Here is my example image with this technique. The description below can also be found accompanied with the image.
I've been working on a new method for amplifying weaker O3 signals in an image that will create a more natural looking O3 while reducing the impact of non-O3 regions of the image.
In the past I have used methods where O3 was stretched more than HA/S2 prior to combination, or selectively masking the O3 areas and cranking up the saturation to make it more visible. The challenge with these methods is that it is very difficult to generate a mask that will allow you to selectively saturate areas of an image without generating mask edge artifacts where some non-O3 signal is boosted while bordering areas of the same signal are not. Color masks are also not effective at sampling areas of the image where the O3 is drowned out by Ha. So using the method below will allow you to enhance O3 signal wherever it is in the image, not just where it is strongest.
I prefer a lazier approach to processing. What I mean by this, is that simple techniques that can be easily applied across the image without too much masking or complex processing gymnastics generally produce a more natural looking result. I also like methods that can be used on separate data sets with minor tweaking of the application.
I recently developed a method to do just this for amplifying O3 regions in targets where Ha and S2 signal is overwhelming. I adopted a PixelMath BoostFactor approach that I learned a few years ago from lightvortexastronomy where Ha signal can be blended into the red channel, which is particularly useful for blending in Ha into targets like M33 or M31 where there are knot like Ha features that can only truly be captured using a narrowband filter.
Using this math I attempted to boost O3 by reapplying the O3 integration into my RGB combined narrowband image. The problem is that it turned the O3 areas purple. I could reclaim somewhat the colors I wanted but this required the processing gymnastics that I wish to avoid. The "ahah" moment came when Alex Ranous described how he blended blue into an SHO image and to avoid the purple result he added the blue into the blue channel and subtracted it from the red channel. It was kind of a face-slap moment for me and the miising piece to my technique.
So here is the process:
-Combine your narrowband channels as you like.
-Stretch your mono O3 channel to bring out the areas you wish to enhance. Clip the areas in the O3 channel that you DONT wish to enhance by moving the shadows slider to the right in Histogram tranformation. It will look something like below:
-Use this O3 channel as a MASK on the color image you wish to enhance. This will ensure that the signal you are blending in is a perfect match for the color image.
-Use this following pixel math equation. Raise or lower the boostfactor to taste. (You need to replace "StretchedO3" with whatever your O3 integration is called.
R: $T - ((StretchedO3 - Med(StretchedO3)) * BoostFactor)
G: $T
B: $T + ((StretchedO3- Med(StretchedO3)) * BoostFactor)
Symbol: BoostFactor=0.5
You can see the before and after below:
The result is a really nice enhancement of O3 in your image, while leaving the rest of the image alone.
One problem with using the O3 mask as it is, is that you will unevenly apply the O3 boost. In areas that are weaker, you will get an compounded weaker application. (The mask is fainter there as well as the signal you are boosting with). If you find that you want a more even O3 boost, I suggest that you Binarize the O3 mask. I was looking closely at my final result and realized that there is a lot of O3 overlap in the "mountainous" Ha Region. So I binarized the O3 mask before I applied the boost pixel math. The result was a more true O3 coverage. Notice the overlap area below has a lot more O3 blue inside it. I really love this transitional region. It could be a cropped image of it's own. The overall result is perhaps a little Bluer than I would prefer, however I left it as is to illustrate this process and result.
Binarized Mask:
Overlap Region with nicely blended O3:
I've been working on a new method for amplifying weaker O3 signals in an image that will create a more natural looking O3 while reducing the impact of non-O3 regions of the image.
In the past I have used methods where O3 was stretched more than HA/S2 prior to combination, or selectively masking the O3 areas and cranking up the saturation to make it more visible. The challenge with these methods is that it is very difficult to generate a mask that will allow you to selectively saturate areas of an image without generating mask edge artifacts where some non-O3 signal is boosted while bordering areas of the same signal are not. Color masks are also not effective at sampling areas of the image where the O3 is drowned out by Ha. So using the method below will allow you to enhance O3 signal wherever it is in the image, not just where it is strongest.
I prefer a lazier approach to processing. What I mean by this, is that simple techniques that can be easily applied across the image without too much masking or complex processing gymnastics generally produce a more natural looking result. I also like methods that can be used on separate data sets with minor tweaking of the application.
I recently developed a method to do just this for amplifying O3 regions in targets where Ha and S2 signal is overwhelming. I adopted a PixelMath BoostFactor approach that I learned a few years ago from lightvortexastronomy where Ha signal can be blended into the red channel, which is particularly useful for blending in Ha into targets like M33 or M31 where there are knot like Ha features that can only truly be captured using a narrowband filter.
Using this math I attempted to boost O3 by reapplying the O3 integration into my RGB combined narrowband image. The problem is that it turned the O3 areas purple. I could reclaim somewhat the colors I wanted but this required the processing gymnastics that I wish to avoid. The "ahah" moment came when Alex Ranous described how he blended blue into an SHO image and to avoid the purple result he added the blue into the blue channel and subtracted it from the red channel. It was kind of a face-slap moment for me and the miising piece to my technique.
So here is the process:
-Combine your narrowband channels as you like.
-Stretch your mono O3 channel to bring out the areas you wish to enhance. Clip the areas in the O3 channel that you DONT wish to enhance by moving the shadows slider to the right in Histogram tranformation. It will look something like below:
-Use this O3 channel as a MASK on the color image you wish to enhance. This will ensure that the signal you are blending in is a perfect match for the color image.
-Use this following pixel math equation. Raise or lower the boostfactor to taste. (You need to replace "StretchedO3" with whatever your O3 integration is called.
R: $T - ((StretchedO3 - Med(StretchedO3)) * BoostFactor)
G: $T
B: $T + ((StretchedO3- Med(StretchedO3)) * BoostFactor)
Symbol: BoostFactor=0.5
You can see the before and after below:
The result is a really nice enhancement of O3 in your image, while leaving the rest of the image alone.
One problem with using the O3 mask as it is, is that you will unevenly apply the O3 boost. In areas that are weaker, you will get an compounded weaker application. (The mask is fainter there as well as the signal you are boosting with). If you find that you want a more even O3 boost, I suggest that you Binarize the O3 mask. I was looking closely at my final result and realized that there is a lot of O3 overlap in the "mountainous" Ha Region. So I binarized the O3 mask before I applied the boost pixel math. The result was a more true O3 coverage. Notice the overlap area below has a lot more O3 blue inside it. I really love this transitional region. It could be a cropped image of it's own. The overall result is perhaps a little Bluer than I would prefer, however I left it as is to illustrate this process and result.
Binarized Mask:
Overlap Region with nicely blended O3:
