ASCOM driver binning

james_screech

How does the ASCOM driver bin data?

For example in 2x2 binning does it add the values for the 4 pixels together and then multiply by and them multiply by 4 to bring the value into the 16 bit range or does it average the 4 pixels and then multiply by 16 to bring the value into the 16 bit range?

James

chad

Hi,

Ofcourse it should be 4, why is it 16?

For an example, for a over exposure image to bin, each pixels of value is 65535, if you multipy by 16, the result value is not 65535, then it is wrong.

Thanks

Chad

james_screech

The camera has a 12bit ADC, so the pixels from the camera are in the range 0 - 4095, so with binning 2x2 you could multiple the individual ADC reading by 16 and average them or sum the ADC readings (giving a range of 0 - 16380) and then multiply by 4.

James

james_screech

chad wrote:

Hi,

Ofcourse it should be 4, why is it 16?

For an example, for a over exposure image to bin, each pixels of value is 65535, if you multipy by 16, the result value is not 65535, then it is wrong.

Thanks

Chad

Sorry if I'm being a little slow, but do you mean that in 2x2 binning you sum the ADC values and them multiply by 4?

James

chad

Hi，

Bin has nothing to do with ADC.

For example, 12 bit ADC, for a pixel, when it is over exposure, it will output 2^12=4096. I think there is no doubt about it. But in computer system, the length of data usually is 8bit, 16bit, 32bit or 64bit. It is not 12bit, so we made a 4bit left move. It means we've magnified the data 16 times. then 4096×16=65536.

Now we have 4 pixels to make bin. Each of them is over exposure. So every one will ouput 4096, it is 12 bit. We have 2 ways to bin.

First, bin them in 12bit, then change the bin result into 16 bit.

Second, change them into 16bit, then bin them.

For method 1: (4096+4096+4096+4096)/4=4096, then move 4 bit left, 4096*16=65536

For method 2: move 4 bit left first: (4096*16 + 4096*16 + 4096*16 + 4096*16)/4 = 65536

For bin, usually, it is sum and multiply, but in some special case, maybe other calculations. But it usually do not affect the ADC.

Thanks

Chad

james_screech

chad wrote:

Hi，

Bin has nothing to do with ADC.

For example, 12 bit ADC, for a pixel, when it is over exposure, it will output 2^12=4096. I think there is no doubt about it. But in computer system, the length of data usually is 8bit, 16bit, 32bit or 64bit. It is not 12bit, so we made a 4bit left move. It means we've magnified the data 16 times. then 4096×16=65536.

Now we have 4 pixels to make bin. Each of them is over exposure. So every one will ouput 4096, it is 12 bit. We have 2 ways to bin.

First, bin them in 12bit, then change the bin result into 16 bit.

Second, change them into 16bit, then bin them.

For method 1: (4096+4096+4096+4096)/4=4096, then move 4 bit left, 4096*16=65536

For method 2: move 4 bit left first: (4096*16 + 4096*16 + 4096*16 + 4096*16)/4 = 65536

For bin, usually, it is sum and multiply, but in some special case, maybe other calculations. But it usually do not affect the ADC.

Thanks

Chad

We're going around in circles here, so which way does the ASCOM driver use? If it IS USEALLY sum and multiply but in special cases multiply, sum and divide, what are the special cases?

James

chad

Hi，

Usually, hardware bin, we will use the method 1 to make the binning.

And Software bin will use the method 2 to make binning.

Special case, software bin, in the past, some of our customers said the bin2 it not obvious compared with no binning, it leads them make some mistakes, so we just make sum without multiply when binning at that time.

Thanks

Chad