CMOS Hardware Binning

[baj (Brad)]

I recently acquired an ASI533M and I was surprised to see a hardware binning option in sharpcap.



In the past I have used software binning which improved my SNR but had unusual side effects on my image sampling / scale in terms of astrometric measurements so I was curious to see how this hardware binning feature worked.



It looks to me like the hardware binning 2x2 clearly changed my image sampling so that I was undersampled. It also worked well in my astrometric software. If this is the case then for me this is great news since I only image asteroids and comets whose motion is not compensated by sidereal tracking. I rely on undersampling to minimize trailing in my images whilst keeping round stars.



Has anyone been experimenting with hardware binning?



If it works the way I think it does then I would love to hear from anyone with an IMX571 sensor such as the ASI2600. For me this camera in 3x3 hardware bin mode would be perfect for a 12 inch F3 newtonian. I would sell my EdgeHD11 and Hyperstar to get this camera if it could give pixels that were effectively 11.28 microns


Brad.

[baj (Brad)]

Thinking about this. I wonder if the following applies

I think hardware binning reduces the resolution at the sensor level so that the amount of data being read from the sensor is smaller. A 1 micron 20 megapixel sensor would become a 2 micron 10 megapixel sensor and thus reduce the file size and amount of data to be read. Along the lines of ccd binning.

I assume that with software binning you will always read the full resolution from the sensor and then after that the software bins the data. In the above example if I software bin 2x2 it will still be necessary to read 20 megapixels before it is then binned by the software

So maybe it is just about reducing the amount of data being read from the sensor. Although I suspect the software binning would not change the well depth of the pixels if they are binned after the image is captured and not before

[Drac0 (Mark)]

Hi Brad,

CMOS cameras like the 533 don't actually do hardware binning, it's just in camera software binning. Whatever binning you use, the sensor collects the same data but it does reduce the data transfer from the camera by binning it first. It makes a huge difference in file size - a bin 2 file will be half the resolution but a quarter of the file size. My 294MM produces a 8288×5644, ~90mb file in "unlocked" bin 1 mode while bin 2 it's 4144x2822 & ~22mb.

I'm still investigating whether there is any advantage at all to using bin 1 and binning it in post or just using bin 2 straight from the camera.

A 1 micron 20 megapixel sensor would become a 2 micron 10 megapixel sensor and thus reduce the file size and amount of data to be read.

In this case it would become a 2 micron 5 megapixel, half the resolution but 1/4 the megapixels.

Cheers,
Mark

[baj (Brad)]

Thanks Mark


It makes me wonder though why there are two options. One for hardware binning and one for software binning. In addition software binning goes to 4x4 but hardware binning stops at 3x3. I would assume there would have to be a difference between the two



I have also discovered that software binning causes problems when using my astrometric software to record positions of asteroids but the hardware binning option worked fine. So another difference.

[baj (Brad)]

Ok. Just checked the manual for the camera which states that hardware binning provides faster frame rates. This is the only difference zwo identify between software binning and hardware binning.



What does this mean? Maybe each frame is smaller in size due to increased pixel size from binning and therefore faster to read. 1 micron 20 megapixel frame down to 2 micron 5 megapixel frame (thanks Mark).


I would love it if it worked this way which is similar to ccd binning.

[Drac0 (Mark)]

Hi Brad,

I actually stand corrected, CMOS do have a form of hardware binning, but it's not the same as CCD & is limited in access.

After comparing CMOS BIN and CCD BIN, let’s look at another concept of BIN- hardware BIN and software BIN. Hardware BIN, as the name implies, is to merge the pixel with hardware, usually refers to the completion of pixel merge on the chip, due to the different chip structure, CCD can do this, but CMOS cannot. CMOS hardware BIN is more like pixel skip. The frame rate will be faster by pixel extraction, but the signal-to-noise ratio is limited. Generally, we apply it to scenes that require high frame rates, such as shooting solar system objects. For deep space shooting, we recommend binning with software on CMOS cameras.
Here we also defines hardware BIN as pixel merge on CMOS sensor, and software BIN as pixel merge in SDK in software. Please notice it is not the same BIN as CCD on chip BIN.

In our ASCOM and ASIImg, ASILive software, the default BIN is software BIN, only ASICAP have hardware BIN option. You can find this option in ASICAP->Control->More (provided that the camera supports hardware BIN).

Source: https://astronomy-imaging-camera.com...-fundamentals/

Still learning,
Mark

[ChrisV (Chris)]

I thought that binning has the advantage of reducing read noise on CCD but not CMOS cameras. On CCD, read noise for one pixel is the same for binned pixels. So if binning 2x2 you effective get 1/4 of the read noise . But this advantage doesn't happen with CMOS.

Yes?

[baj (Brad)]

That is a great link Mark.



I am fascinated by the data comparison between the ASI1600 VS KAF8300. According to this table in software bin 2 mode your pixel size doubles and your full well increases by a factor of 4. I did not expect this. This being the case if you are simply after larger pixels and an increase in full well depth as I am then software bin mode is fine


I am going to experiment with software binning a lot more

[Drac0 (Mark)]

Originally Posted by baj

That is a great link Mark.



I am fascinated by the data comparison between the ASI1600 VS KAF8300. According to this table in software bin 2 mode your pixel size doubles and your full well increases by a factor of 4. I did not expect this. This being the case if you are simply after larger pixels and an increase in full well depth as I am then software bin mode is fine


I am going to experiment with software binning a lot more

Sounds about right - in bin 1 on my 294MM the full well is only 14k while it's 66k in bin 2 mode along with the doubled pixel size. This is part of the quandary I have at the moment - at a very short focal length (<300mm), bin 1 is considered a better "fit", but is there enough visible improvement in the results to make it worth giving away the advantages of bin 2? And how much difference, if any, is there between software binning in camera & capturing in bin 1 then doing bin 2 in post?

I will probably just end up going bin2 in camera until I get a better pc for post, one that can handle the larger files much easier.

Cheers,
Mark