Catalogs and indexes
606 views
Skip to first unread message
Eric SIBERT
Jan 10, 2017, 4:28:05 PM1/10/17
to astrometry
Hi all (and mainly Dustin ;-) ),
In a recent comment and as indicated on the website, you suggested me to
use 4100 indexes. And finally, I completely removed 4200 indexes and get
faster solving with only 4100 indexes.
As far as I understand:
- 4200 indexes are built using 2MASS i.e. with magnitude measured around
2 µm.
- 4100 indexes are built using Tycho-2. Tycho-2 has two magnitudes Bt
and Vt. In the readme file, one can see a "mag -B" flag. The hipparcos
Bt band as a maximum around 425 nm (525 nm for Vt). Is it true Bt or
calculated B?
Amateur astronomers are using CCD or CMOS cameras with a sensitivity
ranging from 400 nm to 1000 nm with maximum between 500 nm and 700 nm.
For those like me that are using DSLR, green pixels used for astrometry
have a maximum between 500 nm and 525 nm. So magnitudes measured at 2 µm
may not be suitable. Due to intrinsic stars properties and interstellar
absorption, magnitudes may be different and relative brightness between
stars may be affected. The latter point may change the way search is
made in indexes. I'm right?
Online version also uses USNO-B catalog. Right?
Which band did you used to build indexes?
(Optical B, R, and I bands (0.4-0.9 microns))
For amateur astronomers, it would be more suitable to use catalogs with
bands around 500 nm. Correct?
For instance, Tycho-2 Vt? better than Bt?
Tycho-2 only account for bright stars.
For fainter one, UCAC-4 seems interesting due to the use of a 579-642 nm
bandpass. It may not be optimal for DSLR but nice for CCD and CMOS
sensitive to H-alpha (656 nm). Do you agree?
Some other catalogs may be useful?
Should I recommend to amateur astronomers that are installing locally
astrometry.net to use only 4100 indexes when available and use 4200
indexes when 4100 indexes are not available?
Can we configure astrometry.net to use 4100 indexes before 4200 ones?
I saw that one can build its own indexes :
http://astrometry.net/doc/build-index.html
I'm slightly afraid. I would like to build indexes using Tycho-2 Vt.
Regards
Eric
In a recent comment and as indicated on the website, you suggested me to
use 4100 indexes. And finally, I completely removed 4200 indexes and get
faster solving with only 4100 indexes.
As far as I understand:
- 4200 indexes are built using 2MASS i.e. with magnitude measured around
2 µm.
- 4100 indexes are built using Tycho-2. Tycho-2 has two magnitudes Bt
and Vt. In the readme file, one can see a "mag -B" flag. The hipparcos
Bt band as a maximum around 425 nm (525 nm for Vt). Is it true Bt or
calculated B?
Amateur astronomers are using CCD or CMOS cameras with a sensitivity
ranging from 400 nm to 1000 nm with maximum between 500 nm and 700 nm.
For those like me that are using DSLR, green pixels used for astrometry
have a maximum between 500 nm and 525 nm. So magnitudes measured at 2 µm
may not be suitable. Due to intrinsic stars properties and interstellar
absorption, magnitudes may be different and relative brightness between
stars may be affected. The latter point may change the way search is
made in indexes. I'm right?
Online version also uses USNO-B catalog. Right?
Which band did you used to build indexes?
(Optical B, R, and I bands (0.4-0.9 microns))
For amateur astronomers, it would be more suitable to use catalogs with
bands around 500 nm. Correct?
For instance, Tycho-2 Vt? better than Bt?
Tycho-2 only account for bright stars.
For fainter one, UCAC-4 seems interesting due to the use of a 579-642 nm
bandpass. It may not be optimal for DSLR but nice for CCD and CMOS
sensitive to H-alpha (656 nm). Do you agree?
Some other catalogs may be useful?
Should I recommend to amateur astronomers that are installing locally
astrometry.net to use only 4100 indexes when available and use 4200
indexes when 4100 indexes are not available?
