Rob Hawley's Pages Almaden Observatory Main Page

M8 and M20 (Lagoon and Trifid Nebulas)

July - September 2013

Reprocessed October 2015

This is a Narrowband for Stars (nb4stars) project.  For more on the theory of what I am trying please consult the nb4stars page.
Two of the most popular star party objects of the summer are star forming regions in Sagittarius Messier 8 and Messier 20.  From a dark site these are some of the few Messier objects visible to the naked eye. Both are complex in their own ways.

M8 is the most southern of the two objects.  The open cluster it is forming (NGC 6530) pokes out from the cloud.  A brighter portion of M8 lies west of the open cluster.  When observing visually there is a distinct dark nebula that separates the two.  Curiously this dark band showed only with extensive processing.

M20 lies a bit to the north.  It is also a star forming region, but is split by a distinct dark band (Barnard 85) giving the impression of 3 separate bright areas.  Hence its common name the Trifid.  M20 also contains a reflection nebula.  This would normally not be captured using my extreme narrowband emission filters.  The Strömgren filters easily caught it.

The filters I used did not capture Oxygen III. That is a major component of both nebula. Thus the images lack the aqua color that should normally be present.  In the future I shoot additional O III which I will add to the green channel.  I enhanced the image taken by the three "RGB" filters by adding additional Hydrogen data. That produced a pink color closer to the real color of hydrogen. I can't shoot this as a pure Narrowband project without losing the reflection nebula of M20.

Getting both a good view of the Reflection nebula and good detail in emission nebulas proved challenging.  I describe the process below.

At the time these were taken my FSQ had a rather severe collimation error.  Thus the corners have poor focus.  The scope got an all expense paid trip to Japan in Winter 2014 to correct this.

During AIC 2015 I learned of a new feature in PixInsight (ColorMask) that allows selective modification of areas of specified colors.  M8 is Red.  I mean really Red. One attendee suggested that one could get a more satisfactory image by selecting a range of red colors and then adding some green and blue to moderate the color.  I made that change and I think it brings out more details.

Full Frame

M8 and M20 using Narrowband for Stars (Reprocessed)

Click on the image above for a full size image.  The annotated image is also presented below.


Processing images is always a tradeoff.  This is just the M8 area reprocessed to moderate the reds (and bring out more detail).

M8 using Narrowband for Stars


M20 presented a different problem.  The DSE and LHE make only minor differences to the image.  The challenge was to preserve the dim reflection nebula.  The nebula only is visible in the sV and sYel filters.  The Ha20 and 3nm Hydrogen filters showed the area with the same general hydrogen return as the rest of the Milky Way in the area.  Adding Hydrogen to the G and B channels kept swamping the reflection nebula. 

Also I found I had to very carefully stretch the image.  Both tasks were accomplished by using a mask that only exposed (hid) the mid range intensities.  When I added the Hydrogen data I made sure that only the brighter regions of the nebula were affected.  Similarly I did a Curve stretch that only affected the reflection nebula.  This preserved both in the final image.

I am showing the original release copy and not the Oct 15 reprocess since I like the original better.

M20 with nb4stars enhanced with 3nm Hydrogen

IC 4685

This is a bright area on the east side of the image.  This portion of the image is greatly affected by my current optical problem, but this shows this is an interesting area.

IC 4685 with nb4stars enhanced with 3nm H

Composing the Images

This was probably the most complex processing task I have encountered so far.  The point of shooting simulated RGB was to simulate color at least in the stars.  I believe my system will correctly reproduce star colors, but since it only collects H alpha and N II it will not reproduce the color of nebulas. Since my N II filter is not currently mounted I cannot calibrate the Nitrogen contribution, Thus the NII data is contained in the Ha20 data, but I don't know how much is there. 

I talked about some of the processing above, but now let me document it in more detail.  All processing was done using Pixinsight 1.8.

Data Collection

Number of Images
sV (Strömgren V)
sYel (Strömgren Y)
20nm Hydrogen (Ha20)
3nm Hydrogen

All filters were Astrodon

While still linear I applied deconvolution and TGVDenoise to sharpen the stars and reduce the noise. By reducing noise at this point there was less to take out in the final steps.

G2V Color Calibration

To insure that stars are the correct colors one images a reference G2V (i.e. sun like) star.  You then balance the colors so this star appears white.  For this project I used HIP 99046 which is east, but the same elevation.  From this I determined that the correction should be


These were used for color correction in PixInsight.  For comparison here is an version of the image with the BV Index of the stars annotated.  The reader can decide for himself if the colors are accurate.

Adding Additional Hydrogen

 The vast amount of Hydrogen captured with the Ha20 left the resulting image swimming in red and not very interesting.  To add some interest and make clear how much of the image was H (as opposed to N II) I wanted to produce something like a true Hydrogen color.  A true Hydrogen color is pinkish

Hydrogen Discharge tube

After some calculations a friend determined that the proper mix to simulate this color is 100% red, 12% Green, and 27% Blue.  So far so good, but I did not want adding H to affect my color calibration.  To prevent this from happening I took my H data and aggressively applied Morphological Transformation to it.  This had the effect of dimming the stars.  I then created a mask that only exposed the bright nebula. 

After color correction the H in the bright nebulas was reduced relative to the 3nm Ha data.  I added the H back in by applying a Max(Corrected_Red, Ha).  Thus if the remaining signal was at least as strong as my 3nm signal it would remain.  I suspect the 3nm mostly replaced the Red data in the nebulas. This made it easy to know how much H to add to the G and B. I just used PixMath to add the MT'd H data with the appropriate weighting.

The mask insured that only the bright nebula was affected. This preserved the reflection nebula from being overwhelmed and insured the star colors were not corrupted.

In retrospect I probably should have used the Vicent Peris technique to add the Ha.  I will try this when I return to this project next spring.


So far the image had been linear.  At this point I could directly apply a HT and brighten it.  However doing so cause the stars to bloat.  To prevent this I first did a partial stretch with the Masked Stretch script.  I then applied HT.

Bringing Out the Structure

The image at this point is rather flat looking.  Most of the detail is hidden in small differences in intensity.   I used several of the sophisticated tools to increase these subtle differences.  The first is HDRMultiScaleTranform which spreads the brightness over a wider range of values.  Then as I discussed above I used tools that detected and highlighted the edge differences.

Final Steps

Finally I used MMT to further sharpen the structures the previous steps disclosed.  Then another pass of TGVDenoise to remove the last noise.


Overview of M8/20

Annotated Image of M8/M20 area

Copyrights for Photos

Creative Commons License
Except as noted, all work on this site by Robert J. Hawley is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 United States License. This permits the non commercial use of the material on this site, either in whole or in part, in other works provided that I am credited for the work.