Affiliation
Astronomical Society of South Australia (ASSAU)
Sat, 11/08/2014 - 13:39
Hi all
I've tried a number of times to upload multiple images (different filters) of the same object for processing by VPhot but only one (the last one uploaded) ends up in my image list.
In this case, it's DSLR g, b, r channels taken from a single image.
Images appear to have uploaded OK, as shown above, but the last one wins:
Notice that I'm selecting All from the Filter drop-down (and yep, I clicked Refresh).
I'm wondering whether the problem relates to the FITS header or something else.
Can anyone suggest what I may be overlooking?
I know why Cal is red, so don't worry about this. I can do photometry on the image.
David (BDJB)
File Upload
U Aql processed images.png46.07 KB
U Aql V image list_0.png90.2 KB
David:
What filter designations appear in the fits headers for your dslr images? You noted that the images are from the b g r color pixels BUT the image that made it is called V. What is the filter designation for the other two images?
I don't think b g and r are valid filters in VPhot.
Ken
Hi Ken
FITS FILTER keywords are not generated by the tool I'm using (IRIS; there are other ways of adding these of course) so currently I'm specifying the filter via the Quick Upload page's drop down menu.
The FITS files in question are each the result of extracting a single color channel (red, green, blue) from a calibrated DSLR image. I am using V, B, and R to denote these.
Plate solving works just fine for each image. It's just that only one appears in the image list, which is puzzling.
I am able to perform differential photometry on such an image with VPhot (e.g. the green or V) to produce an untransformed magnitude. I'm well aware of the need to apply a transformation to go from the DSLR g filter to Johnson V, but at the moment I'm just trying to get more than one image processed and in the list.
My choice of filter name may not be strictly valid, but should this affect what ends up in the image list?
Thanks.
David
>>The FITS files in question are each the result of extracting a single color channel (red, green, blue) from a calibrated DSLR image. I am using V, B, and R to denote these.<<
This implies that you uploaded one image at a time and selected the filter letter from the drop down list before each upload. You did not select a single letter and upload all images (red, green ,blue) at once. Is this correct?
>>Plate solving works just fine for each image. It's just that only one appears in the image list, which is puzzling.<<
This implies that you plate solved each image offline before the upload. Is this correct? The plate solve would not be affected by the filter name. Otherwise, how would you know that the plate solve was succesful if you did not see the image in your VPhot image list?
Ken
Hi Ken
Regarding your first question:
Yes, that's correct.
Regarding your second question:
VPhot (i.e. PinPoint) plate solved the single image that ended up in the image list. I didn't plate solve them offline before upload.
You can, of course, still have an uploaded image in your list that has a WCS or Cal error.
Thanks.
David
This is getting "curiouser"!
Can you add the FITS filter keyword to one missing image header and upload the image again to see what happens?
Also, if you email the image to Geir, he may have an idea.
Ken
Shall do Ken. I have a couple of deadlines in the next day or so, and will return to this soon.
Thanks.
David
Hi again Ken, all
I tried again to upload all 3 images and again their processing status was shown as being OK but only one image (V) appears in my image list.
Note that Cal is OK now because I have made sure this is in the header based upon processing carried out. Also, one other difference this time is that I used the Upload Wizard vs Quick Upload. The end result is the same problem as last time though.
Each FITS header appears below, starting with the V image. The is the one that makes it into the image list.
The keywords in bold are what I have modified vs what comes straight from IRIS.
