Lamplighter05

17 hours ago, vlaiv said:

I identified such effect with bilinear interpolation and small angle rotation - here is brief explanation of what happens in that particular case:

Look at two columns of pixels and very small angle rotation:

Red line represents pixel centers after rotation (after rotation it will be vertical, but since we haven't rotated yet - it is tilted and pixels are at their regular positions).

In order to rotate - we need to calculate pixel values along that red line. Now for each pair of pixels, if we do linear interpolation between its values - at the bottom we will almost exclusively use left pixel value - it will be like 99%:1% ratio and on the top - it will be other way around 99% of right pixel and only 1% of left.

These two extreme cases introduce minimal pixel to pixel correlation, or softening / blur in the image, but as we move along vertical - we progressively move towards 50%:50% case - which perfectly mixes two pixel values or averages them - and that reduces noise the most / adds most blur / cuts off high frequencies the most.

Same thing happens in horizontal - except above diagram is rotated 90 degrees.

This creates rectangular pattern across the image of different levels of pixel to pixel correlation and thus different levels of blur that will pop up after certain types of processing.

Simplest way to see this is to generate random noise, then rotate it by small angle and then clip low values. Standard deviation variation (due to this blur) will produce this rectangular pattern:

rectangular pattern wavelength will depend on rotation angle. Above is produced with 2 degrees rotation.

If polar alignment is not as good - there will be field rotation over the course of imaging session, and subs from the end of session will be slightly rotated compared to those from beginning. In order to align them for stacking - the will need to be rotated back by small angle - which causes above effet

 

Great analysis. Sampling science is fascinating. I'm now wondering what algorithms are used for 'dithering' - simple alt/az offsets will need  a minimum delta to help avoid introducing these pixel boundary conditions. Presumably 'on sensor' pixel binning will mitigate, or having an oversampling sensor to glass setup?

On 28/11/2022 at 13:50, vlaiv said:

Stacking process.

I believe that if data were to be stacked using Siril with say Lanczos interpolation for sub alignment - pattern would go away, but I'd be happier if someone actually confirmed that by doing it.

It looks like it could be a 'rounding error' aka quantisation error in the stacking algo, with slight local variations causing a moiré effect with the data.  It could be rotational calculations on partial pixel coverage as the data values are clipped or 'binned' across pixel boundaries. One way to try 'fixing' it would be to pre-scale the stacking images (prior to stacking)  with a simple 2x  linear interpolation - basically creating a artificial bin2 of the data. Then stack the result and finally scale the result back down by 0.5x. 

16 hours ago, johninderby said:

Good to see another CC owner. 👍🏻👍🏻👍🏻

Yes I was impressed with the build quality of it. Very solid scope.  One thing I found was that sometimes it was difficult to remove the cover if you overtightened it  Hence the knob. 

Indeed, as I already had an TS APO 80mm triplet refractor and am very happy with the build of that, I was really happy to see, and conformed via your thread here this was my next option as opposed to a Mak. I'm looking to the autumn and winter for most of my telescope time ... dew and temperature plus CA my main concerns. Funnily, the CC I received has a red front tube baffle from the current RC as advertised. I did check I hadn't got some factory 'Frankenstein' hybrid - seems not - only cosmetic! Lol. My next 'investment' will be some Vixen to Losmandy mounting saddle, so I can use the broader rail on the mount side rather than the Vixen bar. I think your lid solution is a great addition.

Separately did you tweak the collimation in the end, if so what method did you use? Cheers for any info.

On 12/10/2019 at 10:46, johninderby said:

Minor jobs. 🙂

Got the finder aligned and found I needed the two inch spacer added to focus on something a mile away. 

Found the front cover was a bit awkward to remove as there was nothing to grab and it is a snug fit. So off to the spare parts box for a black stainless steel knob to attach to the cover. Much better. 👍🏻

I took the plunge and invested in one of these too - just arrived today. Very nice indeed. One thing I noticed about the front cover is that it 'locks' in place by twisting once placed into the OTA. Its about a 10 to 20 degree twist to 'lock' and 'unlock'.