10.26
#... |
|---|
|
Hi Folks, First post here - I'm interested in your various techniques for processing NB images. Personally - I primarily use j-P Metsavainio's Tone Mapping method - https://dl.dropboxusercontent.com/u/24702393/Tone%20Mapping_v2.pdf Also - what methods do you use for star removal and replacement of RGB stars? Cheers Andy PS: I've attempted to post an image here of M16 & SH2-54 which I'm having trouble getting enough info from theO3 data http://www.astrobin.com/255908/B/ |
0.00
#... |
|---|
|
Hi Andy, Welcome, To combine my HA, SII, OIII images I use Pixelmath and the imagecombination tool in Pixinsight. First I combine the raw images with the Imagecombination tool. R= SII G=HA B=OIII. Then I remove the green for the most part to bring out the blue color of the OIII. Then I use a formula with Pixelmath to combine the HA, OIII, SII images. This will result in a more RGB like background and star color. Finally I combine the two images with Photoshop. |
8.04
#... |
|---|
|
I use Photoshop throughout with Registar to register the filters (inc binned data). I never had any success with star removal, so tend to just shrink them to match. I never get much colour in RGB stars, so never replaced them on to NB images. This is going to be my next challenge - to get better colour in stars. Carole |
0.00
#... |
|---|
| I've never had good luck with star removal. I'm still working on various techniques in PI but all leave residues of the removal process. I've had far better luck with star minimization .. |
7.93
#... |
|---|
I've never had good luck with star removal. I'm still working on various techniques in PI but all leave residues of the removal process. I've had far better luck with star minimization .. I am in the same boat. I haven't quite figured out star removal in PI, so I resort to deconvolution and star reduction for now. I also usually do all my combinations on linear data, rather than on stretched data, to avoid star bloat and star halo issues. That ends up requiring more meticulous work with the Hue curve to really bring out the color the way I want it, but at least I don't really have star halo issues. |
|
0.00
#... |
|---|
| I only shoot narrowband - so the first thing I consider is what to do with my stars. The issue is star color: Do I stretch hard, and blow out the stars to white ... or mask them and try to preserve star color. It's pretty subjective for me and depends on my mood. Lately, I've been considering star color more though. My last two images I've done partial stretches and then combined, and finished the stretch with the color version. |
0.90
#... |
|---|
Bert Scheuneman: Hi Bert, how do you do that? Do you use the histogram process to reduce its intensity? And: Bert Scheuneman: The key here would be, what formula? I guess it's a linear combination of the three but with different coefficients? cheers, Richard |
|
0.00
#... |
|---|
|
@ Richard To remove the green I use the scnr tool in Pixinsight. If you use Photoshop click the link below for a great green (gradient) removal tool. http://www.deepskycolors.com/archivo/2010/04/26/hasta-La-Vista-green.html I use the formula's explained on the site below (see link) https://pixinsight.com/forum/index.php?topic=3734.0 Bert |
|
0.90
#... |
|---|
|
Thanks for that -- I've been using PixInsight for years but mostly using the same tools all the time. I'd never come across scnr. Great! |
|
0.00
#... |
|---|
|
Hi Andy, I remove the stars from the masters I use for the colour combine. I do it in PI but I don't fuss too much if it's not perfect. A little clone stamping of the colour combined data generally cleans it up enough. I use LinearFit to match the NB intensities to get a starting point for the colour combine (with LRGBCombination.) The Luminance is derived from combining the original NB masters (with stars.) Apart from some extra colour tweaking, typically using the ColorMask script and CurvesTransformation, the rest of the process is much the same as what I do for RGB data. Cheers, Rick. |
2.10
#...
