"Binning" question

[Spoon]

Hi all,

Quick question, I'm purchasing a CCD tomorrow and was just wondering, what is this "binning" talk about? I don't get what it means. Can anyone do a "binning for dummies" walk through please.

Cheers

Cam

[iksose7]

Basically binning will increase sensitivity of the CCD but at the cost of resolution have a read through Sara's post about binning, good read and will give you an idea about what you may want to do in the future!

http://swagastro.weebly.com/to-bin-or-not-to-bin.html

Callum

[Spoon]

Basically binning will increase sensitivity of the CCD but at the cost of resolution have a read through Sara's post about binning, good read and will give you an idea about what you may want to do in the future!

 

http://swagastro.weebly.com/to-bin-or-not-to-bin.html

[lensman57]

Hi all, Quick question, I'm purchasing a CCD tomorrow and was just wondering, what is this "binning" talk about? I don't get what it means. Can anyone do a "binning for dummies" walk through please. Cheers Cam

You can only Bin the pixels from a Mono CCD BTW, it just doesn't work with OSC.

A.G

[Spoon]

You can only Bin the pixels from a Mono CCD BTW, it just doesn't work with OSC.

 

A.G

Hi,

Yeah the camera I'm getting is a mono

Cam

[gkec]

Basically binning will increase sensitivity of the CCD but at the cost of resolution have a read through Sara's post about binning, good read and will give you an idea about what you may want to do in the future!

http://swagastro.weebly.com/to-bin-or-not-to-bin.html

I think that article misses the point of binning and draws some incorrect conclusions.

First of all you have to consider how binning is performed.  If you imagine a CCD as an array of buckets (i.e the pixels) and they fill with water (in reality electrons) depending on the number of photons that land in the bucket.

At the end of the columns of rows of buckets there is an extra row of empty bigger buckets and at the left hand end of that row there is a single bigger bucket which we can call the measuring bucket.

In a normal 1x1 bin readout what happens is that the first row of buckets is emptied into the bigger row of buckets, the 2nd row into the first and so on until the last row is emptied into the last but one and is left empty.  So in effect the image shifts down by one row.

Now a similar process happens to the row of big buckets in that the first big bucket is emptied into the measuring bucket, the second bucket into the first, and so on so the image is shifted one pixel to the left.

The amount of water in the measuring bucket is measured and that value is taken as the value of the bottom left hand pixel.  The measuring bucket is then emptied and the left most bucket in the big bucket row is emptied into the  measuring bucket, the next bucket is emptied into the left most bucket so the row again moves one pixel to the left and the measuring bucket is measured again to get the value of the pixel on the bottom row second from the left.

This is repeated for the whole row.  Once the whole row has been measured and is empty. Then first row of buckets,which hold what was in the second row of buckets, are again emptied into the bigger bucket and the whole process repeated to measure the second row.

This rather tedious process is repeated until all the buckets have been measured.

Right now for 2x2 binning.

The process is the same but instead of emptying one row of bucket into the bigger buckets and going through the measuring process on the row the second row of buckets is also emptied into the single row of bigger buckets.  So the image moves down by two rows and the big row of buckets hold the water from two rows of buckets.

Similarly for the row, two big buckets are emptied into the measuring bucket  so that there is 4 buckets of water in the measuring bucket.  Again this bucket is measured and the value taken as the 4 pixels in the bottom left hand corner.  Again this is repeated for all the buckets.

One thing is immediately apparent is that you can have m x n binning where m and n is any number you like so long as the buckets don't over flow.

So we can bin -- so what?

Firstly the measurement process is not 100% accurate.  You can visualize his as the water sloshing around a bit whilst you are trying to measure it.  If say you can measure the water level to say 1 cm then the measurement is more accurate if you measure 4 buckets worth at once than measuring each individual bucket.  This is known as readout noise.  The readout noise is the same whether you measure a single pixel or a binned pixel.

Also the measurement take time compared to all the bucket emptying.  So it is quicker to measure 4 buckets at once rather than each bucket.  When you have a few million buckets it adds up.

Now back to the article.

It is claimed that a binned pixel is more sensitive than a single pixel.  Well you are measuring 4 buckets of water rather than just one. Of course you get more signal.  You should get, assuming equal illumination, 4 times more signal.  However only 3 was measured.  Lost a whole bucket of water somewhere.

There are two explanations for this.  

1) To measure this properly you have you use a flat field with constant illumination so that all the pixels get roughly the same amount of light.

2) The bucket emptying mechanism is done hurriedly so water is split. I sort of imagine an "It's a knockout" competition where you have to empty buckets dressed as a dragon on a moving roundabout, for those who remember that.

This is true in practice as the electronics that effectively do the bucket emptying have to strike a balance between speed and accuracy.  Too fast and you empty stuff all over the floor, too slow and it takes forever

The binned pixels were shown to have less noise. Well statistically the noise reduces by the square of the sample count. So a sample count of 4 gives a times 2 noise reduction. Which was sort of what was measured (50 cf 70), the 15 difference was quite possibly the readout noise difference.  This is not altogether a direct consequence of binning on chip, you could do this after acquisition in software and get the same result.  To compare directly you should compare CCD binning with software binning and I don't think this was done in in this case.

