The 80/1.7 GF on the GFX 100S has *more* DOF than calculators say

Started 2 months ago | Discussions
JimKasson
MOD JimKasson Forum Pro • Posts: 33,941
The 80/1.7 GF on the GFX 100S has *more* DOF than calculators say
4

Not much more, but more, at least at f/2 and subject distances of 101 and 55 meters:

https://blog.kasson.com/the-last-word/are-dof-calculators-useful/

And the DOF is slightly asymmetrical, to boot.

Jim

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
Rodenmg
Rodenmg Regular Member • Posts: 384
Re: The 80/1.7 GF on the GFX 100S has *more* DOF than calculators say
2

Lol  Not biting today  

 Rodenmg's gear list:Rodenmg's gear list
Nikon D610 Nikon D810 Nikon D500 Sony a6300 Nikon Z7 +25 more
MOD Hamiltionian Contributing Member • Posts: 929
Re: The 80/1.7 GF on the GFX 100S has *more* DOF than calculators say

JimKasson wrote:

Not much more, but more, at least at f/2 and subject distances of 101 and 55 meters:

https://blog.kasson.com/the-last-word/are-dof-calculators-useful/

And the DOF is slightly asymmetrical, to boot.

Jim

"""If we focus on the front tree and look at the back tree, we should see the same amount of blur that we see if we focus on the back tree and look at the front tree.''''

True as long as f << S1, where S1 is the distance to the focal plane in the objects space, and definitely true in this case.

Spherical aberration is much more likely the source of the asymmetry.  Especially since this is designed as a portrait lens and such an asymmetry is a feature.

-- hide signature --

Stay Calm and Carry Cameras
https://thecentercolumn.com/

JimKasson
OP MOD JimKasson Forum Pro • Posts: 33,941
Re: The 80/1.7 GF on the GFX 100S has *more* DOF than calculators say

Hamiltionian wrote:

JimKasson wrote:

Not much more, but more, at least at f/2 and subject distances of 101 and 55 meters:

https://blog.kasson.com/the-last-word/are-dof-calculators-useful/

And the DOF is slightly asymmetrical, to boot.

Jim

"""If we focus on the front tree and look at the back tree, we should see the same amount of blur that we see if we focus on the back tree and look at the front tree.''''

True as long as f << S1, where S1 is the distance to the focal plane in the objects space, and definitely true in this case.

Right.

Spherical aberration is much more likely the source of the asymmetry. Especially since this is designed as a portrait lens and such an asymmetry is a feature.

I would agree that SA is the most likely suspect. What is interesting to me is that the asymmetry is slight, at least at these largish blur circles.

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
MOD Hamiltionian Contributing Member • Posts: 929
Re: The 80/1.7 GF on the GFX 100S has *more* DOF than calculators say

JimKasson wrote:

Hamiltionian wrote:

JimKasson wrote:

Not much more, but more, at least at f/2 and subject distances of 101 and 55 meters:

https://blog.kasson.com/the-last-word/are-dof-calculators-useful/

And the DOF is slightly asymmetrical, to boot.

Jim

"""If we focus on the front tree and look at the back tree, we should see the same amount of blur that we see if we focus on the back tree and look at the front tree.''''

True as long as f << S1, where S1 is the distance to the focal plane in the objects space, and definitely true in this case.

Right.

Spherical aberration is much more likely the source of the asymmetry. Especially since this is designed as a portrait lens and such an asymmetry is a feature.

I would agree that SA is the most likely suspect. What is interesting to me is that the asymmetry is slight, at least at these largish blur circles.

Yeah, does seem pretty slight.  I suppose you can only have some much SA and still meet the expectations for sharpness and contrast for a modern MF lens.

-- hide signature --

Stay Calm and Carry Cameras
https://thecentercolumn.com/

Erik Kaffehr
Erik Kaffehr Veteran Member • Posts: 5,844
Re: The 80/1.7 GF on the GFX 100S has *more* DOF than calculators say

JimKasson wrote:

Hamiltionian wrote:

JimKasson wrote:

Not much more, but more, at least at f/2 and subject distances of 101 and 55 meters:

https://blog.kasson.com/the-last-word/are-dof-calculators-useful/

And the DOF is slightly asymmetrical, to boot.

Jim

"""If we focus on the front tree and look at the back tree, we should see the same amount of blur that we see if we focus on the back tree and look at the front tree.''''

True as long as f << S1, where S1 is the distance to the focal plane in the objects space, and definitely true in this case.

Right.

Spherical aberration is much more likely the source of the asymmetry. Especially since this is designed as a portrait lens and such an asymmetry is a feature.

I would agree that SA is the most likely suspect. What is interesting to me is that the asymmetry is slight, at least at these largish blur circles.

I would agree on SA, but must say, I don't have any ponies in this race.

I would think that the DoF concept is good for understanding the concept of in and out of focus rendition. Also, I would suggest that it is good that DoF-tools are decent predictor for transition between in focus and out focus.

But, all that stuff has been known for like 100 years, so it doesn't come as a great surprise.
I would say that DoF calculators may be useful, if we have the knowledge to use them correctly.

Best regards

Erik

-- hide signature --

Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles

JimKasson
OP MOD JimKasson Forum Pro • Posts: 33,941
At f/2.8...