Can we configure astrometry.net to use 4100 indexes before 4200 ones?
I saw that one can build its own indexes :
http://astrometry.net/doc/build-index.html
I'm slightly afraid. I would like to build indexes using Tycho-2 Vt.
Regards
Eric
Dustin Lang
Jan 11, 2017, 10:12:11 AM1/11/17
to astrometry
Hi,
If you set the scale ranges correctly (--scale-low, --scale-high, or -L, -H), adding more index files shouldn't slow it down *too* much.
It's also possible to created different "astrometry.cfg" files that only tell the solver about some index files; you can then use them with solve-field --config X.cfg
Yes, it should work best to have a roughly matching bandpass between the index and images; the brightness ordering is used to choose which stars to put in the index, and also determines the order that stars from the image are checked.
The 2MASS indexes performed about as well as USNO-B with SDSS r-band images in my tests, if I recall correctly. And they have simpler licensing terms.
Pan-STARRS-1 would be interesting to look at too, and then GAIA once it becomes more complete and uniform in later data releases.
I don't recall how the Tycho-2 indexes were built, but if you found mention of "-mag B" then that's probably right. That would be just the B_t column, nothing fancy. You're probably right that one built with V_t would work better.
The 2MASS-based ones use the J-band brightnesses, ~1.2 microns.
The web version, yes, uses the USNO-B indexes, which were built using the R band.
There is code in the astrometry.net repository for reading the UCAC-4 catalog, and I know people have built index files from them before. Don't know if anyone has published anything.
If you try to build some index files, please do report any successes or failure, especially if you find any errors or things that are unclear in the documentation!
cheers,
--dustin
If you set the scale ranges correctly (--scale-low, --scale-high, or -L, -H), adding more index files shouldn't slow it down *too* much.
It's also possible to created different "astrometry.cfg" files that only tell the solver about some index files; you can then use them with solve-field --config X.cfg
Yes, it should work best to have a roughly matching bandpass between the index and images; the brightness ordering is used to choose which stars to put in the index, and also determines the order that stars from the image are checked.
The 2MASS indexes performed about as well as USNO-B with SDSS r-band images in my tests, if I recall correctly. And they have simpler licensing terms.
Pan-STARRS-1 would be interesting to look at too, and then GAIA once it becomes more complete and uniform in later data releases.
I don't recall how the Tycho-2 indexes were built, but if you found mention of "-mag B" then that's probably right. That would be just the B_t column, nothing fancy. You're probably right that one built with V_t would work better.
The 2MASS-based ones use the J-band brightnesses, ~1.2 microns.
The web version, yes, uses the USNO-B indexes, which were built using the R band.
There is code in the astrometry.net repository for reading the UCAC-4 catalog, and I know people have built index files from them before. Don't know if anyone has published anything.
If you try to build some index files, please do report any successes or failure, especially if you find any errors or things that are unclear in the documentation!
cheers,
--dustin
Eric Sibert
Jan 11, 2017, 10:41:50 AM1/11/17
to astro...@googlegroups.com
Dustin Lang <dstn...@gmail.com> a ?crit :
And yes, my underlying question is how to speed up plate solving that
I'm using on field with a tablet. Indeed, my main problem on field is
when it doesn't solve at all. I added the --depth 10,20,30,40 option
to avoid a too depth search that is usually without success. The use
of 4100 indexes seems also a good way to have successful searches.
> It's also possible to created different "astrometry.cfg" files that only
> tell the solver about some index files; you can then use them with
> solve-field --config X.cfg
Ok.
> Pan-STARRS-1 would be interesting to look at too, and then GAIA once it
> becomes more complete and uniform in later data releases.
By the way, I discovered that GAIA DR1 and TGAS can be download as (a
bunch of) FITS files. So it should be not too difficult to make
indexes with it?
GAIA is using its own G-band that seems to correspond to its CCD sensors.
> I don't recall how the Tycho-2 indexes were built, but if you found mention
> of "-mag B" then that's probably right. That would be just the B_t column,
> nothing fancy. You're probably right that one built with V_t would work
> better.
