GrantBoxer
Well-known member
I have heard that creating a superluminance from OSC images may enhance the detail in OSC images. Can anyone comment on this and if this is the case, suggest a work flow to create a superluminace image?
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
Superluminance for OSC images
- Thread starter GrantBoxer
- Start date
As I understand it "superluminance" is generated by integrating all your exposures (L, R, G and B) and then using the result as your luminance channel. The idea being that all these exposures would reduce SNR in the resulting "superluminance". I am not sure if OSC benefits from this approach as there is no L but you can certainly separate your R, G and B channels and then integrate them. The results can be then swapped for L using LRGBCombination.
yeah - i suppose making a pseudo-luminance lends itself well to a processing style where you enhance the L image as much as possible (pulling out details, sharpening, etc.) and merge that with and RGB image which is not necessarily processed for sharpness but is color noise reduced, etc. sort of like how people sometimes bin their color 2x2 and then use this lower-res image to add color to the L limage.
for a mono camera, juan has described in another thread (some time back) that simply integrating together all the R/G/B subs will create a pseudo-L, and the SNR weighting in ImageIntegration will properly weight the contributions from each filter's image. you can also integrate your R,G,B masters instead of the subs; same effect. for an OSC, to emulate this you'd have to do as described above on your RGB master image.
another technique is to go into RGBWorkingSpace and set your gamma and RGB weights to 1 and apply that to your RGB image, then extract L* from the image. if you don't fix the weights then you've extracted kind of a "perceptual L" image since the RGB weights in an RGB image are intended to match the color response of the human eye... and a CCD camera is different. but don't forget to undo that change on your RGB image before continuing to process it.
as far as PI is concerned, many of the tools can operate on the L* of an RGB image without having to extract it. so in theory there's lots of processes you can do to an RGB image where you are working on the "implicit" L* without having to extract it from the image.
rob
for a mono camera, juan has described in another thread (some time back) that simply integrating together all the R/G/B subs will create a pseudo-L, and the SNR weighting in ImageIntegration will properly weight the contributions from each filter's image. you can also integrate your R,G,B masters instead of the subs; same effect. for an OSC, to emulate this you'd have to do as described above on your RGB master image.
another technique is to go into RGBWorkingSpace and set your gamma and RGB weights to 1 and apply that to your RGB image, then extract L* from the image. if you don't fix the weights then you've extracted kind of a "perceptual L" image since the RGB weights in an RGB image are intended to match the color response of the human eye... and a CCD camera is different. but don't forget to undo that change on your RGB image before continuing to process it.
as far as PI is concerned, many of the tools can operate on the L* of an RGB image without having to extract it. so in theory there's lots of processes you can do to an RGB image where you are working on the "implicit" L* without having to extract it from the image.
rob
In case anyone is interested in the discussion on Synthetic Lums, here is the thread: http://pixinsight.com/forum/index.php?topic=6042.msg41076#msg41076
Note that since having that discussion with Juan, I converted to only taking quality 1x1 binned RGBs and using them to build synlums and couldn't be happier with the results, especially since it allows me to build great data sets in 3/4s the time.
Best,
Jim
Note that since having that discussion with Juan, I converted to only taking quality 1x1 binned RGBs and using them to build synlums and couldn't be happier with the results, especially since it allows me to build great data sets in 3/4s the time.
Best,
Jim