Dave's Astrophotography Blog - Making Pretty Pictures from Scientific...
Transcript of Dave's Astrophotography Blog - Making Pretty Pictures from Scientific...
Making Pretty Pictures from Scientific Data
by David Ault
Many Sources Available
• Sloan Digital Sky Survey (SDSS)http://data.sdss3.org/fields
• Hubble Legacy Archive (HLA)http://hla.stsci.edu/hlaview.html
• ESO Image Archivehttp://archive.eso.org/eso/eso_archive_main.html
• Space Telescope Science Institutehttps://archive.stsci.edu/cgi-bin/dss_form
• Local Observatories• For example, I know the professor at Southwestern University and have helped some of his
students with processing image data for posting on their website
• And many more!• Spitzer, WorldWideTelescope and probably many more I couldn’t find or remember
• Remember to give the appropriate credit!
Sloan Digital Sky Survey
• Uses Sloan filters• Wideband but not centered around
RGB• Steep cutoff with little overlap
between filters
• Anyone can access the data
• Very easy to use interface, especially for mosaic data
• Downloads all files individually
• Easy to determine exactly what portion of the spectrum is captured in each image
Sloan Digital Sky Survey
• Example Search for NGC524• 1 – shows the Search by object
name field
• 2 – the preview image
• 3 – shows the download links, there’s one for each filter
1
2
3
Sloan Digital Sky Survey
• There is also a mosaic tool• 1 – The entry form
• Provide the desired central location, image scale and field of view
• 2 – The results, with small previews
• 3 – Download button
• There are also controls for which filters and which tile to select so you can target your downloads
1
2
3
Hubble Legacy Archive
• Mix of wideband and narrowband filters and
• Anyone can access the data• Uses a cart system to accumulate
data into one big compressed archive
• Good documentation on their site• Easy to determine exactly what
each file represents (i.e. filter, imager, etc.)
• Contains some pre-processed data that has mixed colors as well as the raw calibrated data
Hubble Legacy Archive
• Results for NGC524• 1 – The ‘Help Center’ button gives
you a lot of great information about decoding everything
• 2 – A lot of information is included in the file name for ease of use when constructing your images
• 3 – Select All button
• 4 – Add Selection to Cart
• There’s a good description of what each column represents at the bottom of the page
1
1
3
4
Hubble Legacy Archive
• 1 – Cart results. It shows the file names including filter information target imaging system, etc.
• 2 – Download button. Downloads the data as one large zip file
• Data on all Hubble instruments and filters can be found here:
http://www.stsci.edu/hst/wfc3/documents/handbooks/currentIHB/c06_uvis06.html
1
2
ESO Image Archive
• Mix of wideband and narrowband filters and
• Must sign up for an account (free)
• Some data locked for scientific research & publication
• Download tools are flaky. I had to use their download script on a Linux machine.
• Good documentation on what each imaging system is capable of
ESO Image Archive
• This archive encompasses data from multiple installations, instruments and filters, making it more confusing to find the data you want
• Clicking on a column will take you to a section of the documentation describing what it means
• I’ve marked a few columns:• 1 – Selection, so you can target
specific downloads• 2 – Type showing where this is an
image, spectrum, etc• 3 – Filter
1 2 3
Southwestern University
• Uses Johnson Filters• Wideband but not centered
around RGB
• Gradual cut-off with a fair amount of overlap between filters
• 16” DFM Ritchey-Chrétien
• SBIG ST-8300 Monochrome camera
• Bortle Scale 8 skiesDr. Mark Bottorff does some public outreach
Normalizing Data
• If the histograms of your raw data are very different, then when blending the data the results will not be what you expect
• PixInsight has a tool called LinearFit that does this, but it can be done in most imaging programs
• You can see to the right what happens when you stretch raw Hubble data the same amount
Three raw frames with the same stretch applied
Dealing with alternate spectra in an RGB world• Most scientific data is not RGB• Sloan, Johnson, Washington, Strömgren and various IR and narrowband
filters are common• SHO is a common method of combining narrowband data from Sulfur-II,
Hydrogen-α and Oxygen-III where S, H & O are mapped to R, G & B respectively
• You can also approximate the spectra of typical RGB filters using overlap of the spectra of the scientific filters
• Subtractive methods can also work, so if you have L, G & R filters you can infer a B channel based on: 2*L – (G+R)
• Since all astrophotography is technically false color in terms of what the human eye is capable of seeing, there is no right or wrong way to combine the data to make RGB images
Channel Substitution (aka SHO, HSO, etc.)
