Thank you for that recognition my friend! :-) There's always a lot of behind the scenes work to get a video ready, the recording part that we end up seeing is almost always the shortest part of the whole process, haha! Thank you for your support and clear skies!!
In the now old days with CCD camera's, binning genuinely helped S/N as binning reduced the readout noise of 4 pixels to a single readout noise of the 1 superpixel. But with CMOS camera's that is not the case and all that happens is a software bin, averaging adjacent pixels and reducing image size, so there is no benefit. I would always recommend imaging at bin 1 and if you feel the image looks a little oversampled, then bin/reduce image size in your chosen software to suit. Nothing is lost.
Nice work explaining this. One thing you are not mentioning which I see no one talk about is the fact that increasing the binning to a higher value increase SNR at the expense of resolution which can be a very powerful tool in processing. For example In processing you can leverage that to mitigate noise in really faint part of an object that often does not have a lot of structure in these regions. bin an image at bin 4, stretch it ( you can stretch it further because of higher snr), resample the image to bin 1 value using interpolation and then blend all the super faint stuff back into the main image at bin 1… you are essentially using variable resolution workflow which dynamically lower resolution and increase SNR where details is not there but noise is.
thanks for putting examples behind the numbers luke! hardware binning did have a purpose with CCD sensors, when electrons were added up per row it actually did do more than just downsampling in terms of noise. but with CMOS sensors, there’s no difference between binning while shooting and just downscaling your image afterwards, unless you’re worried about storage space or something. and as you’ve shown, even when very “oversampled”, there’s still plenty of gains to be had by using everything your sensor is capable of!
Thank you Gabe!! I'm glad you found this enjoyable mate, it certainly seems like oversampling in the traditional sense is not something to be worried about so much anymore! :-)
Shouldn’t you also take into account the seeing? If the seeing is poor, the oversampled image isn’t really capturing the detail that it feels like? And of course some of the post-imaging “sharpness” from drizzle and/or deconvolution is sort of artifactual, using math to guess what the missing values are. But it did seem like your sharpened images on bin=1 were “better” than what you got from bin=2, even though I suspect your pixel scales were far lower than your seeing? Very interesting!
I was planning on googling binning when this popped up on my homepage! Talk about perfect timing!! Super helpful and answered all of my questions. Thank you so much!!
Your comparison videos have so very much helped me make good imaging settings decisions. Have also helped me understand the different concepts to begin with. Thank you, Luke.
Perfect explanation Luke! And great example of how it is always better to be over sampled than under even if you have to drizzle to get yourself oversampled. Thanks and clear skies mate!
This was great information and the demonstration was perfect. I’ve never done binning because I simply thought that I would lose resolution. However, I am inclined to experiment with this using my ASI533 on the C8. Thank you
Very interesting results Luke. I can definitely see the individual results from the different levels of binning. I've always wondered what would happen if I did any binning on my images. Might have to give it a try and see. Great video my friend! Clear skies!!!!
Thank you mate!! I definitely feel it's always worth a try on your own system for this kind of thing! :-) Hope you have a lovely weekend my friend!! All the best, Luke
Just what I needed to wake me up this morning Luke 😀! Got my brain thinking! Very well explained and demonstrated! Great vid mate, we need more of these to help us understand the technical side of this crazy hobby 🤣 Clear skies!
Great work! I enjoyed the livestream you did while preparing for this video. It’s nice to hang out while imaging or when the weather is bad. Either way is a win!
Very very interesting. Especially the part about BlurX and Bin1 vs Bin2. I normally go for Bin2 with my ASI 294MM PRO and EdgeHD 9.25 because of over-sampling but I'll absolutely experiment with Bin1 now. Thanks!
Very informative and interesting. Love all the comparison videos you're doing, there are a lot of misconceptions in this hobby by people that recommend stuff because it works for them, or that's what someone told them to do, but in many cases it could be good or bad. It's nice that we get to test some of these things with actual data. Thank you.
