About the subject

η Carina is the bright star in the middle. Around that star, a great nebula known as the Great Carina Nebula is visible in the southern hemisphere.

From my location, this nebula never sets, but buildings obscure it during the night for half a year, and the Carina season, as I call it, is just starting over here.

Carina is in a very busy region of the milky way, lots of very interesting targets nearby.

η Car itself is a variable double star, yellow in color visually, and varies widely in magnitude from +6.21 to +11.1. It recently (circa 1840) had a great outburst that made it the second brightest star in the sky for a while, though it no longer is as bright. The outburst produced a gas cloud called the "Homunculus Nebula", not visible here.

The great carina nebula is large and bright enough that it can be seen with the naked eye from dark locations. It's one of the largest diffuse nebulae in the sky. It surrounds several star clusters, including Tr 14, Tr 16, Collinder 232, all of which are prominently visible near the center, Collinder 228 above.

Out of frame but very near, the Gem cluster, NGC 3532 and NGC 3114. It's an amazing area for wide shots, might as well try that next ;-)

About acquisition

First project with brand-new automation modules from cvastrophoto.

- Plate solving for centering and framing (not that it was necessary for this image)

- Drift guiding

- Automated capture and dithering

In total, 102 subs were captured, out of which many had to be dropped due to post-dither instability. Software bugs just made post-dither stabilization fail, but with a few tweaks and fixes I expect much lower drop rates.

Sadly, the telescope's collimation seems off. I didn't notice it during the previews.

It needs more integration time, so I went hard on NR to compensate. Will get more subs next chance I get, to improve the dim details.

Drift guiding

Drift guiding is a variation on the common guiding techniques I'm using within cvastrophoto to improve image quality. Instead of sending correction pulses at a fixed cadence between guide exposures, a "drift" model is updated with information from those exposures, but corrections are performed continuously based on the predictive model.

This enables both shorter and longer guide exposures without sacrificing guiding accuracy, especially on mounts that require fast correction cadences like mine.

The guiding software also doesn't try to keep a "guide star" fixed in its location. It in fact has no concept of guide star, as it works with image correlation techniques. It's what I call an "all star" guider. This effectively allows very short guide exposures without chase seeing.

Guide pulses try to keep the image steady during imaging exposures, by compensating low-frequency drift (first-order derivative for now), and try to recenter in-between or at dither intervals, since large random shifts can still happen due to various factors (differential flexure for instance). It ignores high-frequency random shifts that are uncorrectible anyway, and trying to correct them just makes things worse, especially with high-backlash mounts.

The technique is rather CPU hungry, but it can still run on a Raspberry Pi 4 running out of a USB battery pack.

Version B processing

Added 1h 36m of Ha and created an HaRGB composite

Used 2 layers:

- HaGB

- HaRGB

The HaRGB has better star colors, the HaGB has more nebula detail. Combined both layers by taking the color of HaRGB and the luminosity of HaGB, with quite a few tweaks along the way:

HaGB:

- Diffusion NR - 40 steps

- HDR stretch

- Extract Lum

- Curves to raise black level

HaRGB:

- Diffusion NR 40 steps

- HDR stretch

- Chroma NR

Composite (gimp):

- HaRGB base layer

- HaGB lum, combine value at ~50%