R = HEIIG = NIIB = OIII
R = NIIG = HEIIB = OIII
R = (HEII + NII)/2G = OIIIB = OIII
Overlap of Spectra From Astrodon LRGB and Sloan Filters
0
10
20
30
40
50
60
70
80
90
100
300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 1050 1100
Tra
ns
mit
tan
ce
%
Wavelength (nm)
Luminance
Red
Green
Blue
u'2
g'2
r'2
i'2
z'2
0
10
20
30
40
50
60
70
80
90
100
300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 1050 1100
Tra
ns
mit
tan
ce
%
Wavelength (nm)
Luminance
Red
Green
Blue
Johnson U
Johnson B
Johnson V
Johnson R
Johnson I
Overlap of Spectra From Astrodon LRGB and Johnson Filters
Overlap of Spectra From Human Cones and Sloan Filters
0
10
20
30
40
50
60
70
80
90
100
300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 1050 1100
Tra
nsm
itta
nce
%
Wavelength (nm)
u'2
g'2
r'2
i'2
z'2
Human Blue
Human Green
Human Red
Area Overlap Between Filters• The first graph shows that the z’2 filter has no overlap with any
of the LRGB filters
• The g filter almost entirely overlaps B & G and because the cut-offs are so steep colors will appear cyan instead of blue or green
• Since we rarely see green objects in astronomical scenes we can blend it with the other colors using various methods
Area Overlap Between Filters
• The average neutral blend picks between the minimum of the green value and the average of the red and blue (taken from PixInsight SCNR algorithm)
• If the green value is higher this indicates the would have been predominately green so it is replaced, when lower it’s value is protected• g’ = min(g, (r+b)/2)
• In addition to the min protection I usually blend it back in some percentage with the original green data• g’ = α*g + (1- α)*min(g, (r+b)/2)
• This blend can be duplicated in Photoshop using three layers. In this case the lower layer would be r, the next layer b (with a 50% normal blend between them) and the upper layer would be g with a Darken blend mode (the blend factor is equivalent to α in the equation above).
Area Overlap Between Filters
• My favorite combination for Sloan filters is below
• It uses a blend between Sloan g & r overlapping with Astrodon RGB and Sloan g & r overlapping with Human perception
• It also uses an average neutral blend on the green channel
Average Neutral Blend
No blend With blend (α=0.25)
No blend
With blend(α=0.25)
No blend With blend (α=0.5)
2 Color Blends
• Here I only used the Sloan r & g filters
• I used the same average neutral blend for the green channel to make it look more natural
R = r
tg = 0.97*g + 0.03*r
G = 0.25*tg + 0.75*min(tg, (r+g)/2)
B = g
Getting Really Fancy – Hue Shifts
• In addition to different blending methods you can also use curves and levels to adjust the color
• I started with:R = (HEII + NII)/2
G = OIII
B = OIII
• Then I used curves to push the pinks towards red and greens towards blue
• This yields something a little closer to balanced RGB results
Getting Really Fancy – Different Field of View
• You will need to do 3 things to combine data from different field of view• Align the data• Match the intensities – Linear Fit• Create a mask for blending
• Because the fields of view are so different, I used the PixInsight DynamicAlignment tool
• The mask was created by stretching the Hubble data so that most of the inner nebula was white then blurring the image slightly
Hubble & Spitzer data mapped to the same field of view along with a mask
Getting Really Fancy – Different Field of View
Two different views of M57 w/ Spitzer + Hubble Data
Other Examples
M51Southwestern University
Filters: Johnson V,R
M104Southwestern University
Filters: Johnson V,R
Other Examples
M81 & M82SDSS-III DataFilters: Sloan r,g
NGC2903SDSS-III DataFilters: Sloan u,g,r,I,z
Other Examples
NGC3628SDSS-III DataFilters: Sloan u,g,r,i
NGC3628SDSS-III DataFilters: Sloan u,g,r,i,z
Other Examples
ARP188HubbleFilters: F483W,F555W, F814W
ARP188 - CropHubble
Filters: F483W,F555W, F814W
Questions?
M51HubbleFilters: F483W, F555W, F814W
M51 – Central BarHubble
Filters: F483W, F555W, F814W
Backup
Additional Links
• ESO Online Digitized Sky Surveyhttp://archive.eso.org/dss/dss
• MistiMountain Observatoryhttp://www.mistisoftware.com/astronomy/index_fits.htm
• World Wide Telescopehttp://www.worldwidetelescope.org/
• Spitzer Image Datahttp://sha.ipac.caltech.edu/applications/Spitzer/SHA/#id=SearchByPosition&RequestClass=ServerRequest&shortDesc=Position%20Search&isBookmarkAble=true
Other References
• Books• Scientific Astrophotography: How Amateurs Can Generate and Use
Professional Imaging Data by Gerald Hubbell