Really interesting video Luke and as always well explained. For me I think of all the advancements in software the biggest game changer has been blur xterminator. Why anyone would spend the money we do on equipment and not invest a relatively small amount on software like pixinsight and blur xterminator I've no idea. 👍
Excellent presentation Luke. I've been thinking about this quite a lot recently for my RC which has a long focal length as well. I think in my case binning 2x2 would be the best compromise (especially on a full frame sensor) to take advantage of the 2x increase in SNR. Whether or not to drizzle as well would be an interesting experiment as drizzling reduces SNR and may well cancel out the benefits of binning.
Thanks ever so much Pete mate!! - I think you're spot on really, bin2 and using APP to drizzle a small amount (1.2/1.3x) would be a really nice compromise at this focal length, and should help make storage and stacking of those massive full-frame files a little easier!! :-D Clear skies mate!
Thanks for this Luke. Really well explained and helped my understanding tremendously. Very much appreciate your contributions to the broader community. :)
I remember decades ago when the only cameras around were from SBIG and cookbook cameras (for those on a budget) binning was used to gain sensitivity and increase well depth because you were increasing the capture area of the cell. Typically the L channel was bin-1 for max resolution but R, G, & B were binned 2 to pickup sensitivity with a slight loss in resolution (but that did not matter since it was just color info). Have these benefits gone away with our modern CMOS imagers?
Great question mate!! - with CCD sensors binning should be more efficient as they only have one injection of read noise per given read area, CMOS need to read out each individual pixel before binning can occur, leading to as many samples of read noise as there are pixels being read - but the good news is CMOS has such a low read noise in the first place, and the read noise is calculated within a square root, that it equals out in the end and CMOS binning ends up being just as efficient as CCD binning in practice 👍 Hope that helps!
Great video Luke as always. One thing that buggers me though, I would have thought though that the higher you bin, the more light sensitive your resulting pixels would be, i.e. in bin2 you would end up with the quarter of the resolution of Bon1, but with each resulting pixel capturing 4 times as many photons? Sorry for that ignorant question, never done anything elase but bin1.
Hey Paul! It's interesting to do these experiments and see what actually happens isn't it, we should see a near doubling of SNR for each bin level increase but how that actually looks on screen isn't too striking I'd say 👍 I have a fairly new video on 'f ratio myth' which might be interesting for you! Clear skies mate
Interesting Video Luke, thanks. I’ve often heard that one can Capture the RGB data binned 2x2 and LUM at 1x1 for the detail. Is this still a relevant and appropriate technique? 🧐
It's certainly possible, but you may as well capture 1x1 in most cases then simply resample afterwards 👍 (unless the storage/stacking speed benefits make bin2 RGB just too appealing!)
Thank you for the wonderful comparisons Luke. Much appreciated. I’m reluctant to show my ignorance, but what exactly is “over/under” sampling? Seems like it should be self evident, but I still feel I’m missing something in the definitions.😊
Thank you very much indeed Joe!! - RE: under/oversampling, undersampling I suppose could be defined by if point sources are looking pixelated when viewed at your usual viewing size, oversampling is perhaps a bit woolier to define without knowing more about someone's setup and seeing conditions! Cheers! :-D
Very interesting & well explained mate I'm guessing you're shooting this target using a mono camera? I only ask as it makes me wonder why colour cameras have bin options as when you use any of then except bin 1 they shoot mono.
Very interesting, but my self, see more detail on my screen on bining 1*1 In my case, my C8 with my camera is recommended for binning 2*2 haven't tried yet, but will do same experiment once M42 comes back for winter.
Hi Luke, I switch from OSC to mono recently so my question will certainly sound "basic" (if not dumb...). When you shoot in bin 1x1, what scale setting do you use in PI for drizzling in your post-calibration. Is it "2" as most people suggest? Thank you!
Hey David! :-) A good question I think! - I actually don't use PI for drizzle, but AstroPixelProcessor instead, where you could use a droplet of 0.5 and scale of 2 to achieve a 2x drizzle effect, - hope that helps!