·
2
likes
|
|---|
|
Here is some information you may find helpful in working with you NB images. My Tone Mapping Hubble Palette Image Workflow To begin with, I generally follow the technique described by J-P Metsavainio (http://astroanarchy.blogspot.com/2009/11/power-of-tone-mapping.html). Read this over before you start again. I use both PixInsight and Photoshop. Here is the step-by-step process I use for Hubble Palette Tone Mapped images, as best I can describe it. 1. I start with the three narrowband images,H-alpha, Oxygen lll and Sulfur ll. The Ha is usually captured at bin 1. The O3 and S2 can be captured at bin 1 or there is not much signal, bin 2.) These are the calibrated, registered and stacked linear images, usually generated in PixInsight (note: if you bin O3 and S2 you will have to upscale the finalstacked images to match the bin 1 Ha image. 2. Smoothing: I do the initial processing inPixInsight. Using the linear image, I smooth each image using the TGVDeNoiseProcess. There is a YouTube video by David Ault on The Astro Imaging Channel that describes it and various other forms of smoothing. Here is the link ( https://www.youtube.com/watch?v=em7AXdvNHRA ). It’s worth looking at for all types of images. Once you set up the parameters for this and all the related Processes in PixInsight, be sure to save them as aProject so you won’t have to re-input the variables each time. 3. Initial Stretching: This is done usingeither the default STF stretch or the MaskedStretch Process in PixInsight. You can experiment with each one to see which gives you the best result. After the initial stretch, I usually do some fine adjustment using the HistogramTranformation Process. I might also do some enhancement of the nebulosity using the PixelMath Process. I don’t fully understand how to write expressions using PixelMath, but this formula seems to work. Here is the formula, 1-(1-$T)*(1-$T). Before implementing this expression on the non-linear image, you do a simple mask of the image. 4. You should now have a reasonable non-linear image for each of the NB images. You can save them in FITS format to preserve your individual steps and as a TIFF file to work on them in Photoshop. 5. Initial work in Photoshop: At this pointI open each in PS and do the following. A.)First duplicate the background Layer, then open the Levels on that layer and move the left slider to remove any flat spots in the histogram. If it’s flat there is no data there. I always create a duplicate layer for any major change in case something goes wrong and I have to start again. B.) If there are any obvious gradients I use the Filter RC-Astro GradientXTerminator tool. This is a Russell Croman tool you can purchase online. C.) If needed, I smooth the image using the ImageNomic Noiseware Filter. This plug-in is also available online. You can also take the image back into PixInsight and reduce the noise using the TGVDeNoise process mentioned earlier. Now you should have a multi-layered PS document where the top layer is the best version of your NB image. 6. Star Removal: Make a copy of that top layer and label it Starless. Do this for each of the three NB images. Now you are going to remove stars from each Starless layer in each image. Removing the stars is simple but time consuming. I recommend you start with Annie’s Actions. This is a suite of Actions that you will find useful. Play the Remove Stars (Large Image) from Annie’s Actions on the Starless layer. That will get rid of most of the stars, but will leave lots of star residues. This is the time-consuming part. Take your Spot Healing Brush from your PS Tools menu. Make the size of the brush the size of the largest star residues. Click on each to remove them. Now you have medium star residues. Make you brush a little smaller and remove those star residues. Repeat this process until the star residues are gone, or to the point where your patience runs out. Do this for all three images. The quality of your starless image effects the quality of colored Tone Map. Be patient. 7. The Tone Map: J-P has a certain way to make the tone map. You can follow that process, or mine here. Both end up with a similar result. First take each of the NB Starless layers and copy them. Create a new RGB document of the same size. Open the Channels tab in that document and paste the Starless Ha layer into the Green channel, the Olll into the Blue channel and the Sll into the red channel. This is your initial starless tone map. It’s green. Don’t worry, we are going to fix that. Adjusting the Tone Map: At this point I generally follow the process developed by Bob Franke. Here is the link to that process (http://bf-astro.com/hubblep.htm ). This is relatively straight forward, so there is no reason to repeat it here, EXCEPT FOR ONE MAJOR ITEM. I do the color adjustment on the tone map, not the final image with stars. If you do it on the star image, you will end up with colored halos around your stars. When you are done with the color adjustment you now have a tone map. Look closely at the tone map. If you see any obviously colored star residues, take the Spot Healing Brush and remove them. One more item. The final adjustments to the color of the tone map I do by feel. I end up doing different tweaks to the color depending on how things look to me. There is no right or wrong, you just have to use your artistic sense here. After all these color adjustments you may find that there are some hard-edged blotchy areas. That comes from stretching that color a bit too much. I generally select those areas, give the selection a soft edge by feathering and then doing a Gaussian blur just enough to get rid of the hard color edge. Don’t worry about losing detail. All that comes from your luminosity layer. 