I’d appreciate some information or advice on the advantages or disadvantages of using L + Synthetic L , RGB over using just L, RGB when using Compose to load your Broadband data sets
I’m aware how Synthetic L is created from RGB channels but what happens if your RGB channels are significantly different in regards to overall exposure time ( but same Gain , same temp and same individual subframe time )
Thanks in advance
Martin
Comparison L, RGB and L + Synthetic L , RGB ( Compose )
-
Mike in Rancho
- Posts: 1166
- Joined: Sun Jun 20, 2021 10:05 pm
- Location: Alta Loma, CA
Re: Comparison L, RGB and L + Synthetic L , RGB ( Compose )
Hi Martin,
In most cases, unless through experience I know better what is coming, I will always try L+Synth L first. Occasionally that might not produce a better and cleaner combined L, maybe noisy, or maybe the reintroduction of chrominance in the Color module doesn't look right, in which cases I will then try out L, RGB, which often takes care of that.
But it's really dataset dependent. And of course everything about compose module hinges on SNR. If good enough, why not utilize some SNR found in your RGB filters to augment the L? Unless there ends up being good reason not to. My recent vdB 152 was one such case, being a super faint target with poor SNR out of my bright skies, and so my (still bad!) processing went better with L, RGB than L+Synth L.
You can fine tune your combined L (meaning L+Synth L) using the exposure sliders, using math and/or looking at your sensor QE graphs, or by eye after you've evaluated the relative SNR of everything. Keep in mind that more exposure time in the slider increases the relative weighting (I had that backwards in my head early on) of that channel in the creation of the combined L.
The old school ST default just had everything at 3600 if I remember right. That would result in relative weighting of 50% L and 50% RGB. Broken down by filters that would be 1/2 L and 1/6 each RGB. In 1.9, the exact total exposures of each filter are automatically filled in from the FITS header, if there.
But even so, consider it a starting point. You may find that the true exposure times do not appropriately weight for the actual SNR of a particular filter. Blue might be a reasonable culprit here. But I tend to do broadband LRGB as ST intends. I have, however, used this technique in weighting the RGB Synth L in narrowband (so really the SHO), usually by increasing the exposure (weighting) of the Ha versus something weak or noisy like Oiii that has poor SNR. The chrominance of course is unaffected by such weighting and comes in during Color as recorded, to be dealt with using the bias sliders and matrices.
Keep in mind, as you probably already know, that even if we have (or ST reads in) the correct exposure times, the numbers really don't matter. What matters is that those numbers, whatever they are, set up relative weighting.
I would avoid, if possible, having your RGB exposures be that different. Not only for Synth L, if used, but for clean color when the chrominance layer is overlaid and composited on top of the processed luminance. They don't have to be perfect, sometimes subs have to be thrown out, but I would think with mono you should try to reasonably be in the ballpark of 3:1:1:1 LRGB.
In most cases, unless through experience I know better what is coming, I will always try L+Synth L first. Occasionally that might not produce a better and cleaner combined L, maybe noisy, or maybe the reintroduction of chrominance in the Color module doesn't look right, in which cases I will then try out L, RGB, which often takes care of that.
But it's really dataset dependent. And of course everything about compose module hinges on SNR. If good enough, why not utilize some SNR found in your RGB filters to augment the L? Unless there ends up being good reason not to. My recent vdB 152 was one such case, being a super faint target with poor SNR out of my bright skies, and so my (still bad!) processing went better with L, RGB than L+Synth L.
You can fine tune your combined L (meaning L+Synth L) using the exposure sliders, using math and/or looking at your sensor QE graphs, or by eye after you've evaluated the relative SNR of everything. Keep in mind that more exposure time in the slider increases the relative weighting (I had that backwards in my head early on) of that channel in the creation of the combined L.
The old school ST default just had everything at 3600 if I remember right. That would result in relative weighting of 50% L and 50% RGB. Broken down by filters that would be 1/2 L and 1/6 each RGB. In 1.9, the exact total exposures of each filter are automatically filled in from the FITS header, if there.
But even so, consider it a starting point. You may find that the true exposure times do not appropriately weight for the actual SNR of a particular filter. Blue might be a reasonable culprit here. But I tend to do broadband LRGB as ST intends. I have, however, used this technique in weighting the RGB Synth L in narrowband (so really the SHO), usually by increasing the exposure (weighting) of the Ha versus something weak or noisy like Oiii that has poor SNR. The chrominance of course is unaffected by such weighting and comes in during Color as recorded, to be dealt with using the bias sliders and matrices.
Keep in mind, as you probably already know, that even if we have (or ST reads in) the correct exposure times, the numbers really don't matter. What matters is that those numbers, whatever they are, set up relative weighting.
I would avoid, if possible, having your RGB exposures be that different. Not only for Synth L, if used, but for clean color when the chrominance layer is overlaid and composited on top of the processed luminance. They don't have to be perfect, sometimes subs have to be thrown out, but I would think with mono you should try to reasonably be in the ballpark of 3:1:1:1 LRGB.
Re: Comparison L, RGB and L + Synthetic L , RGB ( Compose )
Hi Mike,
Thanks for the detailed overview
I suppose like most things in this hobby , try both alternatives and as always look for improvements in SNR
As suggested I’ll aim for 3:1:1:1 ratio , give or take a few subs here and there.
It’s early days yet but the 3 or 4 targets I’ve imaged using my Mono set up , I’ve have noticed a definite reduction in noise floor and cleaner signal compared to the 2600MC OSC on the same targets.The level of SNR is of course dependent on so many variables but so far this 2600MM is really encouraging.
Thanks again
Martin
Thanks for the detailed overview
I suppose like most things in this hobby , try both alternatives and as always look for improvements in SNR
As suggested I’ll aim for 3:1:1:1 ratio , give or take a few subs here and there.
It’s early days yet but the 3 or 4 targets I’ve imaged using my Mono set up , I’ve have noticed a definite reduction in noise floor and cleaner signal compared to the 2600MC OSC on the same targets.The level of SNR is of course dependent on so many variables but so far this 2600MM is really encouraging.
Thanks again
Martin