@Joaquin: Thank you for the good explanation. You hit the nail on the head.

The indi_pylibcamera issue #65 is closed. Prof Huster got raw Bayer images. The tool he used interpreted the data as monochrome. That was also the reason for the checkerboard pattern in the lighter regions of his image.

The header of the FITS image has an attribute "BAYERPAT" which tells the color order in the Bayer pattern. A monochrome FITS image does not has this attribute. As far as I know this makes the difference between monochrome and raw Bayer images in FITS format.

KStars has a setting to enable debayering before showing FITS images on the screen. That allows to see colored images in the preview window. It does not affect the stored image data. Likely CCDCiel has a similar feature.

For my opinion using raw Bayer images is preferred over RGB images. All color cameras I know use Bayer filter. The RBG images are generated in an image signal processor which does the debayering. But this image signal processor usually calculates with integer numbers which leads to loss of information. This is not critical for good exposed images. But in astrophotography the interesting parts of the images are extremely underexposes: the faint nebulas between bright stars are often hidden by the camera noise. Only a special processing makes these details visible:

• Correction of pixel errors: You should use Dark frames to subtract individual pixel bias. If you are a professional you can also use Flat frames to compensate for individual pixel sensitivities. I do not have the equipment for doing Flat frames. But the HQ camera has very homogeneous pixel sensitivity and my optics do not produce vignetting. So I don't use Flat images.
• Debayering: The best debayer algorithms are nonlinear. That's the reason why you need to do debayering after the correction of linear pixel errors.
• Aligning and stacking: This improves your noisy pixel data: the mean grows with the number of frames while the standard deviation grows with the square root of the number of frames. The stacking of hundreds of frames makes details visible which were hidden by noise in the single frames.
• Postprocessing: Here you correct colors/contrast and apply filter.

All these steps should be done with floating point arithmetic! There is a lot of good astro software available for doing this. My preferred free software for deep-sky images is Siril .

The 2x binning of the RPi HQ camera is aware of the Bayer pattern and done in hardware. It only bins pixel of the same Bayer color. The resulting is still a raw Bayer color image.

The indi_pylibcamera indi_pylibcamera driver runs in my setup (2 telescopes with 3 cameras: 2x RPi3 and one RPi0) since over a year without any major issues. In particular of the bad seeing conditions (very bad light pollution due to a near town) I am proud of the pictures I make. See attached image of M27, made of 68 exposures with each 20 second.

Read More...