8. The Luminosity Layer: In almost all cases, the luminosity layer is composed mostly of you Ha image. J-P describes his way of constructing that layer. I have a few variations on that. Here is what I do. Take you Ha image/layer (after all your adjustments) and copy it to a new document called Luminosity. Copy the starless image for O3 and paste it on top of the Ha layer. In the Layers tab, set the combine to Screen. Set the Opacity to between 15 and 25%, it’s up to you. Do the same for the S2 Starless image. Now you have three layers. Flatten the image and adjust the histogram in the Levels menu so there is no flat spot on the left of the histogram. Move the left slider as needed. This is your Luminosity layer. Make whatever adjustment you deem appropriate to make it appealing. I have no specific instruction here. It just must appeal to you eye. 9. The Hubble Palette Image. You now have a colored Tone Map and a Luminosity image. Copy the Luminosity image and paste it on top of the Tone Map. Set the top layer to Luminosity. This is you Hubble Palette image. At this point you might want to make certain adjustments to the Tone Map layer and or the Luminosity layer. This is strictly up to you. There is no right or wrong. Just make it fit your eye. 10. One more item. Where are the colored stars in this Hubble Palette tone mapped image? There are none. The stars are white. First, star color has no meaning in NB Hubble Palette images. So white stars are fine. Second, IMHO, RGB stars do not match the color palette of a Hubble Palette image. Third, stars from the Ha image are usually tight. RGB stars are usually much more bloated and detract from the final image. Well that’s just my opinion. And as Dennis Miller says after a rant, “I could be wrong.” Good luck. Eric |
|
10.26
#... |
|---|
|
Hi Eric, Thanks so much for taking the time to post this valuable in depth workflow. I'll be sure to reread this carefully to see where my NB processing can be improved. Cheers & clear skies  Andy |
|
0.00
#... |
|---|
|
Thanks, Eric! I used a quick and rough version of your technique to reprocess my NGC1499 image and I think it came out quite a bit better. I used MLT instead of TGVDN (got to figure that process out) and I don't have the ImageNomic Noiseware PS Filter so there is probably some noise that could be cleaned up. My star removal could have been a bit cleaner as well but I definitely see the benefit of using it along with the tone/color mapping. http://astrob.in/274666/B/ Thanks again, David |
|
2.10
#... |
|---|
|
David, Not bad at all. The California Nebula is a good target to do the Hubble Palette image. There is decent emissions in all three of the NB channels. Be sure to post the Ha image separably. There is a lot of detail there. And keep up the good work. I have done 1499 a few times myself. Here are three images showing the development of the Hubble Palette image. http://www.astrobin.com/223805/B/?image_list_page=3&nc=&nce= Hydrogen-alpha http://www.astrobin.com/224479/?image_list_page=3&nc=&nce= The Tone Map http://www.astrobin.com/224052/B/?image_list_page=3&nc=&nce= The Hubble Palette image Eric |
0.00
#... |
|---|
|
Hey great detailed explanation Eric, i'll give a few of your own processes a try soon. And i agree with your latter stages, where inspiration and mood can influence the feel of an image in the end. You can often get in a Zen kind of state as you apply certain tweaks, going one direction or the other. I'm finding Straton very useful in removing stars, trying certain levels first before deciding on the final choice. What do others think of this tool? Lately im very impressed with Adobe lightroom for certain tweaks at final stages. I'll often do certain adjustments in that and then add them as a layer in the PS master, feathering them in as a layer as needed. |
9.03
#... |
|---|
|
Hello, Let me join the discussion. I have a lot of trouble with the star removal. I have tried Annies. Straton and by PI erosion etc. and it will work for me with a lot of effort for the HA images. I remove residue with the spot healing brush in PS. I did the Pelican this way and it came out, BUT, there were not many stars in the Pelican. If you shoot around the edge of the Milky Way there are just tons of stars and tons of artifact. O-III for me is really hopeless then. Maybe there are some magic settings for Straton, but I cannot find them. I communicated with the developer and he has not magic bullet. I have sorta given up and just try to get the stars about the same size across the narrow band images... that is not easy either. I use the debloating techniques with MLT. BTW, I agree the noise removal the earlier the better is important. I have come to completely rely on MureDenoise as soon as the images are stacked. I do not have to do much later. If you have noise in the original images and stretch that.... Nice thread... all the references are here and thanks Eric. JY |
