[Elphel-support] binning modes, sum and average, how to change them?

[Biel Bestué de Luna]

yes I understand this but since I have two front ends one color the other
BW any color degradation might not affect the BW isn't it?

-------------------------------------------

I've been doing some tests tonight, and what I got is the following:

the light increment that you get, while pixel binning (by summing), it's
the same that you get when you're not doing any pixel binning but you
either, double the gain, or double the exposure.

so:

EXPOS = 10000
GAIN = 1
PB = summing

is the same as:

EXPOS = 20000
GAIN = 1
PB = averaging/deactivated

or also the same as:

EXPOS = 10000
GAIN = 2
PB = averaging/deactivated

that's quite cool because if you do use the pixel binning you get a really
great light increment without the increment of noise!





2012/10/13 Oleg <support-list at support.elphel.com>

> Biel,
>
> Do you have a datasheet for the sensor? - it's MT9P001 - if not, get it))
> There's a description on how binning works.
> They mention some legal values for binning and skipping:
> "Only certain combinations for binning and skipping are allowed"
> "Column_Bin=3 -> Legal values for Column Skip: 3 "
>
> Do you set binning (BIN_HOR,BIN_VERT) with the skipping(=decimation)
> (DCM_HOR,DCM_VER) as well?
> Or in you experiments are BIN_HOR=DCM_HOR and BIN_VERT=DCM_VERT?
>
>  1 - why the effect varies in the 1/4th decrement and the 1/8th decrement?
>>       in 1/1th (no binning) there's no light increment (that's logical)
>>       in 1/2th and 1/3th there is one light increment (ok)
>>       in 1/4th there is a super light increment (why?)
>>       between 1/5th and  1/7th there is only one increment (ok)
>>       in 1/8th there is again a super light increment (why?)
>>
>
> I'd say some of the modes didn't work as they should have - due to the
> sensor or the camera software. I haven't tried it.
>
>  2 - this lights increment happen before any "gain" is applied to the
>> sensor signal isn't it?
>
>       this means that this light increment doesn't add noise to the image
>> isn't it?
>>
>
> Yes, the binning should be happening before the analog gain. The pixel
> readout noise goes lower, but the (photon) shot noise is summed or averaged
> depending on the mode.
>
> Note that with binning you get a different type of image degradation
> comparing to the image sensor of the equivalent lower resolution.
>
> Check out this picture<http://community.elphel.com/pictures/binning_diagram.png> -
> an example for 2x binning.
> With binning enabled the effective area of a pixel increases but the
> different color pixels areas interfere with each other. At the same time
> the data is processed like it was obtained from a sensor with lower
> resolution (with a bayer pattern). So, the lower binning coefficient the
> better.
>
> Best regards,
> Oleg Dzhimiev
> Electronics Engineer
> phone: +1 801 783 5555 x124
> Elphel, Inc.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://support.elphel.com/pipermail/support-list_support.elphel.com/attachments/20121014/9217b688/attachment-0002.html>