Encouraged by the results obtained on M51 with my Samyang 135 paired with a borrowed ASI 178 MM, I bought an ASI 178 MC to try to produce a color version. After two years spent on improving my acquisition & reduction techniques, here is the result. I hope you'll like it!

The unguided subs were acquired over 8 nights with my Samyang 135 @ F2.8 on my AZ-EQ6 driven by EQMOD with PEC correction enabled and dithering at every frame.

I initially thought that the halos around stars in my first BW attempt were caused by out of band IR. I thus installed an L2 Astronomik in the beam. With the color sensor, the now obiously red halos were still present, meaning that they were caused by in band emission, not IR.

Focusing with either a zoomed live view or a Bahtinov mask, I was systematically finding a best focus position resulting in similar green and blue FWHMs (blue being a bit sharper) and a much wider red channel.

At that focus position, a star close to M51 in a raw stack is shown in the top row of rev. B for each channel. The stack was bayer drizzled at x4 resolution with 50% droplet size. The FWHM values are estimated from Gaussian fits (dashed lines). The red channel is 2.3 times wider than the other two, which are only 1.4 native pixel wide (3.8"/pixel assuming that the true focal length is 130 mm given 2.4 micron pixels and the scale returned by Astrobin's plate solving).

I could not find a way to separate the channels in the live views of the acquisition S/W that I'm using. I finally found a better focus position for the red channel using my laptop screen in night mode, thus showing mostly the red channel of the debayered live view.

The second row of rev B. shows that at that second focus position, the FWHM in the red channel was half its original value (6.2" instead of 12.8" ), and the green and blue channels were degraded from ~5.3 to ~6.4". This implied that in order to achieve the best spatial resolution I needed to use the green and blue channels from focus position #1 and the red channel from focus position #2, as shown in the third row of rev. B. The resulting stack is shown in rev. C (raw, luminance stretch only). The total integration time is 19.4 hours for the green and blue channels and 15.6 hours for the red.

Using an OSC camera channel by channel is of course extremely inefficient. To build the red channel, 1/4 of the pixels is used and with drizzle droplets shrunk by 50%, each one covers 1/4 of its parent pixel. I thus takes a minimum of 16 dithered subs to completely fill the plane. I wanted to see how far I could go with the simplest setup -the Samyang and an OSC- and at this stage the longitudinal chromatism is the limiting factor. A mono camera and filters could provide better focusing while being much more efficient.

The merged focus stack of rev. C was then processed with wavelet regularized Richardson-Lucy deconvolution. The bottom row of rev. B shows a deconvolved star with a FWHM of ~3.1", or 0.8 native pixel. The final crop was obtained by luminance stretch, saturation and two passes of unsharp mask at 20 and 4 pixels.

A BW version is compared to my first attempt in rev. D.

An animation of the increased resolution provided by the drizzle algorithm is given in Rev. E. Note that the native resolution frame was obtained by 4x4 binning of the final deconvolved image, not by processing at the native resolution.