is to adjust these images so that they play with each other better. There are a number of techniques for doing so- but I will not presume I am correct yet and expound on them unnecessarily.

-adam

[pfile]

no i think you are right adam, i just was trying to start at the beginning... the linear master indicated to me that it needed to be fixed before proceeding.

rob

[pfile]

by the way the non-LN L master looks fantastic, needs some DBE but i think if you reprocessed using that image you'd have an easier time of things. regardless you do have to take care for what adam is talking about; seems like the stretch you had for the L was too much for the RGB.

rob

[Rastaban]

by the way the non-LN L master looks fantastic, needs some DBE but i think if you reprocessed using that image you'd have an easier time of things. regardless you do have to take care for what adam is talking about; seems like the stretch you had for the L was too much for the RGB.

rob

I'll give it another go with the non-LN masters and post my results. Thanks!

[Rastaban]

Okay, I think I'm getting there:

[pfile]

yes that is an improvement. i looked at the data more and i feel that there are some pretty significant gradients in the R, G and B masters which proved pretty difficult to remove. i think they limit the processing somewhat.

i did kind of a quick and dirty processing and came up with this... maybe it is worth evaluating the RGB subs and seeing if there are a number of frames with gradients and re-taking some of the RGB under better conditions.

rob

[Rastaban]

yes that is an improvement. i looked at the data more and i feel that there are some pretty significant gradients in the R, G and B masters which proved pretty difficult to remove. i think they limit the processing somewhat.

i did kind of a quick and dirty processing and came up with this... maybe it is worth evaluating the RGB subs and seeing if there are a number of frames with gradients and re-taking some of the RGB under better conditions.

rob

Wow that looks great  I took at look at my subs and I think that I may have had some thin clouds in some of them. This was at a dark site (Bortle 4) and was clear as far as I recall but it's possible there was some transparency changes during the evening. I did note in the Blink review that there were some subs that were definitely brighter than the other ones in that set. I'll give it another go with just the cleanest of subs.