I'd say go it depends a bit on your scope and how much detail it can resolve, worth giving a go at bin1 for sure, dependant on target try drizzle too if you're going for max resolution! :-)
What happens if you have the situation of oversampled data, choose to bin to gain speed and signal:noise… then take the image that is essentially 1/4 the size and go Image 200% (splitting that binned 2x2 pixel back into 4 pixels), then run deconvolution and finish processing?
For anyone wondering the sampling ratios on display here -
Bin1 - 0.28
Bin2 - 0.56
Bin3 - 0.84
Bin4 - 1.12
Sounds like the 1-2 arc-sec/pixel sweet spot needs to be re-examined.
There are several hours of work in your video presentation Luke, bravo. Continue your excellent work.
Thank you for that recognition my friend! :-)
There's always a lot of behind the scenes work to get a video ready, the recording part that we end up seeing is almost always the shortest part of the whole process, haha!
Thank you for your support and clear skies!!
In the now old days with CCD camera's, binning genuinely helped S/N as binning reduced the readout noise of 4 pixels to a single readout noise of the 1 superpixel. But with CMOS camera's that is not the case and all that happens is a software bin, averaging adjacent pixels and reducing image size, so there is no benefit. I would always recommend imaging at bin 1 and if you feel the image looks a little oversampled, then bin/reduce image size in your chosen software to suit. Nothing is lost.
Nice work explaining this. One thing you are not mentioning which I see no one talk about is the fact that increasing the binning to a higher value increase SNR at the expense of resolution which can be a very powerful tool in processing. For example In processing you can leverage that to mitigate noise in really faint part of an object that often does not have a lot of structure in these regions. bin an image at bin 4, stretch it ( you can stretch it further because of higher snr), resample the image to bin 1 value using interpolation and then blend all the super faint stuff back into the main image at bin 1… you are essentially using variable resolution workflow which dynamically lower resolution and increase SNR where details is not there but noise is.
thanks for putting examples behind the numbers luke! hardware binning did have a purpose with CCD sensors, when electrons were added up per row it actually did do more than just downsampling in terms of noise. but with CMOS sensors, there’s no difference between binning while shooting and just downscaling your image afterwards, unless you’re worried about storage space or something. and as you’ve shown, even when very “oversampled”, there’s still plenty of gains to be had by using everything your sensor is capable of!
Thank you Gabe!! I'm glad you found this enjoyable mate, it certainly seems like oversampling in the traditional sense is not something to be worried about so much anymore! :-)
Shouldn’t you also take into account the seeing? If the seeing is poor, the oversampled image isn’t really capturing the detail that it feels like? And of course some of the post-imaging “sharpness” from drizzle and/or deconvolution is sort of artifactual, using math to guess what the missing values are. But it did seem like your sharpened images on bin=1 were “better” than what you got from bin=2, even though I suspect your pixel scales were far lower than your seeing? Very interesting!
I was planning on googling binning when this popped up on my homepage! Talk about perfect timing!! Super helpful and answered all of my questions. Thank you so much!!
You're so welcome!
Finally a video on this topic. I'm well undersampled at 3" per pixel, but drizzling 2x helps a lot!
Awesome to hear you've been waiting for a vid like this mate! :-D hope it was enjoyable!
The great people of the internet needed this video. Thank you for such a wonderful video and explanation!! You rock :D
Haha! :-D Thanks ever so much Sarah!! Really glad you enjoyed it!
Your comparison videos have so very much helped me make good imaging settings decisions. Have also helped me understand the different concepts to begin with. Thank you, Luke.
Thanks so much for watching!! :-D
Perfect explanation Luke! And great example of how it is always better to be over sampled than under even if you have to drizzle to get yourself oversampled. Thanks and clear skies mate!
Thanks ever so much Dave mate!!
Excelent work Luke.
In the end, it depends how much resolution can you get from your sky with your gear.
This was great information and the demonstration was perfect. I’ve never done binning because I simply thought that I would lose resolution. However, I am inclined to experiment with this using my ASI533 on the C8. Thank you
Thank you so much for watching mate! glad it was useful :-) Clear skies!
That was a lot of work Luke, well done! And fascinating as well!