> The 2MASS-based ones use the J-band brightnesses, ~1.2 microns.
Ok, no as far as I imagined.
> The web version, yes, uses the USNO-B indexes, which were built using the R
> band.
So USNO-B is used in the visible light range.
> If you try to build some index files, please do report any successes or
> failure, especially if you find any errors or things that are unclear in
> the documentation!
Yes, I may try building indexes with Tycho-2 Vt band or with TGAS. I
will let you know.
Thank you very much for your answers
Eric
> If you set the scale ranges correctly (--scale-low, --scale-high, or -L,
> -H), adding more index files shouldn't slow it down *too* much.
Yes, I usually set scale ranges.
> -H), adding more index files shouldn't slow it down *too* much.
And yes, my underlying question is how to speed up plate solving that
I'm using on field with a tablet. Indeed, my main problem on field is
when it doesn't solve at all. I added the --depth 10,20,30,40 option
to avoid a too depth search that is usually without success. The use
of 4100 indexes seems also a good way to have successful searches.
> It's also possible to created different "astrometry.cfg" files that only
> tell the solver about some index files; you can then use them with
> solve-field --config X.cfg
> Pan-STARRS-1 would be interesting to look at too, and then GAIA once it
> becomes more complete and uniform in later data releases.
bunch of) FITS files. So it should be not too difficult to make
indexes with it?
GAIA is using its own G-band that seems to correspond to its CCD sensors.
> I don't recall how the Tycho-2 indexes were built, but if you found mention
> of "-mag B" then that's probably right. That would be just the B_t column,
> nothing fancy. You're probably right that one built with V_t would work
> better.
> The 2MASS-based ones use the J-band brightnesses, ~1.2 microns.
> The web version, yes, uses the USNO-B indexes, which were built using the R
> band.
> If you try to build some index files, please do report any successes or
> failure, especially if you find any errors or things that are unclear in
> the documentation!
will let you know.
Thank you very much for your answers
Eric
Dustin Lang
Jan 11, 2017, 11:14:38 AM1/11/17
to astrometry
Gaia TGAS is basically the same as Tycho-2 (not as many stars), and Tycho-2 has more than enough precision for nearly any job.
Gaia DR1, with >1 billion sources, has a weird selection function. You can get a sense of what it contains, here:
http://legacysurvey.org/viewer/?gaia-dr1
Later Gaia data releases will be more uniform and complete. For now, I think Pan-STARRS is more interesting. But certainly needs a strong cut on Ndetections to be reliable.
cheers,
--dustin
Gaia DR1, with >1 billion sources, has a weird selection function. You can get a sense of what it contains, here:
http://legacysurvey.org/viewer/?gaia-dr1
Later Gaia data releases will be more uniform and complete. For now, I think Pan-STARRS is more interesting. But certainly needs a strong cut on Ndetections to be reliable.
cheers,
--dustin
Eric Sibert
Jan 12, 2017, 6:13:24 AM1/12/17
to astro...@googlegroups.com
> Gaia TGAS is basically the same as Tycho-2 (not as many stars), and Tycho-2
> has more than enough precision for nearly any job.
I may be interested in TGAS if it contains Gaia G-band magnitude.
> has more than enough precision for nearly any job.
One possible drawback is the epoch. Gaia DR1 is J2015.0 and I suppose
the same for TGAS. What will append to astrometry.net if we build an
index with a catalog not at J2000.0?
> Gaia DR1, with >1 billion sources, has a weird selection function. You can
> get a sense of what it contains, here:
> http://legacysurvey.org/viewer/?gaia-dr1
But for now and even in middle-term, I think Tycho-2/TGAS will provide
enough stars for my work.
Regards
Eric
Eric Sibert
Jan 12, 2017, 10:33:40 AM1/12/17
to astro...@googlegroups.com
> Gaia TGAS is basically the same as Tycho-2 (not as many stars), and Tycho-2
> has more than enough precision for nearly any job.