As for FWHM you are measuring the width at different resolutions so that needs to be taken into account.  Exercise for the reader as it is too late too think.

[Spoon]

Basically binning will increase sensitivity of the CCD but at the cost of resolution have a read through Sara's post about binning, good read and will give you an idea about what you may want to do in the future!

 http://swagastro.weebly.com/to-bin-or-not-to-bin.html

[ollypenrice]

Let us know your focal length, focal ratio and pixel size. This will allow us to give you a realistic estimate of the usefulness or otherwise of binning. In the end, however, you need to do what Sara did and test the effects for yourself. There are countless times I've found theory not to accord with practice in this game and in the end you look at the pictures, not the numbers.

Olly

[Uranium235]

Use an FOV calculator loaded with the gear youre going to use, set it to 2x2 bin mode then it will give you the resolution. Anything up to 3.5" p/p is quite passable for working on large nebulae, but for galaxies I still prefer 1x1.

http://www.12dstring.me.uk/fov.htm

Try not to think of it as sensitivity or more photons collected - its more like an improvement in the S/N ratio (more signal, less noise). But, there is no substitute for just going out and trying it!

[swag72]

I just decided to try to quantify the difference that I was seeing with my kit between 1x binning and 2x binning. I was certainly doing nothing other than reporting my specific findings in a quest to get better data.

I am certainly no scientist and would not suggest for one minute that my findings would be the same as anyone else's. I feel that for my personal kit, I was able to show my gains. I think that this is a worthwhile exercise for anyone to do if they want to try to ascertain gains of binning within their specific setup.

When this was posted elsewhere,it got very bogged down in the scientific minutia, which I for one failed to understand and grasp.

If I now know that I get a better signal by x amount when I bin compacted to when I don't, then I feel that I am personally better placed to make a judgment regarding how I want yo gather data to achieve my aims.

I hope that the link posted to my experimentation didn't confuse any issues, I certainly wanted to keep it simple and to the point..... Sadly something that I cannot say for any ensuing discussions that have resulted in any place that this was posted I'm a simple girl, who likes to keep it simple and basic..... That is my level!

[ollypenrice]

I just decided to try to quantify the difference that I was seeing with my kit between 1x binning and 2x binning. I was certainly doing nothing other than reporting my specific findings in a quest to get better data.

I am certainly no scientist and would not suggest for one minute that my findings would be the same as anyone else's. I feel that for my personal kit, I was able to show my gains. I think that this is a worthwhile exercise for anyone to do if they want to try to ascertain gains of binning within their specific setup.

When this was posted elsewhere,it got very bogged down in the scientific minutia, which I for one failed to understand and grasp.

If I now know that I get a better signal by x amount when I bin compacted to when I don't, then I feel that I am personally better placed to make a judgment regarding how I want yo gather data to achieve my aims.

I hope that the link posted to my experimentation didn't confuse any issues, I certainly wanted to keep it simple and to the point..... Sadly something that I cannot say for any ensuing discussions that have resulted in any place that this was posted I'm a simple girl, who likes to keep it simple and basic..... That is my level!

Bravo. You have two important qualifications underpinning your piece. 1) You actually did the comparison in practice. 2) You take a great many excellent pictures. You didn't get into the theoretical side, you measured what you captured and told us. What you found was not quite what I was expecting you to find because your gain in S/N was more than I've come to expect from my setups (or my previous ones. I haven't measured the 11000 and don't have a FL worth trying it binned anyway, at the moment.)

Olly

[Scott]

Let us know your focal length, focal ratio and pixel size. This will allow us to give you a realistic estimate of the usefulness or otherwise of binning. In the end, however, you need to do what Sara did and test the effects for yourself. There are countless times I've found theory not to accord with practice in this game and in the end you look at the pictures, not the numbers.

Olly

I just decided to try to quantify the difference that I was seeing with my kit between 1x binning and 2x binning. I was certainly doing nothing other than reporting my specific findings in a quest to get better data.

I am certainly no scientist and would not suggest for one minute that my findings would be the same as anyone else's. I feel that for my personal kit, I was able to show my gains. I think that this is a worthwhile exercise for anyone to do if they want to try to ascertain gains of binning within their specific setup.

When this was posted elsewhere,it got very bogged down in the scientific minutia, which I for one failed to understand and grasp.

If I now know that I get a better signal by x amount when I bin compacted to when I don't, then I feel that I am personally better placed to make a judgment regarding how I want yo gather data to achieve my aims.

I hope that the link posted to my experimentation didn't confuse any issues, I certainly wanted to keep it simple and to the point..... Sadly something that I cannot say for any ensuing discussions that have resulted in any place that this was posted I'm a simple girl, who likes to keep it simple and basic..... That is my level!

