The images linked in the previous post by my friend André were taken with the 2m Faulkes South Telescope, Siding Spring.
We were trying to do some photometry on the SN2016adj, which is located inside the Centaurus A Gx.
It appears that, for some reason, in the last of the 3 images, the centroid can't be set to the target.
The SN is the small star very near a bright star, some distance to the right of the center of the galaxy (it looks like a bright double star).
When trying to put an aperture to this star, the centroid stick to the nearest bright star. The only way to go through this is to disable centroid detection, which is not totaly satisfying.
More, the same centroid trouble is to be found in some images taken with a 1m tel.
Here is a link to one of the faulty images :
It seems the centroid detection is false when the targeted star flux is small vs that of the near bright star.
By the way, the astrometrical plate solving, which I tried for testing purpose failed for all of the one-meter-telescope images. AIJ seems to be not able to find enough stars for submission success, no matter the setting entered in the « Astrometry settings » screen ; plus, many hot pixels were detected as stars (this can be controled by adjusting the noise tolerance well above 1 : 5 to 10 seems better). Fortunately those images where already plate solved at the observatory, so it don't really matters here. Anyway do you have any advice to deal with this, should this happening for other non-solved images, as to detect more efficiency the field stars before submission ?
When I click on the links sent previously by André, I see a French language website with a file name and size in the upper right corner. Unfortunately, my French is extremely poor, and the names seem to be only text and do not have a link. There are several
links on the page that all appear to be advertisements, and I am concerned about clicking on a link that might somehow install a virus on my machine.
I'm not sure I absolutely need the images to provide suggestions. Centroid algorithms are iterative in nature and tend to grab the brightest source nearby after a few iterations. The best way to do time series differential photometry in those cases is to place
a dummy aperture T1 on a bright star which will not loose the aperture. Then place T2 on the star you are interested in measuring with centroid turned off (each aperture can independently be centroided or not). To make the second aperture a target aperture,
press shift when you place the aperture. Zoom in very close on the star and carefully center the aperture by hand in the first image. Then place any comparison stars, with or without centroid enabled. When you start the processing of images, the image shift
will be determine by the centroided apertures and the non-centroided apertures will be shifted by the same amount. For this to work, the first (dummy in this case) aperture must have centroid enabled.
If you are doing single aperture photometry on a single image and fell like you need centroid enabled, you might try the alternate "Howell" centroid method that can be selected in the aperture settings panel (click
above an image, the select
). It may or may not provide better results. The standard method works best for defocused stars and the Howell method works best for focused stars.
To help with the hot-pixels when doing astrometry, in the astrometry settings panel enable the median filter and increase the radius beyond the default value of 2. This should remove most hot pixel detections. You can also limit the maximum peak value allowed
to be considered a source. This may help if the hot pixels are all the way to the max ADU of the detector. I wouldn't raise the noise tolerance for an image with few stars (you might eliminate some faint stars) to address hot pixels. In fact you might try
lowering it to 0.1. It's better to use the other settings mentioned above to cause hot pixels to be ignored. If there are very few stars, you may need to set the approximate RA and Dec of the center of the field, assuming that is known.
Thank you so much for helping !
With the advices you provided , I was able to suppress all hot pixels from the centroids submission to astrometry.net, and to increase subtantially the number of stars, so the submission was successful and the most recalcitrant plate was solved. That's great !