And indeed, completeness of TGAS is an issue:
> has more than enough precision for nearly any job.
http://galaxymap.org/drupal/node/232
" DR1 release notes:
Many bright stars at G≲7 are missing from Gaia DR1;
Sources close to bright objects are sometimes missing;
High proper motion stars (μ>3.5 arcsec yr-1) are missing;
Extremely blue and red sources are missing;
The net effect is that no naked eye stars and few stars close to the
Sun are included in TGAS"
Typically stars that I need to perform wide field plate solving.
So, I will stick with Tycho-2.
4100 indexes readme mentions a tycho2-cut.fits file. Is this file
available somewhere for download ?
Eric
Dustin Lang
Jan 12, 2017, 10:50:20 AM1/12/17
to astrometry
Eric SIBERT
Jan 12, 2017, 4:30:40 PM1/12/17
to astro...@googlegroups.com
Le 12/01/2017 à 16:50, Dustin Lang a écrit :
> You can find tycho2-cut.fits on http://data.astrometry.net/
So, I tried to build indexes under Ubuntu (in VirtualBox). I installed
> You can find tycho2-cut.fits on http://data.astrometry.net/
astrometry.net from repo (version 0.67).
I downloaded tycho2-cut.fits.
The table header of tycho2-cut.fits looks like:
XTENSION= 'BINTABLE' / FITS Binary Table Extension
BITPIX = 8 / 8-bits character format
NAXIS = 2 / Tables are 2-D char. array
NAXIS1 = 32 / Bytes in row
NAXIS2 = 2557501 / no comment
PCOUNT = 0 / Parameter count always 0
GCOUNT = 1 / Group count always 1
TFIELDS = 6 / No. of col in table
TFORM1 = '1D ' / Format of field
TTYPE1 = 'RA ' / Field label
TUNIT1 = 'deg ' / Physical unit of field
TFORM2 = '1D ' / Format of field
TTYPE2 = 'DEC ' / Field label
TUNIT2 = 'deg ' / Physical unit of field
TFORM3 = '1E ' / Format of field
TTYPE3 = 'MAG_BT ' / Field label
TUNIT3 = 'mag ' / Physical unit of field
TFORM4 = '1E ' / Format of field
TTYPE4 = 'MAG_VT ' / Field label
TUNIT4 = 'mag ' / Physical unit of field
TFORM5 = '1E ' / Format of field
TTYPE5 = 'MAG_HP ' / Field label
TUNIT5 = 'mag ' / Physical unit of field
ORIGIN = 'ESO-QFITS' / Written by QFITS
DATE = '2008-02-25T15:15:58' / [UTC] Date of writing
TTYPE6 = 'MAG ' / label for field
TFORM6 = '1E ' / format of field
END
Both Bt and Vt magnitudes are available.
Indeed, in the 4100 indexes built parameter, there was no mention of a
specific magnitude (this was a misunderstanding from me: "-S mag -B 0.1").
So I tried with "-S MAG_VT" and built successfully new indexes.
I used new indexes with my own windows software through cygwin install
of astrometry.net. It worked :-)
On one of my DSLR picture (20°x15°) I did a comparison between the
different indexes (xx08 to xx19 for all).
- 4200 : first solve 17 matches / verify 134 matches
- 4100 : first solve 36 matches / verify 151 matches
- Tycho/Vt : first solve 40 matches / verify 158 matches
So, there seems to be an improvement.
I'm happy ;-)
Eric
Dustin Lang
Jan 12, 2017, 5:18:31 PM1/12/17
to astrometry
Good work!
Eric SIBERT
Jan 13, 2017, 3:05:48 PM1/13/17
to astro...@googlegroups.com
For those who are interested, they can download my Tycho-2 Vt magnitude
indexes at:
http://dl.free.fr/st6FqJZaT
(a single zip file with indexes from 4119 down to 4107).
Regards
Eric
indexes at:
http://dl.free.fr/st6FqJZaT
(a single zip file with indexes from 4119 down to 4107).
Regards
Eric
Reply all
Reply to author
Forward
0 new messages