Photometric color calibration

I love the photometric color calibration (PCC) option in Siril. When it works, it produces a very pleasant images with well balanced color on stars. The problem is “when it works”. I find that is extremely sensitive to a lot of details and we receive so little information about it (for example it never works for my short focal images)

Take for example, this image of M101 from Tadej Skofic (https://www.youtube.com/watch?v=RrZ5u5TaFlM). I’m trying to process it using Siril and as it has an important colour cast, my first option was to use the PCC.

So I selected M101 coordinates as the center of the image. For the image scale I’ve uploaded the image to Astrometry.net to plate solve and get the image scale. As you can see in the solution page (http://nova.astrometry.net/user_images/3496462) the image scale es 1.77 arctic/pixel so I have put a focal length of 1000mm and a pixel size of 8.58 to get the same image scale (1.77).

But when I try to solve and apply the color correction it failed with the generic error. Looking at the logs I don’t get any clue

fetch_url(): http://cdsweb.u-strasbg.fr/cgi-bin/nph-sesame/-oI/A?M101
initializing CURL
fetch_url(): http://vizier.u-strasbg.fr/viz-bin/asu-tsv?-source=NOMAD&-out.meta=-h-u-D&-out.add=_r&-sort=_r&-out=%20RAJ2000%20DEJ2000%20Vmag%20Bmag&-out.max=200000&-c=0.000000+0.000000&-c.rm=74.6&Vmag=<12.32
log: Findstar: processing...
log: Catalog NOMAD size: 145 objects
using min_scale 1.569752  max_scale 1.969752
using trans_order 1
atFindTrans: iter_trans unable to create a valid TRANS
Give another try with scale changes
atFindTrans: iter_trans unable to create a valid TRANS
initial call to atFindTrans fails
using min_scale 1.569752  max_scale 1.969752
using trans_order 1
atFindTrans: iter_trans unable to create a valid TRANS
Give another try with scale changes
using trans_order 1.
Initial trans structure:
TRANS: a=-1.080093591e+03 b=4.654133693e-02 c=5.150562423e-01 d=-7.510087844e+02 e=7.510499296e-02 f=1.622294980e-01 sig=1.3793e-22 Nr=1 Nm=0 sx=0.0000e+00 sy=0.0000e+00
atRecalcTrans: only 1 stars in list(s), require at least 6
atRecalcTrans fails on matched pairs only
log: Plate Solving failed. The image could not be aligned with the reference stars.

Any idea on how I can improve the reliability of the option?

I’ve forget to upload the image :slight_smile:

Be careful, our tool is not dedicated to do same astrometry work as astrometry.net.
You can’t apply it on non-linear images. This is on purpose because it is non-sense as photometry can only be done on linear image.

I know that is not so general astrometric solution, I was talking about astrometric only as a reference for the image scale.

I’ll try again with a linear version of the image (this jpg is stretched to be visible here). I didn’t know that it did not work with non linear images.

Saludos,
Rafa Barberá

I’ve been able to perform a calibration on the linear image. Still the process is very sensible to image scale (too much?). Only when I tweaked the focal length (millimetre by millimetre) I’ve got a success calibration.

Oh no, you don’t have to move the focal mm by mm. If it does not work, change the value but not that small.
On my side it is ok 98%.

One more question about the PCC. I’ve been taking some images of the C/2019 Y4 (ATLAS) comet. It’s a fast moving one, so I need to stack the images aligning to the comet nucleus. The problem is that once aligned, there are no useful stars on the frame to perform the PCC. I can use the standard star alignment to obtain a decent colour balance… but a blurred comet. When we align on the nucleus, I’ve a sharp comet but a wrong colour. Are there any way to transfer the K0, K1, K2 values computed omn the WB command from one image to other?

2 Likes

You can use the former color calibration tool. It works well in many cases.

could you explain what a non-linear image is in the context of processing astrophotos ? I process my DSLR images with siril in always the same manner. Mostly it works fine, but for some images, it completely refused to work with a split second of time. I dont think this is because the image is non-linear. That doesnt make sense. It also doesnt even attempt to find stars, or states that it found only a single star when there are hundreds in the image.

Linear images are images which histogram have not been modified, the pixel values are original.

Stars are often not found when there are too many black pixels in the image, caused by misappropriate use of bias and darks, or too many white pixels in the image, causing stars to be saturated, caused by overexposure. Settings for star detection can be changed to try to improve it, there’s a shortcut button to access it from the registration tab, otherwise it’s in the PSF analysis window.

1 Like