Thank you mate!! :-D
Very interesting results Luke. I can definitely see the individual results from the different levels of binning. I've always wondered what would happen if I did any binning on my images. Might have to give it a try and see. Great video my friend! Clear skies!!!!
Thank you mate!! I definitely feel it's always worth a try on your own system for this kind of thing! :-)
Hope you have a lovely weekend my friend!!
All the best,
Luke
Just what I needed to wake me up this morning Luke 😀! Got my brain thinking! Very well explained and demonstrated! Great vid mate, we need more of these to help us understand the technical side of this crazy hobby 🤣 Clear skies!
Thank you so much for your kind message Simon!! :-D I really appreciate it mate!
Great work! I enjoyed the livestream you did while preparing for this video. It’s nice to hang out while imaging or when the weather is bad. Either way is a win!
Thanks mate!! It was nice to share the stream with people :-D Thanks for watching!!
Very very interesting. Especially the part about BlurX and Bin1 vs Bin2. I normally go for Bin2 with my ASI 294MM PRO and EdgeHD 9.25 because of over-sampling but I'll absolutely experiment with Bin1 now. Thanks!
Cheers mate!! - I think it's well worth an experiment with deliberate oversampling!! :-) Clear skies!
Very informative and interesting. Love all the comparison videos you're doing, there are a lot of misconceptions in this hobby by people that recommend stuff because it works for them, or that's what someone told them to do, but in many cases it could be good or bad. It's nice that we get to test some of these things with actual data.
Thank you.
Agreed mate! - nothing beats an actual test for me, they're a lot of fun too :-) Clear skies!
Hi luke great explanation and very easy to follow thanks Luke for taking the time to make these videos
Thanks so much for taking the time to watch and comment Tony!! It's very much appreciated my friend :-)
A superb video Luke and a proper simple explanation of this topic. Great! :)
Thanks mate!! :-D glad you liked it!
Great explanation Luke! I’ve watched a few vids on this over the years and it’s been as clear as mud to me! Finally fully get it. Cheers! 👍🏻Noddy
Excellent to hear it mate!! Thank you for watching and sharing! :-D
Really interesting video Luke and as always well explained. For me I think of all the advancements in software the biggest game changer has been blur xterminator. Why anyone would spend the money we do on equipment and not invest a relatively small amount on software like pixinsight and blur xterminator I've no idea. 👍
Agreed mate, bxt is absolutely the best money I've spent so far in astro! :-) Clear skies to you!
Excellent presentation Luke. I've been thinking about this quite a lot recently for my RC which has a long focal length as well. I think in my case binning 2x2 would be the best compromise (especially on a full frame sensor) to take advantage of the 2x increase in SNR. Whether or not to drizzle as well would be an interesting experiment as drizzling reduces SNR and may well cancel out the benefits of binning.
Thanks ever so much Pete mate!! - I think you're spot on really, bin2 and using APP to drizzle a small amount (1.2/1.3x) would be a really nice compromise at this focal length, and should help make storage and stacking of those massive full-frame files a little easier!! :-D
Clear skies mate!
Great video, Luke! Your graphs were perfect - not crappy at all, but very illustrative. Thanks for the great explanations.
I appreciate that Greg! - Thanks so much :-)
Thanks for this Luke. Really well explained and helped my understanding tremendously. Very much appreciate your contributions to the broader community. :)
Thanks ever so much mate!! :-) Clear skies to you!
I remember decades ago when the only cameras around were from SBIG and cookbook cameras (for those on a budget) binning was used to gain sensitivity and increase well depth because you were increasing the capture area of the cell. Typically the L channel was bin-1 for max resolution but R, G, & B were binned 2 to pickup sensitivity with a slight loss in resolution (but that did not matter since it was just color info). Have these benefits gone away with our modern CMOS imagers?
Great question mate!! - with CCD sensors binning should be more efficient as they only have one injection of read noise per given read area, CMOS need to read out each individual pixel before binning can occur, leading to as many samples of read noise as there are pixels being read - but the good news is CMOS has such a low read noise in the first place, and the read noise is calculated within a square root, that it equals out in the end and CMOS binning ends up being just as efficient as CCD binning in practice 👍
Hope that helps!
