What is wrong with my flats?

[Zocky]

Daniel, is this the first time you have that kind of problem, or you have vignetting on older images too?

[DanielF]

Daniel, is this the first time you have that kind of problem, or you have vignetting on older images too?

This is only the second image I've done with this camera.
The first actually had the same problem, but I just thought it maybe was because of the full moon that time. I gave up on that image...


I have previously imaged with a DSLR and I did not have this problem with that camera.

/Daniel

[Zocky]

I have previously imaged with a DSLR and I did not have this problem with that camera.

With DSLR you have different back focus, so now you camera is in the different position. I still think your secondary mirror could cause this type of vignetting.

[Carlos Milovic]

Are the darks already bias substracted? Could you post a raw dark sub?

[DanielF]

Are the darks already bias substracted? Could you post a raw dark sub?

Hi.
I guess the dark gets bias subtracted in the PP script?

Here's a single dark sub. No calibration.
https://dl.dropboxusercontent.com/u/8802413/M45%20problem/Dark_001.fit

/Daniel

[DanielF]

I have previously imaged with a DSLR and I did not have this problem with that camera.

With DSLR you have different back focus, so now you camera is in the different position. I still think your secondary mirror could cause this type of vignetting.

If that is the case, shouldn't the flats still correct it even though the vignetting has increased?

[Carlos Milovic]

To me, all your calibration frames looks ok (although, I would recomment taking A LOT of them. Bias specially, is very noisy, and you can see the same patterns in the dark). The flats from the box and the sky flat are very similar (you can judge by yourself by substracting the bias to them, and then divide one by the another with PixelMath, using a expression like this: "$T*mean(flat)/flat" and disable rescaling.

As I said earlier, the problem seems to be with your lights. They have an extra pedestal, that does not match your bias. If there is nothing wrong with the capture settings, or file format, you may also check for parasital light, although for the pedestal value I doubt that this is the problem.
To fix this, as a workaround, open your main bias. Then, create a new ImageContainer, with all the raw light frames in it. Then use PixelMath with the following equation: "$T-mean(bias)", where bias is the identifier of your master frame image. Do not rescale. And apply this instance to the ImageContainer. It should create new files with the bias pedestal substracted. Finally, run the new light through the batchprocessing script, or the image calibration process.

[Phil Leigh]

I think I understand the asymmetric vignetting now...
http://www.skyandtelescope.com/howto/diy/3306996.html

[DanielF]

To me, all your calibration frames looks ok (although, I would recomment taking A LOT of them. Bias specially, is very noisy, and you can see the same patterns in the dark). The flats from the box and the sky flat are very similar (you can judge by yourself by substracting the bias to them, and then divide one by the another with PixelMath, using a expression like this: "$T*mean(flat)/flat" and disable rescaling.

As I said earlier, the problem seems to be with your lights. They have an extra pedestal, that does not match your bias. If there is nothing wrong with the capture settings, or file format, you may also check for parasital light, although for the pedestal value I doubt that this is the problem.
To fix this, as a workaround, open your main bias. Then, create a new ImageContainer, with all the raw light frames in it. Then use PixelMath with the following equation: "$T-mean(bias)", where bias is the identifier of your master frame image. Do not rescale. And apply this instance to the ImageContainer. It should create new files with the bias pedestal substracted. Finally, run the new light through the batchprocessing script, or the image calibration process.

Hi Carlos.
I did as you suggested and then ran a new round in the batch preprocessing script and now it looks better. As you said it still contains a gradient but that can probably be fixed with DBE. Do you think you could explain why this is working? You first subtract the mean bias and then I suppose bias is subtracted again in the script, am I right?



I don't know what could have happened during capture of my lights. I use Nebulosity to control the camera and it doesn't let me adjust gain or offset, just the exposure time so I don't really know how to fix this issue.

/Daniel

[Phil Leigh]

Do you take your darks with the camera on the scope?

[DanielF]

Do you take your darks with the camera on the scope?

No, not this time, I removed the camera and covered the opening, as well as leaving the shutter closed while exposing.

[Carlos Milovic]

Phil, I don't believe there is a problem with the darks. They seem good enough, and they do not require to be on the scope at all.

Daniel, the overcorrection is a clear indicator that there is a pedestal problem somewhere, either on the flats or the lights. The "model" behind our images is that they consist on several sources:

Img_adquired = (Img_ideal+noise)*uneven_illumination*pixel_sensitivity + bias + dark_current + noise

Since we can capture the external sources in the flat, dark and bias frames, we end with the following equation to recover the "ideal" image (of course, ignoring the noise):

img_calibrated = ( img_adquired - bias - (dark-bias) )/(flat - bias)

If we have an additional pedestal value in the lights, we end with:
img_calibrated' = img_calibrated + pedestal/(flat-bias)
(we have the same image as before, plus a gradient from the inverse of the flat... i.e. an over correction).

Also we may have these problems with a negative pedestal in the flats.

I found the value by experimentation. It kind of makes sense that the pedestal is the mean of the bias, since that is the electrical pedestal value, although I don't know why it is doubled in your images. Also I experimented with a gain factor 2x (dividing the lights by two), and then calibrating, but it seems that in that case the bias and dark frames do not match the data.

That's it. I encourage you to investigate this thing further. Try making the capture in another software, and see if the problems still appear.

[TeeJay]

Hello Daniel

I don't read the hole thread, but for me it looks like the most usual problem with reflectors... Light which does not travel the right optical path. A Newton for example suffers from "light from behind". If there is any light source (laptop) behind the telescope, it causes strange illumination issues, especially with flats… I always put a shoppingbag on the rear end of the tube and the problem is gone.

Regards
TeeJay

[Phil Leigh]

This is exactly what I was getting at with my question about the darks...
If there is a light leak (for example via the focuser drawtube and/or the rear cell in the case of a Newtonian) it can show up in darks taken on the scope if the shutter is open and it can show up in the flats to a small extent (they are normally very short exposures) and in the lights.

I have had to add some tape to my Baader steeltrack focuser to block a light leak and several other folks on the forum I use have reported the same problem with various focusers and scopes.

The problem was first found by looking at on-scope darks taken with DSLR's which leave their shutter wide open during dark acquisition.

The problem really messes up the image calibration process. Obviously it will vary each time because the incident light finding its way into the scope/camera is always different (e.g. from the moon etc).

I'm not saying this IS the problem here - just that it is a situation where image acquisition appears to have worked correctly but images will not calibrate/integrate correctly.
regards
Phil

[Carlos Milovic]

Yes, it may be parasital light, but is really strange that it matches the bias pedestal... Anyway, it is a thing that must be discarded too with some darks in the scope.