[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[gnuastro-commits] master cd2fca2: Book: better explanation of pre-convo
From: |
Mohammad Akhlaghi |
Subject: |
[gnuastro-commits] master cd2fca2: Book: better explanation of pre-convolved stars in first tutorial |
Date: |
Thu, 23 Jul 2020 13:47:15 -0400 (EDT) |
branch: master
commit cd2fca23eaa6cf502cb89a714a81ebd8b14ff8aa
Author: Mohammad Akhlaghi <mohammad@akhlaghi.org>
Commit: Mohammad Akhlaghi <mohammad@akhlaghi.org>
Book: better explanation of pre-convolved stars in first tutorial
In the first tutorial, we build four stars (as a point source) in the model
image. But hadn't really explained the fact that they only cover a single
pixel and this caused confusion for new readers: when you open the image in
DS9 and look at the full image at once, you don't see the four stars.
With this commit, this important issue is clarified a little more to
hopefully avoid this confusion for future readers.
This issue was mentioned by Samane Raji.
---
THANKS | 1 +
doc/announce-acknowledge.txt | 1 +
doc/gnuastro.texi | 30 ++++++++++++++++++++----------
3 files changed, 22 insertions(+), 10 deletions(-)
diff --git a/THANKS b/THANKS
index 5cdbccf..69d11f6 100644
--- a/THANKS
+++ b/THANKS
@@ -69,6 +69,7 @@ support in Gnuastro. The list is ordered alphabetically (by
family name).
Marcel Popescu mpopescu@iac.es
Bob Proulx bob@proulx.com
Joseph Putko josephputko@gmail.com
+ Samane Raji samaneraji@gmail.com
Teymoor Saifollahi teymur.saif@gmail.com
Joanna Sakowska js01093@surrey.ac.uk
Elham Saremi saremi@ipm.ir
diff --git a/doc/announce-acknowledge.txt b/doc/announce-acknowledge.txt
index c290d42..5633ca5 100644
--- a/doc/announce-acknowledge.txt
+++ b/doc/announce-acknowledge.txt
@@ -5,6 +5,7 @@ Carlos Allende Prieto
Leindert Boogaard
Alexey Dokuchaev
Raúl Infante Sainz
+Samane Raji
Joanna Sakowska
Zahra Sharbaf
Ole Streicher
diff --git a/doc/gnuastro.texi b/doc/gnuastro.texi
index 10d4d3b..eb8b0a0 100644
--- a/doc/gnuastro.texi
+++ b/doc/gnuastro.texi
@@ -1628,6 +1628,7 @@ He wants the output to be 499 pixels by 499 pixels, so he
can put the center of
Looking at his drawings of it, he decides a reasonable effective radius for it
would be 40 pixels on this image pixel scale, he sets the axis ratio and
position angle to approximately correct values too and finally he sets the
total magnitude of the profile to 3.44 which he had accurately measured.
Sufi also decides to truncate both the mock profile and PSF at 5 times the
respective radius parameters.
In the end he decides to put four stars on the four corners of the image at
very low magnitudes as a visual scale.
+While he was preparing the catalog, one of his students approached him and was
also following the steps.
Using all the information above, he creates the catalog of mock profiles he
wants in a file named @file{cat.txt} (short for catalog) using his favorite
text editor and stores it in a directory named @file{simulationtest} in his
home directory.
[The @command{cat} command prints the contents of a file, short for
``concatenation''.
@@ -1679,21 +1680,28 @@ $ls
@cindex Oversample
@noindent
-The file @file{0_cat.fits} is the PSF Sufi had asked for and @file{cat.fits}
is the image containing the other 5 objects.
-The PSF is now available to him as a separate file for the convolution step.
-While he was preparing the catalog, one of his students approached him and was
also following the steps.
-When he opened the image, the student was surprised to see that all the stars
are only one pixel and not in the shape of the PSF as we see when we image the
sky at night.
-So Sufi explained to him that the stars will take the shape of the PSF after
convolution and this is how they would look if we didn't have an atmosphere or
an aperture when we took the image.
-The size of the image was also surprising for the student, instead of 499 by
499, it was 2615 by 2615 pixels (from the command below):
+The file @file{0_cat.fits} is the PSF Sufi had asked for, and @file{cat.fits}
is the image containing the main objects in the catalog.
+The size of @file{cat.fits} was surprising for the student, instead of 499 by
499 (as we had requested), it was 2615 by 2615 pixels (from the command below):
@example
$ astfits cat.fits -h1 | grep NAXIS
@end example
@noindent
-So Sufi explained why oversampling is important for parts of the image where
the flux change is significant over a pixel.
-Sufi then explained to him that after convolving we will re-sample the image
to get our originally desired size/resolution.
-To convolve the image, Sufi ran the following command:
+So Sufi explained why oversampling is important in modeling, especially for
parts of the image where the flux change is significant over a pixel.
+Recall that when you oversample the model (for example by 5 times), for every
desired pixel, you get 25 pixels (@mymath{5\times5}).
+Sufi then explained that after convolving (next step below) we will
down-sample the image to get our originally desired size/resolution.
+
+Sufi then opened @code{cat.fits} [you can use any FITS viewer, for example,
@command{ds9}].
+After seeing the image, the student complained that only the large elliptical
model for the Andromeda nebula can be seen in the center.
+He couldn't see the four stars that we had also requested in the catalog.
+So Sufi had to explain that the stars are there in the image, but the reason
that they aren't visible when looking at the whole image at once, is that they
only cover a single pixel!
+To prove it, he centered the image around the coordinates 2308 and 2308, where
one of the stars is located in the over-sampled image [you can do this in
@command{ds9} by selecting ``Pan'' in the ``Edit'' menu, then clicking around
that position].
+Sufi then zoomed in to that region and soon, the star's non-zero pixel could
be clearly seen.
+
+Sufi explained that the stars will take the shape of the PSF (cover an area of
more than one pixel) after convolution.
+If we didn't have an atmosphere and we didn't need an aperture, then stars
would only cover a single pixel with normal CCD resolutions.
+So Sufi convolved the image with this command:
@example
$ astconvolve --kernel=0_cat.fits cat.fits
@@ -1714,7 +1722,9 @@ $ls
@end example
@noindent
-When convolution finished, Sufi opened the @file{cat_convolved.fits} file and
showed the effect of convolution to his student and explained to him how a PSF
with a larger FWHM would make the points even wider.
+When convolution finished, Sufi opened @file{cat_convolved.fits} and the four
stars could be easily seen now.
+It was interesting for the student that all the flux in that single pixel is
now distributed over so many pixels (the sum of all the pixels in each
convolved star is actually equal to the value of the single pixel before
convolution).
+Sufi explained how a PSF with a larger FWHM would make the points even wider
than this (distributing their flux in a larger area).
With the convolved image ready, they were prepared to re-sample it to the
original pixel scale Sufi had planned [from the @command{$ astmkprof -P}
command above, recall that MakeProfiles had over-sampled the image by 5 times].
Sufi explained the basic concepts of warping the image to his student and ran
Warp with the following command:
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- [gnuastro-commits] master cd2fca2: Book: better explanation of pre-convolved stars in first tutorial,
Mohammad Akhlaghi <=