@@lukomatico Awesome to hear. Heck might be an idea for a future video....one that shows possible sensitivity advantages of binning (wink-wink).
Great analysis Luke, thanks. Apologies if I missed it, but how would you rate the seeing during the capturing of this data?
Thanks mate! - it was very much just an average night! :-) Clear skies!
like 227 👍🏻
Thank you! Cheers mate!!
Fantastic info , well presented . Thanks Luke !
Great video Luke as always. One thing that buggers me though, I would have thought though that the higher you bin, the more light sensitive your resulting pixels would be, i.e. in bin2 you would end up with the quarter of the resolution of Bon1, but with each resulting pixel capturing 4 times as many photons? Sorry for that ignorant question, never done anything elase but bin1.
Hey Paul! It's interesting to do these experiments and see what actually happens isn't it, we should see a near doubling of SNR for each bin level increase but how that actually looks on screen isn't too striking I'd say 👍
I have a fairly new video on 'f ratio myth' which might be interesting for you!
Clear skies mate
Interesting Video Luke, thanks.
I’ve often heard that one can Capture the RGB data binned 2x2 and LUM at 1x1 for the detail. Is this still a relevant and appropriate technique? 🧐
It's certainly possible, but you may as well capture 1x1 in most cases then simply resample afterwards 👍 (unless the storage/stacking speed benefits make bin2 RGB just too appealing!)
Thank you for the wonderful comparisons Luke. Much appreciated. I’m reluctant to show my ignorance, but what exactly is “over/under” sampling? Seems like it should be self evident, but I still feel I’m missing something in the definitions.😊
Thank you very much indeed Joe!! - RE: under/oversampling, undersampling I suppose could be defined by if point sources are looking pixelated when viewed at your usual viewing size, oversampling is perhaps a bit woolier to define without knowing more about someone's setup and seeing conditions!
Cheers! :-D
Very interesting & well explained mate I'm guessing you're shooting this target using a mono camera? I only ask as it makes me wonder why colour cameras have bin options as when you use any of then except bin 1 they shoot mono.
Hey Tich mate!! Aye this was taken with a mono cam, osc binning is an interesting subject all of itself for sure!
Clear skies bud! :-D
Excellent explanation nice job mate 💪🏼💪🏼💪🏼
Thank you mate!! :-D
Thanks Luke.
Very interesting, but my self, see more detail on my screen on bining 1*1
In my case, my C8 with my camera is recommended for binning 2*2 haven't tried yet, but will do same experiment once M42 comes back for winter.
Thanks for sharing mate!!
Hi Luke great video mate
Glad you enjoyed it mate!! 👍👍
Hi Luke, I switch from OSC to mono recently so my question will certainly sound "basic" (if not dumb...). When you shoot in bin 1x1, what scale setting do you use in PI for drizzling in your post-calibration. Is it "2" as most people suggest? Thank you!
Hey David! :-) A good question I think! - I actually don't use PI for drizzle, but AstroPixelProcessor instead, where you could use a droplet of 0.5 and scale of 2 to achieve a 2x drizzle effect, - hope that helps!
Bins 💡 Thanks mystery solved 😀
Ok Luke - question for ya….
With 115mm with
.8 reducer in bortle7
What Bin and Drizzle level will help my images best?
Using PI, BXT, etc..?
I'd say go it depends a bit on your scope and how much detail it can resolve, worth giving a go at bin1 for sure, dependant on target try drizzle too if you're going for max resolution! :-)
What happens if you have the situation of oversampled data, choose to bin to gain speed and signal:noise… then take the image that is essentially 1/4 the size and go Image 200% (splitting that binned 2x2 pixel back into 4 pixels), then run deconvolution and finish processing?
how did you get on with the backyard universe upgrade?
Good bit of kit mate really, haven't had the newt out a great deal to be honest but last o checked it's held collimation and is ready to go 👍
Interesting.
Thanks so much mate!! :-D
Dithering just seems like pixel shifting.
aows tha Bin
Ah not bin too bad sithee! 😂