Monofied DSLR

[bilgebay]

Dear Bernd,

Thank you for your detailed analysis. This is extremely helpful for sure. I shot these frames approx 10 days ago and I remember having great difficulty in finding the right exposure time for not saturating the R and B pixels while illuminating the G pixels sufficiently. Still, I suspect I might have saturated R and B channels. I will collect some new data when I go to my observatory in 10-12 days time.

As for the checkerboard pattern I didn't do anything other than calibration. I have uploaded one of the calibrated frames to DB --> https://www.dropbox.com/s/6gu1x1s6srnzasd/m101_30sec_1x1__0006_c_cc_dm.rar?dl=0

The checkerboard is clearly visible in the uncalibrated Lightframe

I can also try shooting with camera's original firmware, without any hacks. But I think, it is generally accepted that Nikon plays with pixel values too.

Lastly, I cleaned the sensor after I completed the test shots. The flats are much cleaner now... should have done it prior to the test shots but I was in a rush to see if the camera would produce any useable data or not.

Thanks again! Much appreciated your help.

Sedat

 

[bilgebay]

Maybe this screenshot can explain why the checkerboard pattern disappeared after the calibration ?

 

[bulrichl]

I would like to take a deeper look at this. Can you please upload the following data: some (say 4) uncalibrated light frames plus the corresponding normally calibrated light frames?

Bernd

 

[bilgebay]

Sure, here are the raw files --> https://www.dropbox.com/s/nem2fpxqmpnudoq/RAW.rar?dl=0
and here are the calibrated files --> https://www.dropbox.com/s/0a6y245n5liwff6/Calibrated.rar?dl=0

Thanks

 

[bulrichl]

OK, it's obvious why a normal calibration already will remove the checkerboard pattern:

The deviation between R/B channels and G channels affects the signal in light, dark and flat frames. Since it is a multiplicative effect, this deviation is canceled out in the flat field correction process. So you need not apply the workflow that I outlined for image calibration, the normal image calibration is just fine. There is only one benefit of producing the modified MasterFlat: one can judge the outcome of the removal of micro lenses and color filters from the sensor.

Bernd

 

[bilgebay]

My conclusion is the firmware hack is not doing what it's supposed to do and still Nikon is multiplying the R and B values by~4 to make up for the G pixels... I will look into this more and come back here with my findings.

 

[bulrichl]

Obviously this "multiplication" is performed before the analog to digital conversion. This can be recognized in the histogram:
1) In the histogram, at 16-bit (64K) setting and a horizontal zoom of 400, there are no gaps in case of the R and B channels.
2) Saturation of the signal occurs at 16383 ADUs, for ALL channels (range of the 14-bit ADC: [0,16383].

So actually the "multiplication" is an amplification of the analog signal (a voltage) of the R and B channels, and it is done in the hardware.

Bernd

 

[bilgebay]

Then the question arises as to why the bias pixel values remain unchanged...

I really appreciate your input but I don't feel comfortable continuing this discussion here as it drifted from a PI issue towards a Nikon issue. Maybe I should start a thread on Cloudy Nights DSLR Forum. If the moderators are fine I'll be happy to go on here of course.

 

[bulrichl]

This is admittedly a very special problem. I understand that you want to continue the discussion in an appropriate Nikon forum. I only wanted to show that PixInsight is able to assist you in such special cases as well, and I'm glad when it helped. If you want, you can keep me up-to-date via personal message.

Bernd

 

[Juan Conejero]

Hi Sedat,

If the moderators are fine I'll be happy to go on here of course.

Of course, please don't hesitate to continue with this topic here. It is a very special and interesting problem, as Bernd has pointed out, and the solutions implemented with PixInsight are undoubtedly relevant here.

 

[bilgebay]

Thank you Juan. With pleasure.

Best wishes

Sedat

 

[bilgebay]

Dear Bernd,

I have shot dark, bias and flat frames with both the original firmware and the NickonHacker patched firmware and uploaded those files to DB.

https://www.dropbox.com/s/d92694vkadodko4/Nikon_SGP.rar?dl=0

With the Nikon Firmware things look even more weird to me.

Flat frames were shot with a 7nm Ha filter in front of a Nikon 35mm lens.

Next week I will be able to test this setup under the stars with proper length exposures. Hope to come up with a nice integration.

Cheers

Sedat

 

[bulrichl]

These frames look totally different from the Master Calibration files that you uploaded previously. Now there is NO multiplicative factor, but (in case of the frames captured with the "TrueDark" Firmware) only a small difference of about 21 - 22 ADU between R/B and G channels:

             R/B     G
bias   0 s   127    148
darks 30 s   137    159

I guess that different settings in the camera caused the previously deviating behavior. Anyway, keep the "TrueDark" Firmware with the current camera settings. Presumably astro images captured with these settings will come out successfully.

The important difference between Nikon Firmware and "TrueDark" Firmware is: bias and dark frames captured with the Nikon firmware are severely clipped and therefore useless. Please check it for yourself comparing a bias each of Nikon and "TrueDark" firmware in histogram (horizontal zoom: 300) and statistics (option 'Unclipped' disabled, watch count(%)). Bias and dark frames captured with the "TrueDark" firmware are not clipped.

Bernd

 

[bilgebay]

Hi Bernd, thank you for looking into this once again. Other than changing the ISO from 1600 to 200 I have changed nothing else on the camera. I'm suspecting that ISO setting has a contribution to the multiplication coefficient of the R/B pixel values. While researching this issue I have read lots of forum posts and noticed that the recommended ISO is 200 for this camera. So I decided to use it on all the calibration frames. As you are suggesting, I will keep all the settings as they are and start shooting some Ha frames next week and share my results here.

I have noticed the clipping especially on the master dark that's why I opted for sending the individual subs to you. When I integrated the darks more than half of the pixels were being rejected by low and high rejection paramaeters which I have never experienced to date with any of my other cameras.

Thanks and clear skies

Sedat