Hi all
Are there any “golden” rules or ways of defining how much RGB vs. L data is needed on a target?
Are there any “golden” rules or ways of defining how much RGB vs. L data is needed on a target?
Ratio between RGB and L data
Engaging
Nothing based on science or any golden rule but I do between 2x and 3x more time on L than RGB with very great results.
6h of L and 2h of RGB would be some of my typical numbers.
6h of L and 2h of RGB would be some of my typical numbers.
Concise
1:1, same scale.
Basic theory is, if you have 6 hours, take 3 hours L 50% + RGB 3 hours (1 hour each) 50%
for Colorful nebula : increase RGB %. (2 hours L + RGB total 4 hours)
for dark and distance galaxy : increase L %. (4 hours L + RGB total 2 hours)
for Colorful nebula : increase RGB %. (2 hours L + RGB total 4 hours)
for dark and distance galaxy : increase L %. (4 hours L + RGB total 2 hours)
andrea tasselli:
1:1, same scale.
@andrea tasselli - by 1:1 do you mean the same total amount of RGB as for L?
I think for these discussions it's important to really quantify with examples.
Saying 1:1 L to RGB might mean they take 3 exposures of L and 1 exposure each of RGB (LLLRGB). They might also mean they take 1 exposure of L and 1 exposure each of RGB (LRGB).
Saying 1:1 L to RGB might mean they take 3 exposures of L and 1 exposure each of RGB (LLLRGB). They might also mean they take 1 exposure of L and 1 exposure each of RGB (LRGB).
Well Written Helpful Insightful Concise
I think for these discussions it's important to really quantify with examples.
Saying 1:1 L to RGB might mean they take 3 exposures of L and 1 exposure each of RGB (LLLRGB). They might also mean they take 1 exposure of L and 1 exposure each of RGB (LRGB).
Good point @Shinpah
Christian Bennich:andrea tasselli:
1:1, same scale.
@andrea tasselli - by 1:1 do you mean the same total amount of RGB as for L?
1 hour for 1 hour of RGB (each). Then add the RGB and put that as L.
Wrote a little time exposure app that can guide you based on desired SNR. You can get a good sense by comparing exposure time per filter.
https://clearskies.go.ro
Sometimes I get power cuts and the site goes down, but there is a download link for the standalone windows version.
https://clearskies.go.ro
Sometimes I get power cuts and the site goes down, but there is a download link for the standalone windows version.
For my point of view, for:
LRGB: 3h of L for 3h of RGB
SHO with LRGB stars: same time for each LRGB max (between 30 mn and 1h) with SHO (upper than 10 h)
LRGB: 3h of L for 3h of RGB
SHO with LRGB stars: same time for each LRGB max (between 30 mn and 1h) with SHO (upper than 10 h)
I shoot for twice the number of L as for any one of R,G, or B, i.e. if I get 100 lum's, then I'm looking for 50 each of R, G, & B. But that was just a guess at what 2 to 1 meant when I saw it recommended way back when.... Also, I realize it's really more about exposure time than number of subs, and my Lums are typically 240s and RGB 300s, but since it's all a bit arbitrary anyhow, number of subs is easier to keep track of.... : )
It's certainly possible that what was meant was twice as many Lums as R, G, & B combined, so 300 lums for my 150 RGB's, which if you're getting the detail from the Lums and just painting color on top with RGB's probably makes better sense..... More exposure time means more detail, but how does exposure time correlate with color?
Alternatively, does it make any sense to skip Lum acquisition altogether, use a synthetic Lum for processing, and invest the saved time in more RGB?
Cheers,
Scott
It's certainly possible that what was meant was twice as many Lums as R, G, & B combined, so 300 lums for my 150 RGB's, which if you're getting the detail from the Lums and just painting color on top with RGB's probably makes better sense..... More exposure time means more detail, but how does exposure time correlate with color?
Alternatively, does it make any sense to skip Lum acquisition altogether, use a synthetic Lum for processing, and invest the saved time in more RGB?
Cheers,
Scott
Helpful Engaging
Scott Badger:
Alternatively, does it make any sense to skip Lum acquisition altogether, use a synthetic Lum for processing, and invest the saved time in more RGB?
Since the luminance filter captures roughly three times the photons per unit time as any one of the RGB filters, a synthetic lum constructed of RGB will have much less SNR than the lum filter in the sameimaging time. So this isn't a great idea and defeats the benefits of LRGB imaging.
I also use the 50/50 rule, i.e., 50% of time on L, 50% of time on total of RGB. This means the L image as roughly 1.7 times the SNR of the RGB. If have often felt that this is not adequate, i.e., have wished for more RGB time. One technique to improve the color SNR is to use an OSC for color capture (if imaging from a dark site) since the OSC is more efficient than mono for RGB. The LRGB images I am happiest with had their color data taken with an OSC.
Helpful Insightful Concise
I like to use about 2x the L of each color channel. So 3 hours each of RGB I'd do 6 hours of L. And then almost always create a SuperLuminance from the RGB and L masters.
How that varies depends on the target, the data quality and what processing challenges I'm having. If it is a faint object that is not super colorful I might go with more Luminance. For really bright objects where the color really defines the object I might use 1:1:1:1. For star clusters I don't use Luminance because the goal is to retain star color as best as possible and there is no challenge of capturing enough light.
I guess it works out like this:
3:1:1:1 = Faint galaxies and nebulae like M64
2:1:1:1 = General targets
1:1:1:1 = Very bright colorful objects like M20
0:1:1:1 = Star Clusters like M13
Kevin
How that varies depends on the target, the data quality and what processing challenges I'm having. If it is a faint object that is not super colorful I might go with more Luminance. For really bright objects where the color really defines the object I might use 1:1:1:1. For star clusters I don't use Luminance because the goal is to retain star color as best as possible and there is no challenge of capturing enough light.
I guess it works out like this:
3:1:1:1 = Faint galaxies and nebulae like M64
2:1:1:1 = General targets
1:1:1:1 = Very bright colorful objects like M20
0:1:1:1 = Star Clusters like M13
Kevin
Helpful Engaging
Arun H:Scott Badger:
Alternatively, does it make any sense to skip Lum acquisition altogether, use a synthetic Lum for processing, and invest the saved time in more RGB?
Since the luminance filter captures roughly three times the photons per unit time as any one of the RGB filters, a synthetic lum constructed of RGB will have much less SNR than the lum filter in the sameimaging time. So this isn't a great idea and defeats the benefits of LRGB imaging.
I also use the 50/50 rule, i.e., 50% of time on L, 50% of time on total of RGB. This means the L image as roughly 1.7 times the SNR of the RGB. If have often felt that this is not adequate, i.e., have wished for more RGB time. One technique to improve the color SNR is to use an OSC for color capture (if imaging from a dark site) since the OSC is more efficient than mono for RGB. The LRGB images I am happiest with had their color data taken with an OSC.
We've briefly had a discussion in the past r:e OSC vs mono rgb efficiency, but I'm not confident in the complete accuracy of the spreadsheet/CN thread used for that assertion (or at least in one particular case - the 571 QE curve).
To answer the poster:
For LRGB imaging I do 2:1:1:1 LRGB and integrate all the exposures together to use as the luminance data.
I think there's pretty questionable use in a 1:1:1:1 ratio if you're solely using the lum by itself since it probably won't be particularly less noisy than the luminance data from the RGB subs.