Something wrong with flats

[amitshesh]

I have captured flats with sharpcap for a long time now. Suddenly now I am struggling.

I have captured flats using sharpcap using both sky illumination as well as my LED panel. The histogram says OK and I confirmed that the white curve is in the middle. But then when I apply the flats, I get this reverse illuminated image (attached below). I do not know what I did, or which sharpcap setting (if any) I changed to get this.

I'm using a QHY 533C on a C8 with the starizona night owl 0.4x reducer.

When I was just randomly trying things to see if I can fix this, I disabled darks. Suddenly the image became much better! There was vignetting as I expected, but this reverse halo effect vanished. Sure enough bringing the darks back reproduced this image. Now I am really lost as to what is happening. I don't understand what is causing my flats to fail, and how disabling darks is "helping".

Capture_00001_WithDisplayStretch-reduced.png (896 KiB) Viewed 3666 times

I'm also attaching the camera settings saved by sharpcap during this capture.

Capture_00001_WithDisplayStretch.CameraSettings.txt

(1.4 KiB) Downloaded 108 times

Lastly, here is the master flat.

MasterFlat_18_29_51_offset=4.979%-reduced.png (840 KiB) Viewed 3666 times

[amitshesh]

And here are the camera settings that were used to capture the master flat (forgot to attach it in my original message).

MasterFlat_20_55_32.fits.CameraSettings.txt

(1.17 KiB) Downloaded 91 times

[oopfan]

Something to consider that your MasterFlat settings file says that FrameType=Bias:

[QHY533C]
CameraSerialNumber=516bc14bbaefa3829
FrameType=Bias

[admin]

Hi,

flats causing overcorrection is usually associated with the pedestal signal level of the light frames not being properly accounted for in the flat correction process. This is usually handled in one of two ways

* (Optimally) Using dark frames when you capture the light frames
* Using the percentage value recorded in the filename of the flat frame saved by SharpCap

In your case, since you are using a dark frame, we need to concentrate on that. You need to check the camera settings used for the dark frame and make sure that they match those being used for the light frames, otherwise the dark subtraction will leave a constant offset value in the resulting image, which will then get (incorrectly) corrected by the flat, leading to the effect you are seeing.

Check that the following settings are the same (at minimum) between your dark and light frames:

* Exposure and gain
* Offset/black level/brightness
* White balance settings
* Bit depth and read mode (if your camera has read modes)

To check, look at the average brightness of your dark frames (in terms of where the histogram peak is). That should match the position of your histogram peak in light frames pretty closely - if they are significantly different you have an issue in this respect.

cheers,

Robin

[amitshesh]

Thanks Robin! I played around with my settings two nights ago.

1. I retook my darks. Indeed that solved the problem. I still do not know what caused my earlier darks (they were from October, for a cooled camera) to not work.

2. I remembered that I had changed the white balance settings when I was attempting to take pictures of the lunar eclipse.

I have two camera settings:

1. gain 70 exposure 2s: this is what I use to go to a target, platesolve, etc.
2. gain 60 exposure 30s: this is what I use for live stacking.

I used my latest darks, with the flats I took a week ago (the ones that precipitated the above problem).

When doing live stacking, I found the images to be very bad (bad thresholding and banding unlike anything I had seen before). The live stacking histogram showed that the red and green channels were being badly clipped (peaks were beyond the left edge), and the green peak was visible. Changing the white balance settings from 10 each to 15 seemed to fix this at least partly. When I was between targets with the 1st camera settings above, a white balance of 15 caused the above reverse-correction effect. But bringing them down to 10 each made this go away.

So now I'm confused how the white balance settings in the image control work, and specifically whether those settings are important even when I am taking calibration frames (darks and flats). I have not had a chance to retake my flats and try, but should I ensure that my darks and my flats are taken with the same image control settings and I should stick to those numbers during imaging?

Most of all, I do not remember what my "original" white balance settings were, and there is no option to "reset" them. I did try disabling the "load camera settings from previous session" but that did not work.

Finally, does sensor analysis affect/change any of these settings? I have another thread about sensor analysis, but I'm wondering if that analysis changed something in the settings.

If it helps, I'm using a QHY 533C camera. I did switch to my other camera, an ASI 662MC and found that the white balance settings and range were different. So this is a camera-specific setting which makes it even harder for me to determine what values to use.

[admin]

Hi,

for most cameras, the white balance settings apply a multiplication factor to the pixel values after they are transferred to the PC - for instance if you have increased the blue balance then maybe all the blue pixel values are being multiplied by 1.25 to brighten the blue channel. Using your ZWO camera, the 'default' white balance values that do nothing are 50/50. Unfortunately QHY cameras don't have that sort of standardization, and different models have different ranges and different 'default neutral' values, so I can't easily tell you what the defaults are for your camera.

There is usually a way to find out... Set the camera to RAW8 mode and show the main histogram. Get the camera showing a scene that has plenty of mid level pixels (not too much under/over exposed, so that the histogram runs across the entire graph). Now adjust the colour balance and keep watch for either spikes or gaps in the histogram of the colour channel you are adjusting - spkies means values higher than their neighbours, often you will get a few normal values, then a spike. Alternatively you may get gaps where the count drops every few steps along the graph. Spikes are usually a sign that the colour balance control is reducing the brightness, gaps that it is increasing it. Also, when the colour balance is reducing the brightness you may find that the histogram stops short of the right hand side of the graph, even when overexposed. This adjustment lets you find the neutral colour balance setting where there are no gaps/spikes and the histogram for that channel reaches all the way to the right. As far as I know, you should expect the same neutral balance value to apply to all 3 channels.

Hope this helps,

Robin

[amitshesh]

Can you confirm that the white balance settings need to match for lights, darks and flats? Or that these settings do not matter for the calibration frames?

[admin]

Hi,

the white balance settings for the light frames must match those for the dark frames. Additionally, if you are using 'dark flats' when creating your master flat frame then the white balance settings for the flat frames must match those for the dark flat frames. There is no need for the settings used for darks/lights to be the same as those used for flats/dark flats though.

cheers,

Robin