Two people who certainly know their craft so I won't be argueing the point with either . I'm just curious though, I notice Sarahs kit is vastly different from mine, mine being a 600mm scope and tiny 414ex chip. Although I'm yet to do any lrgb I was under the impression that 2x2 is used on rgb simply (or not) for colour and 1x1 on L for detail. is there any arguement (apart from time) that would suggest that 1x1 across the board is a better option for a small chip with larger pixels (I think)? I get that trial and error is the best teacher but we aren't all blessed with french/spanish weather. 

Thanks for any input and I hope I haven't deviated too far from the op's question

[ollypenrice]

Two people who certainly know their craft so I won't be argueing the point with either . I'm just curious though, I notice Sarahs kit is vastly different from mine, mine being a 600mm scope and tiny 414ex chip. Although I'm yet to do any lrgb I was under the impression that 2x2 is used on rgb simply (or not) for colour and 1x1 on L for detail. is there any arguement (apart from time) that would suggest that 1x1 across the board is a better option for a small chip with larger pixels (I think)? I get that trial and error is the best teacher but we aren't all blessed with french/spanish weather. 

Thanks for any input and I hope I haven't deviated too far from the op's question

Binning colour is often used as a compromise and is OK provided the pixel scale remains reasonable when binned. It has the disadvatage that the RGB-only stars are often 'blocky' and since I try not to apply lum to stars (not always practicable) I do like to have my RGB stars unbinned. There are plenty of well heeled imagers with big scopes of very long focal length who bin everything including luminance because that still gives them a pixel scale close to the resolution allowed by the sky. Personally I doubt I would ever bin colour if it took me below 4 arcsecs per pixel or so. This excludes all the rigs we have here at present.

Olly

[Spoon]

Let us know your focal length, focal ratio and pixel size. This will allow us to give you a realistic estimate of the usefulness or otherwise of binning. In the end, however, you need to do what Sara did and test the effects for yourself. There are countless times I've found theory not to accord with practice in this game and in the end you look at the pictures, not the numbers.

Olly

FL: 800mm

FR: F4

PS: no idea, it's an HX916

Cam

[Spoon]

Anything up to 3.5" p/p is quite passable for working on large nebulae

My camera and scope combo would be 3.45p/p if I have put values in right.

[ollypenrice]

FL: 800mm FR: F4 PS: no idea, it's an HX916 Cam

Pixel size is 6.7 microns? That gives a very sensible 1.73 arcsecs per pixel according to http://www.12dstring.me.uk/fov.htm I'm not getting the same pixel scale as you but maybe I went up a gum tree!

I think this is bang on the money for trying binned colour. Personally I don't do that but I'm in a luxurious position with regard to clear sky time. Pressed for time I might well give binned colour a wizz. 1.73 arcsecs per pixel is close to what I get in the TEC140/Atik 11000. Here's a full size to see how that resolution looks. This is a big chip but the resolution of detail is slightly lower than yours, though nearly the same. http://ollypenrice.smugmug.com/Other/Best-of-Les-Granges/i-NNsK787/0/O/LEO%20TRIPLET%20TEC140%202015%20web.jpg

Olly

My camera and scope combo would be 3.45p/p if I have put values in right.

That is very much what we get from the Tak106/Atik 11000. It's OK and very tolerant of seeing and guiding. It doesn't resolve to the limits of the scope by any means and here I wouldn't bin the colour.

[Spoon]

FL: 800mm FR: F4 PS: no idea, it's an HX916 Cam

 

Pixel size is 6.7 microns? That gives a very sensible 1.73 arcsecs per pixel according to http://www.12dstring.me.uk/fov.htm I'm not getting the same pixel scale as you but maybe I went up a gum tree!

 

I think this is bang on the money for trying binned colour. Personally I don't do that but I'm in a luxurious position with regard to clear sky time. Pressed for time I might well give binned colour a wizz. 1.73 arcsecs per pixel is close to what I get in the TEC140/Atik 11000. Here's a full size to see how that resolution looks. This is a big chip but the resolution of detail is slightly lower than yours, though nearly the same. http://ollypenrice.smugmug.com/Other/Best-of-Les-Granges/i-NNsK787/0/O/LEO%20TRIPLET%20TEC140%202015%20web.jpg

[Uranium235]

FL: 800mm FR: F4 PS: no idea, it's an HX916 Cam

Ahhh, I dont think binning will be your best option then. For it to be worthwhile you need a high pixel count to start off with (8mp or more), the 916 is just 1300x1030 (1.39mp). Therefore the resulting image size would be just too small - it might be ok for binning colour, but for lum you should stick to 1x1 and maintain your image scale.

[nightster]

Maybe I am missing something about the whole pratice of binning. How does it save time? Shorter subs?

[Spoon]

Maybe I am missing something about the whole pratice of binning. How does it save time? Shorter subs?

From what I've read I think it basically reduces resolution but increases sensitivity of the sensor