Processing Flats problem

Dasidius

New member
Hello guys I have an issue that I can seem to resolve. When I process my picture using weighted batch pre-processing I'm getting large white gradients around the edges of my master light. The issue started when I switched to the Optolong L-Pro Filter filter for galaxy season. I have retaken all my bias, flats and dark's with the new filter. My equipment I'm imaging with is the ZWO ASI294MC Pro With a Hyperstar V4 on a C8 scope. I shoot all my shots at -10c, gain 120 and offset zero. My exposure times are five minutes. My target value I'm using for my flats is 25k ADU. I take my bias and dark's both at -10c. I use Sequence generator Pro to capture all my pictures and calibration files. I get this error message when I run batch pre- processing. **Warning: Only master bias will be used to calibrate flat frames (filter = noFilter, Binning = 1, exposure = 0.01s, Length =8) I have tried everything I can think of. Any ideas on what I can do different?
 

Attachments

  • Leo.JPG
    Leo.JPG
    86.5 KB · Views: 83
There is one detail in your description that might be the culprit: offset 0. The default offset for the ASI294 MC Pro is 30.

Please check one of the calibrated light frames (captured with the multiband filter) with ImageStatistics ('Unclipped' unchecked; the value of 'count (%)' shall be near 100 %) and HistogramTransformation (with horizontal zoom, about 50, in order to check the low range of intensities): are the calibrated light frames clipped?

If this is not the cause of the trouble, there is something else that I cannot find out from your description.

Bernd
 
There is one detail in your description that might be the culprit: offset 0. The default offset for the ASI294 MC Pro is 30.

Please check one of the calibrated light frames (captured with the multiband filter) with ImageStatistics ('Unclipped' unchecked; the value of 'count (%)' shall be near 100 %) and HistogramTransformation (with horizontal zoom, about 50, in order to check the low range of intensities): are the calibrated light frames clipped?

If this is not the cause of the trouble, there is something else that I cannot find out from your description.

Bernd
Thanks for the tip on offset, I was not aware of that and will change it ASAP. Where do I find the ImgaeStatistics at?
 
No, obviously the offset isn't the problem. If you want to upload a light, flat, bias and dark frame in FITS format and your MasterFlat, MasterBias and MasterDark in XISF format I can take a look at it.

Bernd
 
There is one detail in your description that might be the culprit: offset 0. The default offset for the ASI294 MC Pro is 30.

Please check one of the calibrated light frames (captured with the multiband filter) with ImageStatistics ('Unclipped' unchecked; the value of 'count (%)' shall be near 100 %) and HistogramTransformation (with horizontal zoom, about 50, in order to check the low range of intensities): are the calibrated light frames clipped?

If this is not the cause of the trouble, there is something else that I cannot find out from your description.

Bernd

Bernd,

I have a question related to offset for the ASI294MC Pro. I use SG Pro and set a Gain of 120 and Offset of 30 for all my shots. I'm trying to use this equation to analyze my images to determine whether my exposures are correct:

Digital Number (DN) = ((Swamp Factor (SF) * Read Noise^2/Gain in e-/ADU) + Offset in ADUs) * 2^16/2^14

I understand all the elements of this equation except maybe the Offset. If I shoot for exposure times between SF=3 and SF=10, the DNs for my images in PixInsight should be between 160 and 255 DN. However, the median DNs for my light images in PI are in the 1250 DN range. And, my Dark frames range from a median DN value of 1912 for a 1.5 sec FlatDark to a median DN value of 1920 for a 300 sec Dark.

So, it seems to me like my actual Offset is somewhere around 1900 DN. But, according to the equation above, if my Offset is 30, the 16 bit Offset displayed in PI should be 120 DN.

Do you know why my actual offset in my photos seems to be about 1900 DN when it should be 120 DN?

PS - If I look at the median DNs for my calibrated frames, they fall within the range described by this part of the equation:
(Swamp Factor (SF) * Read Noise^2/Gain in e-/ADU)

Thanks,

Ted
 
Hi Ted,

you have to differentiate between the offset setting in the driver (e.g. 30) and the shift in ADU that this will generate in the histogram:

Before A/D conversion is performed, a bias voltage (a DC voltage) is added to the analog signal in order to make sure that all analog values are greater than zero. This is necessary in order to avoid clipping of data in the low range during A/D conversion.

Offset is the parameter which controls the zero-point adjustment of the analog signal. It can be set in the camera driver or the acquisition software. Some acquisition software uses the nonstandard FITS keyword OFFSET to record this parameter.

Example: for the ASI294MC Pro, a bias voltage is applied that will generate a shift of
shift = offset x 16 ADU
after A/D conversion. So the default setting of the driver (offset = 30) will apply a bias voltage that generates a shift of 30 x 16 ADU = 480 ADU after A/D conversion. Since these data are further scaled by the camera driver from [0,16383] to [0,65535], i.e. by a factor of 4, finally the resulting shift will be 1920 ADU.

Note that the correlation between offset setting in the camera driver and resulting shift (in ADU) is different for each camera model.

Bernd
 
Last edited:
Hi Ted,

you have to differentiate between the offset setting in the driver (e.g. 30) and the shift in ADU that this will generate in the histogram:

Before A/D conversion is performed, a bias voltage (a DC voltage) is added to the analog signal in order to make sure that all analog values are greater than zero. This is necessary in order to avoid clipping of data in the low range during A/D conversion.

Offset is the parameter which controls the zero-point adjustment of the analog signal. It can be set in the camera driver or the acquisition software. Some acquisition software uses the nonstandard FITS keyword OFFSET to record this parameter.

Example: for the ASI294MC Pro, a bias voltage is applied that will generate a shift of
shift = offset x 16 ADU
after A/D conversion. So the default setting of the driver (offset = 30) will apply a bias voltage that generates a shift of 30 x 16 ADU = 480 ADU after A/D conversion. Since these data are further scaled by the camera driver from [0,16383] to [0,65535], i.e. by a factor of 4, finally the resulting shift will be 1920 ADU.

Note that the correlation between offset setting in the camera driver and resulting shift (in ADU) is different for each camera model.

Bernd

Bernd,

Thanks so much for that clear explanation! At least now I know why my FlatDarks and DarkFrames are over 1900 DN in PI. I guess the only further question is: Am I losing dynamic range by having such a high offset? I experimented with reducing the Offset from 30 down to as low as 6 and noticed that my histogram peak still clears the left side of the graph such that no pixels seem to be getting clipped. So, I could probably go down to an Offset of 6 if it helped me get more dynamic range.

Thanks!

Ted
 
I guess the only further question is: Am I losing dynamic range by having such a high offset?
You should be aware that an offset value that is sufficient to avoid clipping at gain 0 might not be sufficient at higher gain. Therefore sometimes it was recommended to optimize the offset value according to the gain value, i.e. use different offsets for different gain values. However, in my view, this is ridiculous. I guess that ZWO eventually decided to use ONE default offset value that should be appropriate for all gains. The advantage is: lower complexity, because otherwise you have to bear in mind to use the same combination of settings for the calibration frames as for the light frames.

Example:
If you use an offset of 6 with the ASI294MC Pro, the shift will be 6 x 16 ADU x 4 = 384 ADU. The impact on the dynamic range will be:

(65536 ADU - 1920 ADU) / (65536 ADU - 384 ADU) = 97.6 %,

i.e. using an offset of 30, you are loosing 2.4 % of dynamic range compared to an offset of 6. This is almost negligible.

Of course you can use a lower offset value if it is ensured that no clipping occurs. Note however, that the situation can change if you use filters (e.g. narrowband filters).

Bernd
 
Back
Top