Does pixel pitch affect depth of field

[OpticsEngineer]

Thanks. To simulate the fabrics I might just stick with line-pairs as they also generate color fringes. But the concentric circles are a really good addition since the moire patterns really leap out.

 

[Jack Hogan]

Thanks. To simulate the fabrics I might just stick with line-pairs as they also generate color fringes. But the concentric circles are a really good addition since the moire patterns really leap out.

If you can, stick to a sinusoidal target instead of line pairs, it makes Michelson Contrast immediately useful. Lines introduce harmonics, which complicate the readings and related calculations.

https://www.strollswithmydog.com/resolution/#SineTarget

I think it's a good idea to use some MTF-derived metric for acceptable sharpness, in fact it's pretty well a standard in many disciplines. MTF50 is a decent first approximation when pixel peeping in photography, per my previous link.

However, the desired application needs to be kept in front of mind. For instance MTF50 is very sensitive compared to e.g. CPIQ acutance (see the first plot I showed upthread), it takes a trained eye to see an MTF50 difference of 10%. And noise around the relative spatial frequencies in photographic MTFs show as undulations of the curve, which makes MTF50 more sensitive to noise. Machine vision may be more or less exacting than photography.

To John and Bob's point, in photographic DOF discussions one cannot discount the human viewer when determining the threshold for acceptability. In the limit the CoC is the smallest visual angle we can detect. You can say that the capture/display equipment set some sort of a lower limit on it - but in the end it is a combination (a convolution) of all of the above. Marc Levoy uses 1 pixel CoCs to calculate the DOF of projected images for instance.

Horses for courses I guess.

Jack

PS. The mentioned lens that reaches its best 'sharpness' around f/8 is a Nikon AF-S 70-300mm f/4.5-5.6 VR, near the long end. It 'is the very best modern tele zoom lens to get if you prefer a dedicated tele zoom over a do-it-all zoom and want light weight and moderate price'. Many happy campers use it around these fora.

 

[prsc]

"How are you defining acceptable sharp focus?"

For the purposes of how I am approaching this, whatever sharpness there is at best focus I would define as acceptably sharp. So that is our reference level of contrast.

Then we move away from best focus and we reach points before and behind where the contrast has been reduced to 50% of the reference contrast.

Fair enough. But is it then the pixel size or the changing reference sharpness that affects the DoF? To me it seems like it is more about the latter.

 

[JohnMoyer]

...

However, once one starts playing with this stuff one soon realize that we often err on the wrong side of aperture size, tending to prefer larger f-numbers for safety, therefore dropping sharpness at the scene. Never, ever f/16 for critical shots with today's equipment (ok, maybe just once or twice for special occasions).

Jack

For me, "special occasions" happen more than once or twice.

Maybe this was a "special occasion" because subject movement prevented focus stacking and subject would leave if photographer got too close. It seems to me that in some circumstances F/16 might be the compromise that captures the most data.


Gulf fritillary (Agraulis vanillae) on Zinnia in Norman, Oklahoma, United States on September 15, 2024 ; Canon EOS R5 ; EF100-400mm f/4.5-5.6L IS II USM +2x III ; F Number 16.0 ; 800.0 mm

Using the circle of confusion reported by the camera in the metadata, Focus Distance Upper2.4 m, Focus Distance Lower2.2 m. But for me the depth of field seems much shallower because I view the image on a screen at 100%.

This was processed using the Canon "digital lens optimizer", but I expect that Richardson/Lucy capture sharpening in rawtherapee would produce similar results.

All of the edits done in Canon DPP software:

• RawBrightnessAdj : 0.67
• WhiteBalanceAdj : Daylight
• WBAdjColorTemp : 5200
• ColorSaturationAdj : 1
• UnsharpMaskStrength : 2
• UnsharpMaskFineness : 1.7
• UnsharpMaskThreshold : 3
• LuminanceNoiseReduction : 1
• ChrominanceNoiseReduction : 1
• ColorMoireReductionOn : No
• ColorBlurOn : Yes
• DLOSetting : 75
• GammaWhitePoint : +1.300
• CropX : 1592
• CropY : 1064
• CropWidth : 6600
• CropHeight : 4400

Shot with:

• ExposureCompensation : -2/3
• Quality : CRAW
• CameraISO : Auto
• MeteringMode : Evaluative
• MeasuredRGGB : 1156 1024 1024 277
• WB_RGGBLevelsAsShot : 1967 1024 1024 2123
• ColorTempAsShot : 4803
• WhiteBalance : Auto
• MeasuredEV : 14.00
• MeasuredEV2 : 29

CircleOfConfusion : 0.030 mm calculated by exiftool

DOF : 0.005 m (2.297 - 2.303 m) calculated by exiftool

Exposure compensation was to avoid clipping reds in raw. Post processing increase in "GammaWhitePoint" was to use all of the red data and avoid clipping reds in JPEG.

( Downscaled image is at https://www.rsok.com/~jrm/2024Sep30_birds_and_cats/2024sep15_butterfly_IMG_1542c.html , GraphicsMagick did an additional unsharp mask after scaling with the default algorithm )

--

John Moyer's Home Page

John Moyer home page with prime numbers.

www.rsok.com

John Moyer

 

[Jack Hogan]

"How are you defining acceptable sharp focus?"

For the purposes of how I am approaching this, whatever sharpness there is at best focus I would define as acceptably sharp. So that is our reference level of contrast.

Then we move away from best focus and we reach points before and behind where the contrast has been reduced to 50% of the reference contrast.

Fair enough. But is it then the pixel size or the changing reference sharpness that affects the DoF? To me it seems like it is more about the latter.

DOF is a region of acceptable perceived sharpness. If none of the image falls into that category, there is no DOF.

