DBE / ABE is resulting in "rainbow" colors and increased grainy noise

Hi Frank,
Dr. Frankenkeim said:
I kept thinking about why "the iPad did not work" for my flats and came across the following thread on CN:

https://www.cloudynights.com/topic/667524-preparing-color-balanced-osc-flats-in-pixinsight/

In short, DonR seemed to describe a very similar issue to the one I was experiencing above. That got me thinking...
Click to expand...
I replied to this CN thread as well, and I will repeat it here:

--------------------------------------------------------------------------------
You picked an old and outdated thread to comment on.

1. The "issue" that Don pointed out is not a problem at all. Please see https://pixinsight.com/forum/index....ld-correction-with-data-of-osc-cameras.13261/ .

2. Nevertheless (because this topic was brought up again and again), Juan Conejero decided to implement a separate flat scaling factor for each color channel as an option in the ImageCalibration process. This was accomplished in PixInsight v1.8.8-6 (released in August 2020). Meanwhile the WeightedBatchPreProcessing script (since v1.5.2 , released in December 2020) supports this option as well. With this option enabled (it is enabled by default), there is no change of color cast caused by the flat field correction.

Before you commented on it, the CN thread ended in July 2019 when this option was not yet available in PixInsight's ImageCalibration process.

If your flat frames exhibit 3 widely separated peaks (one for each color channel), this is caused by the light source, the different color sensitivity and the characteristics of the filters in the color filter array of the sensor. This should not pose a problem as long as the weak channel contains enough signal and the strong channel is not clipped. Therefore, the correct exposure of the flat frames is very important, much more for OSC cameras than for monochrome cameras. If this condition cannot be met with the light source you are using, you'll have to take a different light source that generates better color-balanced flats.

Bernd
--------------------------------------------------------------------------------
 
Servus Bernd,

Thank you for your quick reply. I don't disagree with you. The issue was caused by the combination of my unmodified OSC DSLR and the original flat method, which resulted in an underexposed, non-balanced flat. Thats exactly what I wanted to figure out.

The statistic examples that I attached were all processed with version 1.8.8-6+, with all defaults enabled.

Using the "monitor manager" tool to separately change the screen brightness and the monitor R,G,B settings, did allow me to get to a more balanced flat with my existing equipment. For the exposure readout it seems I need to solely rely on PI since all other histogram readouts in PS, APP, DR, and Nikon NX-D show the histogram as overexposed, while statistics gets me a median of ~2k for my 12 bit camera (set to 16 bit in statistics).

Either way I am glad that I figured it out and hopefully this will help somebody else.

Thank you again for your quick reply and greetings from LA. CS

Frank
 
Dr. Frankenkeim said:
For the exposure readout it seems I need to solely rely on PI since all other histogram readouts in PS, APP, DR, and Nikon NX-D show the histogram as overexposed, while statistics gets me a median of ~2k for my 12 bit camera (set to 16 bit in statistics).
Click to expand...
What you observed is normal. All histograms that camera monitors display and many histograms shown by image processing software (meant for daylight photography) are already stretched. However, for astrophotography you need to be able to display an unstretched histogram. When using the Av exposure mode with my Canon EOS 600D (= Rebel T3i), an exposure correction factor of about 2 to 3 is needed. The histogram that is displayed on the camera monitor then seems to be hopelessly overexposed, but the flat frames are correctly exposed.

Bernd
 
FWIW, I have the same issues. More prevelant in Wide Band than Narrow Band, but there as well. If I actually perform a stretch on each image to more or less make them similar in appearance, avoiding losing detail, then combine to make a color image, the rainbow effect is substantially reduced. Then apply dbe/abe and it is better still. One thing I noticed is that if I apply just dbe/abe prior to combining, even with linearizing the images, I get the rainbow efffect. If I do dbe/abe after combining, I then get the rainbow (regardless of linearizing or not) Seems that combining a slightly stretched, non-linear image works better although that may not be the preferred way.
 
You must log in or register to reply here.
Back
Top