Jump to content

Narrowband

Severe colour blotching artifact


TheMan

Recommended Posts

I bought a ZWO asi 294 mc pro a few months ago. I was really happy with it and my 72 ed. Recently I have been getting this colour blotching artifact which I cannot find a cure on how to fix it. I have tried not using the cooler, using flats, darks, dark flats etc. Unfortunately nothing is working. I'm afraid that it won't be curable as other people have been getting the same or similar artifacts. (https://bbs.astronomy-imaging-camera.com/viewtopic.php?f=21&t=8007) Any help is appreciated.

rosette copy.png

Rosette Lights_120sec_1x1__frame34_d.jpg

Link to comment
Share on other sites

Ok, in this particular case - flat file is wrong and such flat file can certainly lead to above artifacts.

Here are my recommendations:

- shoot flat darks as well - maybe you already doing that, but you did not include one flat dark in this set so I figured I should mention it anyway - these are shot in the same way you shoot regular darks - except matched to flats (exposure, gain, offset of flats).

- since you are using OSC camera - pay attention to histogram when shooting flats. Your current flat file has been over exposed and there is clipping to the right on histogram. OSC flat histogram is going to have three peaks instead of one. Common wisdom when shooting flats with mono camera is to have peak at 75-85% of histogram. When doing OSC, you want your right most peak to be in this position.

Here is histogram of your flat:

image.png.732caaea832e939e50d5931958753d61.png

and this is how you want your flat histogram to look like:

image.png.6df367c2c6a312d1bee6f97c0e8d67ab.png

(actually this one is a bit under exposed - you want your right most peak to be not at the middle but about 3/4 to the right).

 

  • Like 2
  • Thanks 1
Link to comment
Share on other sites

On 18/12/2019 at 06:33, TheMan said:

I bought a ZWO asi 294 mc pro a few months ago. I was really happy with it and my 72 ed. Recently I have been getting this colour blotching artifact which I cannot find a cure on how to fix it. I have tried not using the cooler, using flats, darks, dark flats etc. Unfortunately nothing is working. I'm afraid that it won't be curable as other people have been getting the same or similar artifacts. (https://bbs.astronomy-imaging-camera.com/viewtopic.php?f=21&t=8007) Any help is appreciated.

rosette copy.png

Rosette Lights_120sec_1x1__frame34_d.jpg

The suggested solution is to lower the camera offset. Have you tried that?

Link to comment
Share on other sites

1 minute ago, wimvb said:

The suggested solution is to lower the camera offset. Have you tried that?

As far as I could tell from supplied subs, offset is just fine - issue was due to clipping of the flats - look above histogram

Link to comment
Share on other sites

44 minutes ago, vlaiv said:

As far as I could tell from supplied subs, offset is just fine - issue was due to clipping of the flats - look above histogram

It seems from the light frame, that the offset is already low. However, several people reported the same problem, with the same camera model. To me, that suggests a camera issue, rather than a flats issue. The uncalibrated light frame has the same colour blotch. It is also very weak, and needs to be stretched a lot to reveal any signal. 

860899190_120s1x1120gain4offset_d.jpg.ec69a75d09888b8bca96080aa429c7ca.jpg

Edited by wimvb
image added
Link to comment
Share on other sites

8 minutes ago, wimvb said:

It seems from the light frame, that the offset is already low. However, several people reported the same problem, with the same camera model. To me, that suggests a camera issue, rather than a flats issue. The uncalibrated light frame has the same colour blotch. It is also very weak, and needs to be stretched a lot to reveal any signal. 

If we go by darks (and I'm hoping that darks are matched to lights), here is histogram for darks:

image.png.ca2c162994677b5d3e4beab758878fa9.png

Minimum value is way above 0 at 68, histogram is nicely shaped and there is no clipping to the left. Even if short exposure bias has issues with offset - that is not going to impact lights and darks because both have histogram way higher than 0 and no clipping occurs (and bias is not used for calibration here).

But there are couple of things that are surprising now that I've taken a look again at subs:

1. Both darks and flats have following data written in fits header:

image.png.6537f54f24c67b455ce12f281c7136f2.png

What I don't understand is camera gain and EGAIN part. Camera gain of 120 should give us less than 1 e/ADU according to ZWO data on the camera:

image.png.8f4055c04ac09c8c3bc94f5eab4e5f3d.png

For EGAIN of about 4e/ADU, gain setting should be at about 0 gain. This is confusing but can be bug in drivers with EGain being reported for gain 0 rather than actual gain setting (one can calculate actual e/ADU for 0 gain and any gain setting for ZWO cameras because gain is in 0.1dB units).

That aside, only channel that has not got clipping in flats is blue channel, and here is calibrated single sub and blue channel extracted:

image.png.e2d3947150c758440e9e57fa8ef802a8.png

It has classic LP gradient but I don't think it has anything else in there? Let me remove linear gradient to see what is left:

image.png.983eb8bad46b51e5d4acc4f2b4f473d9.png

I would call this reasonably flat fielded. There is little under correction - but that is due the fact that no flat dark was included, just to attest it is flat under correction - here is blue part of the flat:

image.png.2a1abd703b4144f24d607e97f85b4033.png

Hm, that does not explain top left corner.

But here is one more oddity for the end. These are supposed to be raw files straight from the camera, right? ASI294 has 14bit ADC and 16bit FITS is written in MSB - that means all numbers in the image should be divisible with 4.

There are bunch of odd valued pixels in the image and all numbers divisible with 4 are even. This should not happen - either drivers are buggy or image has been tampered with. Maybe there is issue with data transfer or storage (data gets corrupt for some reason).

