m46 2 minutes and 2 hours

[alacant]

Hi everyone

This maybe interesting to the impatient ones among us. It consists of just 2x1 minute exposures stacked and binned to 800x600 to combat the noise. If you need to get closer and get that planetary, it's gonna have to be much longer though.

We'd be interested in any comments on the binning approach for noise. Has anyone had a go at this? You'd be able to get loadsa clusters each session although I'd imagine no use to trap fainter stuff. Still...

Cheers and thanks for looking.

canon dslr + tair3s@300mm and pn208@800mm

[vlaiv]

Binning works, and it works well. It's also precisely defined in terms of SNR gains over original image.

If you have base image, SNR gain will have same factor as number of bin samples. For 2x2 bin you will have x2 SNR of original image, for 3x3 it will be x3 SNR improvement, etc ...

Be sure to use enough precision when doing binning. Don't bin 16 bit data - bin 32bit data.

On a separate note, if you are over sampled and want to bin your data to get to proper sampling resolution, there is better way to bin than regular binning. We can call it sparse binning - it involves splitting original subs prior to stacking into number of subframes - each one containing only certain pixels depending on bin factor - much like bayer pattern contains red, green and blue in separate pixels so would each sub contain N-th pixel in X and N-th pixel in Y (so every second for 2x2 or every third for 3x3, ....). Then you stack all resulting subframes as separate frames - this increases SNR in the same way - by providing more frames to stack (by factor of 4 in 2x2 case, factor of 9 in 3x3 case, ....) but preserves pixel size, or to better put it - reduces pixel size induced blur. This is important for edge cases when you want the most detail - if you are concerned with SNR solely and you are going to bin to higher degree than needed for optimum sampling - this approach is not needed.

 

[RolandKol]

1 hour ago, vlaiv said:

Be sure to use enough precision when doing binning. Don't bin 16 bit data - bin 32bit data.

hmm,

So how does it look from the start?

If I understood/understand correctly, - initial APT or SGP (and probably any other Astro imaging software) subs are saved in up to 16bit format.

Pixinsight has an option to calibrate and save calibrated ones in 32bit format, - but how to bin them after? Further more, you probably speak about binning before calibration.

it sounds like a witchcraft!

 

[vlaiv]

28 minutes ago, RolandKol said:

hmm,

So how does it look from the start?

If I understood/understand correctly, - initial APT or SGP (and probably any other Astro imaging software) subs are saved in up to 16bit format.

Pixinsight has an option to calibrate and save calibrated ones in 32bit format, - but how to bin them after? Further more, you probably speak about binning before calibration.

it sounds like a witchcraft!

 

Depends if you are processing OSC data or Mono+ filters, but here is the simplest approach to get good SNR.

Calibrate, register, stack as you would normally. In the end when you have linear data freshly stacked   you apply bin (x2,x3,x4 to your liking) to get again linear data for each channel. If you are using Pixinsight there is command that does binning and it is called integer resample. Select average method in options. Here is link to docs:

https://pixinsight.com/doc/tools/IntegerResample/IntegerResample.html

You can try it on your data if you have something already stacked in linear phase (so no stretch, deconvolution or denoising done yet) - bin it by several factors and observe differences (x2, x3).

Also, if you are working with high resolutions, you might want to bin image if over sampled. In order to see if its worth binning - measure FWHM, divide with 1.6 and that should be your pixel scale. If you used higher sampling rate, you can bin it and see if you lost any of the detail in the image.

[RolandKol]

Ach, - I was playing with Pix Resample tool incorrectly as did not payed attention to it...  

interesting, - will give a try 100%, - especially for my ugly OIII

[alacant]

13 hours ago, RolandKol said:

Pix Resample tool

Hi. There's also this, which just works. Without the fiddling;)

HTH

[alacant]

Hi everyone

This is an example of binning after stretching. I think that this is easier because you get an idea of how far you need to bin to get the noise to an acceptable level. This is 20 minutes. The resolution is about 1200x800 but I'm not satisfied with the colour. Too magenta I think but hey, the binning looks promising:)

Cheers and thanks for looking.

**EDIT

If anyone has the time, I'd love to see corrected star colour. There's a .tif here TIA.

[Fraunhoffer]

On ‎12‎/‎03‎/‎2019 at 12:32, alacant said:

Hi everyone

This maybe interesting to the impatient ones among us. It consists of just 2x1 minute exposures stacked and binned to 800x600 to combat the noise. If you need to get closer and get that planetary, it's gonna have to be much longer though.

We'd be interested in any comments on the binning approach for noise. Has anyone had a go at this? You'd be able to get loadsa clusters each session although I'd imagine no use to trap fainter stuff. Still...

Cheers and thanks for looking.

canon dslr + tair3s@300mm and pn208@800mm

Im always pleasantly surprised how good a 800x600 picture can be.