Binning - when to do it with DSLR imaging

[iapa]

I use Nebulosity for most of my processing at the moment 'cos I can , not yet up to doing to much with Photoshop yet - that day will come.

When is the best time to apply binning if using a DSLR?

Should this be applied to each image immediately after capture, prior to processing

After stacking

somewhere in between.

Presumably you need to apply the same binning to flats, darks, bias, dark flats?

[michael8554]

It's unusual to bin OSC images from a DSLR after capture, but Craig Stark of PHD fame makes a case when using narrowband filters to capture:

http://www.google.co.uk/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&cad=rja&uact=8&ved=0ahUKEwj-96ajwv7OAhUkI8AKHZPiADQQFggiMAE&url=http%3A%2F%2Fwww.stark-labs.com%2Fcraig%2Fresources%2FArticles-%26-Reviews%2FLineFilterRecon_API.pdf&usg=AFQjCNFmRlL3ySU0VrD1VXWrh_-rZOoMgw

Michael

[ollypenrice]

If you bin a DSLR or other OSC camera output you destroy the colour information. Your pixels are filtered in an array placing RGGB in a pattern over the chip. If you bin 2X2 you will combine an arbitrary collection of those coloured filters into one binned virtual pixel which might combine two reds, a green and a blue or whatever.

(Also the software which handles OSC data is very clever and makes an educated guess as to what is going on in, say, the red channel where it is blocked by the green and blue. It does the same for the other channels.)

Since most people use binning for their colur capture, where resolution is less important, the only case for DSLR binning is in regard to narrowband. See Craig Stark above, in Michael's link.

Olly

[Stub Mandrel]

There's a subtly different alternative in DSS called 'super pixel' that takes each 4-pixel Bayer group and calculates R, G & B from the three colour channels and all together to compute luminance. This creates an image that's 1/4 the size and half the resolution as with binning, but it retains colour data, in fact it claims to be more accurate as there's no interpolation and they say its works well when you have relatively few subs.

[iapa]

Thanks for the responses everyone. It clarifies what I thought: how can you combine pixels reliably with the Bayer matrix, it's effectively just a resizing exercise interpolating from the adjoining pixels to get effectively reduced quality.

Thanks again.

And apologies for yet another daft question - they sometimes seem sensible to me when I ask them tho;.... honest

 

 

[alacant]

Unless you have a gaming laptop with loadsa memory, binning a 256Mb fits from a dslr stack is the only realistic way I've found to be able to process it in a reasonable time frame. OWTTE. At least what I understand as bin: I load the file into StarTools and hit bin. No problem with the colour. Maybe that's not what is being discussed here (?). Cheers.

[SilverAstro]

3 hours ago, Stub Mandrel said:

There's a subtly different alternative in DSS called 'super pixel' that takes each 4-pixel Bayer group and calculates R, G & B from the three colour channels and all together to compute luminance. This creates an image that's 1/4 the size and half the resolution as with binning, but it retains colour data, in fact it claims to be more accurate as there's no interpolation and they say its works well when you have relatively few subs.

That sounds interesting to try, especially as I usually dont have very many subs ( my camera needs manual op for each frame )

Thanks for the tip,   only problem is that I cant find it in my DSS, perhaps that is because I see that I am on 3.3.2 and there is now a 3.3.4 ! I presume it would be under one of the "Stacking Method" tabs ?    Found it Options > Settings > Recommended > use superpixel , also one of the FAQs says to use Bayer Drizzle with it as well , Options > Settings > RAW/Fits >RAW Files (middle panel)

Well well, that should keep me busy for a while trying that out

[dph1nm]

By experience I can tell you that Superpixel is good if your stars are bloated or slightly trailed - helps DSS to detect them.

I guess the correct way to bin DSLR data is to separate the R,G,B channels before debayering then bin each of those before carrying on with the processing. But as this is software binning you really don't gain overall s/n by doing this, so I am not sure it is worth the effort over just binning the debayered colour image at the end.

NigelM

[ollypenrice]

2 hours ago, dph1nm said:

By experience I can tell you that Superpixel is good if your stars are bloated or slightly trailed - helps DSS to detect them.

I guess the correct way to bin DSLR data is to separate the R,G,B channels before debayering then bin each of those before carrying on with the processing. But as this is software binning you really don't gain overall s/n by doing this, so I am not sure it is worth the effort over just binning the debayered colour image at the end.

NigelM

Downsizing (resampling) a conventionally debayered RGB image at the end sounds entirely convincing as a productive noise reducer. As for the rest, too much black magic...

Olly