Pixel binning - huh?

[SrMi]

Advertisement (gone after registration)

6 hours ago, thrid said:

When they pixel bin are they reading the receptors at different gain values?

I'm not sure if this is possible but one idea would be to read a smaller set of receptors to favor highlight detail and combine the two readings. Kind of like a bracketed gain reading.

Fuji did something like this with their SuperCCD, except they had an additional set of dedicated receptors used to capture highlight information. This information was combined with the main receptors and resulted in significantly more highlight retention

Arri one ups this by taking two simultaneous bracketed gain readings at each receptor well, resulting in two exposure readings that are combined on the fly and results in an obscene exposure range. The new Alexa S35 captures a true, fully useable 17 stops of range.

Leica cameras do not pixel bin.

When using quad bayer sensors, the camera could generate real-time HDR by having the upper row use different exposure than the lower row of pixels. I do not know if any camera or smartphone is using it.

 

[SrMi]

7 hours ago, thrid said:

When they pixel bin are they reading the receptors at different gain values?

I'm not sure if this is possible but one idea would be to read a smaller set of receptors to favor highlight detail and combine the two readings. Kind of like a bracketed gain reading.

Fuji did something like this with their SuperCCD, except they had an additional set of dedicated receptors used to capture highlight information. This information was combined with the main receptors and resulted in significantly more highlight retention

Arri one ups this by taking two simultaneous bracketed gain readings at each receptor well, resulting in two exposure readings that are combined on the fly and results in an obscene exposure range. The new Alexa S35 captures a true, fully useable 17 stops of range.

You cannot compare DRs until measured the same way and analyzed whether NR is being applied in the camera.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[01af]

Reducing the resolution—be it via pixel binning, in-camera downscaling, downsizing in post-processing, or using a sensor with less megapixels to begin with—won't help with noise, sharpness, exposure range, or camera shake. It just reduces image size. In a smaller image, the same amount of noise or blur simply will be less obvious to the beholder ... which often gets confused with an actual reduction of noise or blur.

So switching the Leica M11's sensor resolution from 60 MP down to 36 or 18 MP won't improve anything in any way. It will just reduce memory consumption.

[Jared]

12 hours ago, 01af said:

Reducing the resolution—be it via pixel binning, in-camera downscaling, downsizing in post-processing, or using a sensor with less megapixels to begin with—won't help with noise, sharpness, exposure range, or camera shake. It just reduces image size. In a smaller image, the same amount of noise or blur simply will be less obvious to the beholder ... which often gets confused with an actual reduction of noise or blur.

So switching the Leica M11's sensor resolution from 60 MP down to 36 or 18 MP won't improve anything in any way. It will just reduce memory consumption.

I agree it won’t help with sharpness, exposure range, or camera shake, but it DOES help with noise. Shot noise is equal to the square root of the signal. If you bin/down sample/use larger pixels, then SNR improves because per pixel signal increases, and noise goes up slower than signal with larger pixels (however you achieved the larger pixels). That’s not just a reduction in memory consumption. It’s an actual, demonstrable, easily measured improvement in one aspect of technical image quality. It comes at the cost of resolution, of course. 

[thrid]

On 10/21/2022 at 6:51 PM, mirekti said:

There's a reason why Arri costs as much as it does.

Well, the Alexa is a low volume highly specialized piece of kit, but the cost is not just due to the dual gain sensor design... But it would be very interesting if Leica made a lower resolution high dynamic range version of the M that utilized that kind of sensor technology.

 

[thrid]

20 hours ago, SrMi said:

Leica cameras do not pixel bin.

When using quad bayer sensors, the camera could generate real-time HDR by having the upper row use different exposure than the lower row of pixels. I do not know if any camera or smartphone is using it.

It would be interesting if Leica could do something like that, but obviously it would take good amount of new code. If it's even possible with the existing sensor. I assume you would also need a good amount of processing power in the camera. 

Edited October 23, 2022 by thrid

[SrMi]

Advertisement (gone after registration)

2 hours ago, thrid said:

It would be interesting if Leica could do something like that, but obviously it would take good amount of new code. If it's even possible with the existing sensor. I assume you would also need a good amount of processing power in the camera. 

Leica could not do that because it does not have a Quad Bayer sensor that could support that type of operation. It is a matter of hardware support.

[thrid]

2 minutes ago, SrMi said:

Leica could not do that because it does not have a Quad Bayer sensor that could support that type of operation. It is a matter of hardware support.

Got it. Thanks for the clarification.

 

[01af]

vor 21 Stunden schrieb Jared:

... but it DOES help with noise.

No, it doesn't.  Just open your eyes and take a look. Noise in a high-ISO 18 MP Leica M11 picture is no lower than noise in a same-ISO 60 MP Leica M11 picture of the very same subject.

.

vor 21 Stunden schrieb Jared:

If you bin, down-sample, or use larger pixels, then SNR will improve because per-pixel signal increases, and noise goes up slower than signal with larger pixels (however you achieved the larger pixels). 

No-one cares about pixels. Your problem is, you refuse to see the forest (= picture) for all the trees (= pixels). Look at pictures, not pixels! Sure—larger pixels do have better signal-to-noise ratio than smaller pixels. But you keep ignoring the fact that many pixels have better signal-to-noise ratio than few pixels. In the end, it's the size of the sensor, not the pixel size, that determines the picture's signal-to-noise ratio, Everything else being equal, bigger sensors exhibit less noise—no matter if their pixels are bigger or more numerous than the smaller sensor's. That's why bigger-format cameras generally yield better image quality than smaller-format cameras ... no matter if it's film or digital.

Edited October 23, 2022 by 01af

[Santos]

On 10/18/2022 at 3:53 PM, THEME said:

Fascinating post, more please!

So for what circumstances / shooting task / environments you use M11 lower res, @Santos?

Dear Theme,

I almost always have my M11 on the 18MP mode. For Instance, Street, Landscape, Lifestyle, and Model Shoots. The only Reason I ever Change to 60 is when I know I will have to Crop. This may not work for Everyone's Workflow of course, For Instance I will Include an Example using the 135mm APO when I knew I would crop hence using 60mp and One using the Same lens on 18mp. My edits are simple as I under expose and adjust the Black, White, Vibrance Sliders, and White Balance. The Oryges are cropped in hence I used 60mp, while the Maha (a Type of Dear in Arabic) is shot using 18mp Mode and only edited. However, I Plan to Crop the Maha Photo to 4:5 Horizontal since I find it more pleasing (for My Eyes at  least).

Apologies for late Reply I was not Active for a while.

Sincerely,

Santos.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Edited November 10, 2022 by Santos
Added Text