SIMPLE = T
BITPIX = 16
NAXIS = 2
NAXIS1 = 4290
NAXIS2 = 2856
MIPS-HI = 1221
MIPS-LO = 21
MIPS-X1 = 0
MIPS-Y1 = 0
MIPS-X2 = 0
MIPS-Y2 = 0
MIPS-BIX= 1
MIPS-BIY= 1
MIPS-CCD= 0
MIPS-FL = 0
MIPS-SX = 4290
MIPS-SY = 2856
MIPS-NOR= 1.000000
MIPS-CST= 0.000000
DATE-OBS= '2014-08-13T11:35:54.000'
EXPTIME = 20.7490
JD = 2456882.983383964
MIDPOINT= '2014-08-13T11:36:04.375'
FILTER = 'V '
CALSTAT = 'BDF '
RA = '19 28 32.802'
DEC = '-06 54 28.852'
OBJECT = 'U Aql '
END
SIMPLE = T
BITPIX = 16
NAXIS = 2
NAXIS1 = 4290
NAXIS2 = 2856
MIPS-HI = 1165
MIPS-LO = -35
MIPS-X1 = 0
MIPS-Y1 = 0
MIPS-X2 = 0
MIPS-Y2 = 0
MIPS-BIX= 1
MIPS-BIY= 1
MIPS-CCD= 0
MIPS-FL = 0
MIPS-SX = 4290
MIPS-SY = 2856
MIPS-NOR= 1.000000
MIPS-CST= 0.000000
DATE-OBS= '2014-08-13T11:35:54.000'
EXPTIME = 20.7490
JD = 2456882.983383964
MIDPOINT= '2014-08-13T11:36:04.375'
FILTER = 'B '
CALSTAT = 'BDF '
RA = '19 28 32.802'
DEC = '-06 54 28.852'
OBJECT = 'U Aql '
END
IMPLE = T
BITPIX = 16
NAXIS = 2
NAXIS1 = 4290
NAXIS2 = 2856
MIPS-HI = 1102
MIPS-LO = -98
MIPS-X1 = 0
MIPS-Y1 = 0
MIPS-X2 = 0
MIPS-Y2 = 0
MIPS-BIX= 1
MIPS-BIY= 1
MIPS-CCD= 0
MIPS-FL = 0
MIPS-SX = 4290
MIPS-SY = 2856
MIPS-NOR= 1.000000
MIPS-CST= 0.000000
DATE-OBS= '2014-08-13T11:35:54.000'
EXPTIME = 20.7490
JD = 2456882.983383964
MIDPOINT= '2014-08-13T11:36:04.375'
FILTER = 'R '
CALSTAT = 'BDF '
RA = '19 28 32.802'
DEC = '-06 54 28.852'
OBJECT = 'U Aql '
END
Note that the times in each are identical because these were taken as RAW on a DSLR and from there to RGB and red, green, blue channels to create separate FITS. Does VPhot perhaps map unique times as keys to images since CCD images are taken through different physical filters at slightly different times? I may be completely wrong about that but it's easy enough to test that hypothesis. I'll try that later but your input is welcome in the meantime.
David
Tried an experiment. Renamed one V image for one target that I uploaded this AM. Copied the image and changed filter name to B in the copy. Uploaded these two images (renamed original and the copy). Obviously they both have identical times since they are copies but have different filters.
Both renamed images appear in the list of uploaded images with OK status. However, only the B image showed in my image list in my VPhot account. In fact the original V filter image used for copy was missing??
Something is strange with identical times??
Ken
Hi folks,
There is a check in the processing to prevent duplicate images. If someone accidentally uploads the same image(s) multiple times, the same image(s) will show multiple times in the list.
So the check looks in the database to see if there is an image from the same telescope taken at the same time, and if so it is ignored.
Geir
Hi Geir, Ken
Thanks. This is what I suspected. So, we're saying that the current notion of equality between images is defined in terms of time but no additional structural aspect of the FITS header such as filter. Correct?
DSLR photometry is an example in which multiple filters are taken at the same time, something to consider for the future I suppose.
I've just tried updloading the 3 images again but with slightly different times:
Is this likely to suffice to distinguish the images? I didn't change other time fields but I think this is the mandatory one.
David
Hi Ken, Geir
The changed times above still resulted in a single image (V) in my list.
I'll try changing the seconds and upload again.
David
This time I had a PinPoint error.
First time for that one.
I'll try again later.
David
Hi Geir, Ken, all
By using Quick Upload and making sure the times differed by one second each, the 3 images uploaded correctly:
The implication is that currently, VPhot requires image times to differ by at least one second to be considered to correspond to (or in data structure terms: to map to or be a unique key for) different images. Something to consider changing for the future as DSLR photometry (and simultaneous measurement of an image through multiple filters) becomes more common?
Note that if I used the Upload Wizard instead, the images uploaded but each was shown as having a V filter type, probably based upon the first test image uploaded. Is that a bug or a feature?
On that subject, is there an issue tracker for VPhot?
Thanks for persisting with me on this guys.
Now at least the possibility of transformed photometry via VPhot for DSLR images becomes possible, assuming one has the appropriate transformation coefficients.
David