 

 

Link to comment
Share on other sites

36 minutes ago, wimvb said:

It seems from the light frame, that the offset is already low. However, several people reported the same problem, with the same camera model. To me, that suggests a camera issue, rather than a flats issue. The uncalibrated light frame has the same colour blotch. It is also very weak, and needs to be stretched a lot to reveal any signal. 

I see what you mean - it looks like over exposed flat is only part of the story.

What do you think about odd pixel values in the image? I have ASI1600 and it uses 12 bit ADC - every number from raw image that I ever got was divisible with 16 (2^4). I also have ASI178 - it has 14bit ADC - again same thing, only this time all numbers were divisible with 4 (2^2). Never had odd pixel values in raw image.

Link to comment
Share on other sites

19 minutes ago, vlaiv said:

But here is one more oddity for the end. These are supposed to be raw files straight from the camera, right? ASI294 has 14bit ADC and 16bit FITS is written in MSB - that means all numbers in the image should be divisible with 4.

I don't think the drivers have a gain conversion table. I wouldn't trust the e/ADU, only the set value of 120.

18 minutes ago, vlaiv said:

What do you think about odd pixel values in the image?

No idea, really. As you say, the values should be divisible by 4. I'm not sure it's data corruption, because that would more likely corrupt the whole file. 

Link to comment
Share on other sites

Here is another interesting find:

- dark sub in zip above - 0 pixels that are not divisible with 4

- flat sub in zip above - 0 pixels that are not divisible with 4

- light sub - ~37.5% of the pixels have value that is not divisible with 4 (and therefore wrong value if we take it that ASI294 is 14 bit ADC and we compare to other 14bit asi cameras and flat from this camera).

And now for final surprise:

0 green pixels have wrong values

~75% of red pixels have wrong value and ~75% of blue pixels have wrong values

Huh, hold on, is there setting for ASI294 in drivers to adjust B and R value? Maybe this has been fiddled with if exists?

Link to comment
Share on other sites

A guy on my local group was having colour blotch problems with a ZWO colour camera.  I believe he ended up doing away with flats, though how he processes his images now, I have no idea.

I'll contact him, not sure if he is a member on here.

Carole 

Link to comment
Share on other sites

10 minutes ago, vlaiv said:

Huh, hold on, is there setting for ASI294 in drivers to adjust B and R value? Maybe this has been fiddled with if exists?

That wouldn't surprise me too much. The sensors that zwo uses come from various "families". Larger sensors are generally used in consumer cameras. Smaller sensors may be developed for machine vision an surveillance applications. Some basic on sensor image processing isn't unheard of. 

Otoh, that should affect darks and lights alike. 

Edited by wimvb
Link to comment
Share on other sites

8 minutes ago, wimvb said:

That wouldn't surprise me too much. The sensors that zwo uses come from various "families". Larger sensors are generally used in consumer cameras. Smaller sensors may be developed for machine vision an surveillance applications. Some basic on sensor image processing isn't unheard of. 

Otoh, that should affect darks and lights alike. 

Quite right. Strange thing is - it does not affect Flats either.

Btw - I did the same thing you did above - debayer single sub without calibration, and here is my result:

image.png.94d422ae1c706cedff6bfbf3795d8ae9.png

I guess difference is to the fact that you are wiping your image, and this one is just with aligned black point. It sort of looks like light leak maybe?

Is there any chance camera/scope connection is letting some light pass in?

Link to comment
Share on other sites

14 minutes ago, vlaiv said:

Strange thing is - it does not affect Flats either.

In this case, that's correct. But generally, flats are taken under "day time" light conditions. But here there is so much light pollution that even lights have a high intensity level. 

I used DBE in Pixinsight with only eight samples along the diagonals in order to reduce the vignetting. This never affects weak structure such as the blotches. I start to suspect the quality of the bayer layer or the antialiasing layer on the sensor. Ordinary interference as in Newton rings could create blotches. That would not show in darks, and be masked in flats. This would be a very serious quality issue for Sony. 

Link to comment
Share on other sites

1 minute ago, wimvb said:

But here there is so much light pollution that even lights have a high intensity level.

In only 120s with 72mm scope? Don't really think so ...

Offset is low - so it can't be that, average value of dark is around 80e (120 gain is close to unity, so I'll just use 1e/ADU).

On the other hand green background level is about 6000e.

That system is sampling at just a bit over 2"/px (somewhere around 2.3 or so - not sure if it is using reducer or not).

I image from mag 18.5 skies, and my sub 1 minute long, in green filter with 80mm scope and sampling of 2"/px has something like 200e background value, which means that two minutes would be 400e or about x15 less. That is 3 magnitudes difference.

Or 15.5 mag skies. I don't think there is such LP - worst that I could find would be center of Rotterdam with mag16.5

Center of Melbourne is mag17.75 so not so severe.

I think this points again to light leak - that can also raise levels of image.

 

 

Link to comment
Share on other sites

Well, the flat is over exposed, the light frame is quite uniform with a median and mean value of about 0.35, and the dark looks normal in so far that it shows the ordinary ray-burst amp glow and a few hot pixels.

Here's the deBayered light, using superpixel mode, so that no data was interpolated. I have no further ideas to offer.

sreen.jpg.f9e8f6a9a5127889c7d468bca4b1c097.jpg

Link to comment
Share on other sites

13 hours ago, wimvb said:

The suggested solution is to lower the camera offset. Have you tried that?

That's what I've done, I use 10 as the offset and also no bias frames and I use darkflats. I'm not the best processor but achieved this a couple of weeks ago with my Sharpstar 61EDPH.

spacer.png

  • Like 1
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.