M11 DOF COC

[tashley]

Advertisement (gone after registration)

So. Pixel pitch is 3.76 micrometers. 
 

If you were calculating DOF, given the number of other considerations (output type and size for example) what COC  would you use?

Just curious in others’ opinions. 

[SrMi]

5 minutes ago, tashley said:

So. Pixel pitch is 3.76 micrometers. 
 

If you were calculating DOF, given the number of other considerations (output type and size for example) what COC  would you use?

Just curious in others’ opinions. 

1.5 times the sensor pixel pitch for critical work.

[Advertisement]

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

[adli]

 

CoC is depending on how much you want to enlarge and what viewing distance you anticipate.

 

 

[tashley]

16 minutes ago, adli said:

 

CoC is depending on how much you want to enlarge and what viewing distance you anticipate.

 

 

Of which I am aware (see above). I am interested in what others are thinking. 

[adan]

An associated value would be the diffraction limit for 60 Mpixels.

Based on using 24 Mpixels (M10), the images start to get visibly softer starting between  f/8-11 (75mm APO-Summicron). That's with practical viewing, not "theory." (which may be even more limiting).

CoC and diffraction limit can get you coming and going. Stop down to get the smallest CoC - and you get diffraction blur anyway.

There is also the "quantization" effect of digital imaging.

Even if one assumes a CoC smaller than a pixel (which one would think has to be as sharp as one can image), if that small blur circle falls on the junction of two pixels, it will end up be recorded on both (at 50% "influence" ± ) and thus may appear as a subtle blur. But like "motion blur," it rather depends on just how much a blur is split across two pixels - may be 50/50, may be 95/5.

Related to that, of course, is the slight blurring effect of de-mosaicing a color image - swapping luminance data back and forth between neighboring pixels to produce a "yellow" 1-pixel-wide detail in the final picture, from only RGB pixels on the sensor. You probably need to estimate "blur" will be 1.4x more on a color camera than on a Monochrom, all other calculations being equal.

This site (if you scroll down below the APS-vs-FF DoF argument - been there, done that!) makes some estimates of the MTF between neighboring pixels at various blur sizes (CoC) between 0.5 pixel and 2.3 pixels - animated to show the effect of where the detail falls on the pixel array.

https://epxx.co/artigos/dof_en.html

Short version = a COC of 1.5 pixel widths should still produce MTF 50% even in the worst case (bridging two pixels 50/50) - and better if the split is more uneven. So it is playing the odds.

Whereas even a slight increase from that (CoC = 2 pixel widths) drops the BEST case MTF to 40%, and it goes rapidly downhill from there.

So SrMi's suggestion is on-target. You can probably count on a CoC of less than 5.64 microns imaging as a sharp point - most of the time - on a 60 Mp Monochrom sensor.

Add margins of error on top of that for diffraction and de-bayerizing and print magnification and viewing distance.

Edited January 30, 2022 by adan

[tashley]

Thanks Adan. In a more precise way you’ve got to roughly where I’m at but it’s fascinating seeing the detailed working out of it. Much appreciated!