 

[Jack Hogan]

However, once one starts playing with this stuff one soon realize that we often err on the wrong side of aperture size, tending to prefer larger f-numbers for safety, therefore dropping sharpness at the scene. Never, ever f/16 for critical shots with today's equipment (ok, maybe just once or twice for special occasions).

For me, "special occasions" happen more than once or twice.

Point well taken John. My comments to Aaron are to be taken in the context of his article, large prints of landscapes.

Jack

 

[Garry2306]

...

However, once one starts playing with this stuff one soon realize that we often err on the wrong side of aperture size, tending to prefer larger f-numbers for safety, therefore dropping sharpness at the scene. Never, ever f/16 for critical shots with today's equipment (ok, maybe just once or twice for special occasions).

Jack

For me, "special occasions" happen more than once or twice.

Maybe this was a "special occasion" because subject movement prevented focus stacking and subject would leave if photographer got too close. It seems to me that in some circumstances F/16 might be the compromise that captures the most data.


Gulf fritillary (Agraulis vanillae) on Zinnia in Norman, Oklahoma, United States on September 15, 2024 ; Canon EOS R5 ; EF100-400mm f/4.5-5.6L IS II USM +2x III ; F Number 16.0 ; 800.0 mm

Using the circle of confusion reported by the camera in the metadata, Focus Distance Upper2.4 m, Focus Distance Lower2.2 m. But for me the depth of field seems much shallower because I view the image on a screen at 100%.

This was processed using the Canon "digital lens optimizer", but I expect that Richardson/Lucy capture sharpening in rawtherapee would produce similar results.

All of the edits done in Canon DPP software:

• RawBrightnessAdj : 0.67
• WhiteBalanceAdj : Daylight
• WBAdjColorTemp : 5200
• ColorSaturationAdj : 1
• UnsharpMaskStrength : 2
• UnsharpMaskFineness : 1.7
• UnsharpMaskThreshold : 3
• LuminanceNoiseReduction : 1
• ChrominanceNoiseReduction : 1
• ColorMoireReductionOn : No
• ColorBlurOn : Yes
• DLOSetting : 75
• GammaWhitePoint : +1.300
• CropX : 1592
• CropY : 1064
• CropWidth : 6600
• CropHeight : 4400

Shot with:

• ExposureCompensation : -2/3
• Quality : CRAW
• CameraISO : Auto
• MeteringMode : Evaluative
• MeasuredRGGB : 1156 1024 1024 277
• WB_RGGBLevelsAsShot : 1967 1024 1024 2123
• ColorTempAsShot : 4803
• WhiteBalance : Auto
• MeasuredEV : 14.00
• MeasuredEV2 : 29

CircleOfConfusion : 0.030 mm calculated by exiftool

DOF : 0.005 m (2.297 - 2.303 m) calculated by exiftool

Exposure compensation was to avoid clipping reds in raw. Post processing increase in "GammaWhitePoint" was to use all of the red data and avoid clipping reds in JPEG.

( Downscaled image is at https://www.rsok.com/~jrm/2024Sep30_birds_and_cats/2024sep15_butterfly_IMG_1542c.html , GraphicsMagick did an additional unsharp mask after scaling with the default algorithm )

The Cannon upper and lower focus distances are not the far and near depths of field. Although these EXIF tags are proprietary to Canon, I believe, they give you a range for the focus distance, eg https://photography.grayheron.net/2021/08/m3-landscape-bracketing-script-where.html

 

[JohnMoyer]

However, once one starts playing with this stuff one soon realize that we often err on the wrong side of aperture size, tending to prefer larger f-numbers for safety, therefore dropping sharpness at the scene. Never, ever f/16 for critical shots with today's equipment (ok, maybe just once or twice for special occasions).

For me, "special occasions" happen more than once or twice.

Point well taken John. My comments to Aaron are to be taken in the context of his article, large prints of landscapes.

Jack

In that context I agree

 

[DaveR43]

With a different approach, Aaron Priest says his photographer friends ask him why his landscapes are sharper than theirs, such as a 36x24 inch print viewed at ten inches. Hyperfocal Distance, Depth of Field, Circle of Confusion, Diffraction, and Print Size - Aaron Priest Photography (edit: apologies that this link doesn't seem to work anymore) He enters a two-pixel CoC of about 0.009 mm into a DOF calculator while his friends enter “full-frame.” So, Aaron takes photos with a hyperfocal distance three times longer and he captures more detail.

Ll

So far as I can see, none of Aaron Priest's websites are available. The only sites showing his material are managed by other people, and show examples of his panoramic and 360 degree photos, with supporting commentary

 

[OpticsEngineer]

"...is it then the pixel size or the changing reference sharpness that affects the DoF?"

The way I am thinking about this is both are directly linked, but maybe it is most productive to think it is more related to the latter as you suggested.

With the low resolution camera, the achievable sharpness is limited. In the original post, the A7SII camera can't resolve the 14 lp/inch target at all. That pattern is just a blur.

But a higher resolution can resolve more and its reference sharpness is higher. (if we equate sharpness with resolution). So in the original post, the A7RIV can resolve the 14 lp/inch target. And the A7RIV has a shorter depth of field.

Kind of related to this, I have read quite a few posts here on dpreview over in the Canon, Nikon, and Sony forums where people will ask something like, "I have an old lower resolution camera, what should I expect if I buy a newer higher resolution camera?" (usually they mention specific camera models.). And frequently people will answer they made the same upgrade and are really happy with the added detail, but also had to get to used to a much more narrow depth of field. Comments will be like they shot at f/2 like always, and are really happy the eyelashes are sharper than ever before, but now the ear is noticeably less sharp. It seems to surprise a lot of people. Maybe it is because there is so much on the web about how DOF is related to sensor size and pixel size does not matter.