Condensation defect, any way to process out?

cmarcus

Well-known member
I have a condensation patch on my subs and was wondering if there is a way to process it out or minimize it? As you can see it "changes position" pro and post meridian flip (tho' most subs are post flip where the spot is nearer to the business area of the image, arguably I could have cropped otherwise...).

[This has occurred despite setting the camera to cool over 15 minutes. I understand a further procedure is to cool slowly, let it get back up to ambient temperature and then cool again, but I'm a bit late for these subs...)]

TIA for any input
 

Attachments

  • defect2.JPG
    defect2.JPG
    379.2 KB · Views: 91
  • defect1.JPG
    defect1.JPG
    451.2 KB · Views: 95
Hi! Perform two integrations, pre- and post-meridian flip and take look at this post. You'll have to repeat the procedure described there twice, altering the roles of the "clean" and "target" images. Then combine the "corrected" images with ImageIntegration.
 
For me, this doesn't look like a condensation issue at all (I had one with my ASI071MC Pro and it looked quite differently.).

Did you capture flats and apply a calibrated MasterFlat in image calibration?

Bernd
 
Perform a selective rejection by putting zeros over the artifact on one set of the images. This will substitute the real values from the the set where the rejection does not occur during ImageIntegration. Again, observed values with only one run through of ImageIntegration. Furthermore, since the artifact is at the same position- you can apply the zeros to this area by using an ImageContainer so that you do not need to operate on all of the images.

-adam
 
Perform a selective rejection by putting zeros over the artifact on one set of the images. This will substitute the real values from the the set where the rejection does not occur during ImageIntegration. Again, observed values with only one run through of ImageIntegration. Furthermore, since the artifact is at the same position- you can apply the zeros to this area by using an ImageContainer so that you do not need to operate on all of the images.

-adam
Thanks, what do you mean by "putting zeros" over the artificat?
 
For me, this doesn't look like a condensation issue at all (I had one with my ASI071MC Pro and it looked quite differently.).

Did you capture flats and apply a calibrated MasterFlat in image calibration?

Bernd
I applied flats, but only my stock ones (I do them every few months not every night). In the past this defect has been cured by slow cool down, the local people will apply the dessicant again (this is a remote set up).
 
PixInsight uses 0 as Black and 1 as White with everything else in-between. So Adam is saying write "Black" over the artifact and then perform the selected rejection ....
 
Thank you very much, but this is still over my head, how am I "writing Black"? what environment or process are we in at this point?
 
you can use CloneStamp to paint black (0) over the faulty images and then with ImageIntegration you can do the pixel Rejection , to delete the black pixels and replace them with the pixels from the other images. This only works if the artifact is NOT on all images.
 
Last edited:
I applied flats, but only my stock ones
I've never come across the idea of "stock flats" (stock bias and darks, yes). I guess with a permanent observatory, with the camera permanently attached to the scope and the filter(s) permanently in place it might be possible. For I guy like me who has to set the gear up for every observing session, I have no option but to take flats every night.
 
Again: in my view this doesn't look like a condensation issue, but rather like a new dirt particle. You could easily judge it when you take a close look at different light frames: condensation is not static, it is an equilibrium process.

Bernd
 
you can use CloneStamp to paint black (0) over the faulty images and then with ImageIntegration you can do the pixel Rejection , to delete the black pixels and replace them with the pixels from the other images. This only works if the artifact is NOT on all images.
Thanks very much for that, I am making progress on this, and here is my first result attached, the overall effect has worked, but I have this ring... I have also put the high and low rejection from the integration. What do I need to do to prevent this? Presumably something in the clone stamping process is not right? (I used a range mask to get a black source to paint over the suspect area).
TIA for an input.
 

Attachments

  • 1stattempt.JPG
    1stattempt.JPG
    157.5 KB · Views: 59
Thanks very much for that, I am making progress on this, and here is my first result attached, the overall effect has worked, but I have this ring... I have also put the high and low rejection from the integration. What do I need to do to prevent this? Presumably something in the clone stamping process is not right? (I used a range mask to get a black source to paint over the suspect area).
TIA for an input.

Very good job! You are 90% there.
Yes, unfortunately the painting done by clone stamp is not strictly zeros- even at the "hardest" setting.
I asked Juan about this once a few years ago. Given this...

There are two ways to proceed.

1. Try integrating your images again. Use the Large Scale Low rejection with perhaps 2 or 3 layers and a growth of 3 (to try). This will grow the boundaries of bits you painted.
2. When I originally developed this idea for PixInsight the GAME script did not exist. However, now it does and *will* paint black pixels (zeros) at every pixel. You would create a circle mask, using GAME, that covers the blemish (instead of painting). With this mask over each image you need to put black on... you write zeros using PixelMath. Now when you do ImageIntegration it will result in the selective rejection you want without the artifact.

I really need to update my tutorial on this to reflect the improved/easier way.

But bottom line.. you did it correctly.

-Adam Block
 
Hi many thanks. I've had a quick play and procedure 1. seems to work with a higher growth value... I'm not familiar with what's involved in 2, but will try to take a look.
In the meanwhile the on site people are looking into the actual problem....
 
Hi many thanks. I've had a quick play and procedure 1. seems to work with a higher growth value... I'm not familiar with what's involved in 2, but will try to take a look.
In the meanwhile the on site people are looking into the actual problem....

Great! That was my big "epiphany" when I first came up with the idea was to take advantage of the Large Scale feature...and it generally does work. However, there are some times it will not (it depends on the rejection method). The problem is that the cloned/painted pixels are not black at the edges. If the paint was truly zeros everywhere...this would not be an issue. (Hence the newer recommendation I made with #2).

-adam
 
Very good job! You are 90% there.
Yes, unfortunately the painting done by clone stamp is not strictly zeros- even at the "hardest" setting.
I asked Juan about this once a few years ago. Given this...

There are two ways to proceed.

1. Try integrating your images again. Use the Large Scale Low rejection with perhaps 2 or 3 layers and a growth of 3 (to try). This will grow the boundaries of bits you painted.
2. When I originally developed this idea for PixInsight the GAME script did not exist. However, now it does and *will* paint black pixels (zeros) at every pixel. You would create a circle mask, using GAME, that covers the blemish (instead of painting). With this mask over each image you need to put black on... you write zeros using PixelMath. Now when you do ImageIntegration it will result in the selective rejection you want without the artifact.

I really need to update my tutorial on this to reflect the improved/easier way.

But bottom line.. you did it correctly.

-Adam Block
Hi I'm ready to give this another go with a different set, how do 2, I now have GAME. I can figure out to create a circle mask, but not how to write zeros with PixelMath...
FWIW It looks like we have fixed the problem locally for the future...
 
When you use GAME you have an image with white circles.

1. Apply this white circle image to your data as a mask.
2. Now open Pixel Math. The expression you want is "0" (that is it... )
3. Apply this expression (the default destination is to "replace" (change) the image you drag the triangle to)

Now you will have absolute zeros everywhere that is white in your mask.
Be certain when you use GAME to make the mask..you are using a BINARY mask (not the radial gradient one).

-adam
 
Getting back to this now, have had a play but am not getting this right.
First, do I need to register and calibrate all files or do I do this to "raw files".
I assume I separate all the subs that are on the side of the meridian in question (before or after registration and calibration?) and then put them in an image container?
I can see how to create the game mask in Binary, do I apply it to one of the subs in the container and apply pixel math with expression 0 to it?
TIA
Clifford
 
Back
Top