...DOF is still greater than the calculators say it should be, but it is more symmetric.

https://blog.kasson.com/gfx-100s/are-dof-calculators-useful-part-2/

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
MOD Hamiltionian Contributing Member • Posts: 929
Re: At f/2.8...

JimKasson wrote:

...DOF is still greater than the calculators say it should be, but it is more symmetric.

https://blog.kasson.com/gfx-100s/are-dof-calculators-useful-part-2/

Interesting.  Makes sense that we are seeing more symmetry, but I am still somewhat puzzled that we are seeing less dof than the calculators predict.  At the narrower aperture I would have expected it to be pretty close.

I didn't see how the images were processed.  Sharpening?  Something funny going on in the demosaicing?

-- hide signature --

Stay Calm and Carry Cameras
https://thecentercolumn.com/

JimKasson
OP MOD JimKasson Forum Pro • Posts: 33,941
Re: At f/2.8...

Hamiltionian wrote:

JimKasson wrote:

...DOF is still greater than the calculators say it should be, but it is more symmetric.

https://blog.kasson.com/gfx-100s/are-dof-calculators-useful-part-2/

Interesting. Makes sense that we are seeing more symmetry, but I am still somewhat puzzled that we are seeing less dof than the calculators predict. At the narrower aperture I would have expected it to be pretty close.

I didn't see how the images were processed. Sharpening? Something funny going on in the demosaicing?

Lr with defaults except for Sharpness amount 20, radius 1, detail 0, and WB Cloudy.

But whatever the processing was affected all the images, since I used the f/4 shots for the series as the starting place for the Matlab processing.

Of course, I didn't have image of what the lens was putting down on the sensor at 600 nm sampling resolution. I just had the image captured at 3.76 um sampling resolution. I did enlarge that to 900% before doing the CoC simulation, and I used bilinear interpolation for that enlargement. Maybe I lost some sharpness there.

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
JimKasson
OP MOD JimKasson Forum Pro • Posts: 33,941
Bokeh simulation
2

I realized that, with a couple of mods, the code I wrote for the CoC processing could do bokeh simulation.

Anybody for bubble bokeh?

Jim

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
Rodenmg
Rodenmg Regular Member • Posts: 384
Re: Bokeh simulation

That rates right up there with finger nails on a chalk board. 😳

 Rodenmg's gear list:Rodenmg's gear list
Nikon D610 Nikon D810 Nikon D500 Sony a6300 Nikon Z7 +25 more
JimKasson
OP MOD JimKasson Forum Pro • Posts: 33,941
Re: Bokeh simulation

JimKasson wrote:

I realized that, with a couple of mods, the code I wrote for the CoC processing could do bokeh simulation.

Anybody for bubble bokeh?

Some people like it.

https://fujilove.com/the-legend-and-the-bubble-bokeh-review-of-the-meyer-optik-goerlitz-trioplan-100mm-f2-8/

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
Rodenmg
Rodenmg Regular Member • Posts: 384
Re: Bokeh simulation

Onion rings?  To each his own.

Brrrr, like wearing an army issue wool sweater without a t-shirt  

 Rodenmg's gear list:Rodenmg's gear list
Nikon D610 Nikon D810 Nikon D500 Sony a6300 Nikon Z7 +25 more
R Holt New Member • Posts: 24
Re: The 80/1.7 GF on the GFX 100S has *more* DOF than calculators say

I would like to have seen an image taken with the settings shown in this screenshot from dofmaster. Such an image, if the foreground and background trees are sharp when displayed at 100% magnification, would be convincing - depth of field calculators are useful. It is important to choose an appropriate circle of confusion and this is where some discovery may be required. I have chosen 2.5 * pixel pitch, but there is no consensus as to the right size.

JimKasson
OP MOD JimKasson Forum Pro • Posts: 33,941
Re: The 80/1.7 GF on the GFX 100S has *more* DOF than calculators say

R Holt wrote:

I would like to have seen an image taken with the settings shown in this screenshot from dofmaster. Such an image, if the foreground and background trees are sharp when displayed at 100% magnification, would be convincing - depth of field calculators are useful. It is important to choose an appropriate circle of confusion and this is where some discovery may be required. I have chosen 2.5 * pixel pitch, but there is no consensus as to the right size.

For the right price, I could do that.

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
R Holt New Member • Posts: 24
Re: The 80/1.7 GF on the GFX 100S has *more* DOF than calculators say

Haha, you don't have to convince me. For all the nonbelievers it wouldn't take much to focus on the background in live view at high magnification and stop down until the foreground appears sharp. Then, check the depth of field calculator.

MOD Hamiltionian Contributing Member • Posts: 929
Re: At f/2.8...

JimKasson wrote:

Hamiltionian wrote:

JimKasson wrote:

...DOF is still greater than the calculators say it should be, but it is more symmetric.

https://blog.kasson.com/gfx-100s/are-dof-calculators-useful-part-2/

Interesting. Makes sense that we are seeing more symmetry, but I am still somewhat puzzled that we are seeing less dof than the calculators predict. At the narrower aperture I would have expected it to be pretty close.

I didn't see how the images were processed. Sharpening? Something funny going on in the demosaicing?

Lr with defaults except for Sharpness amount 20, radius 1, detail 0, and WB Cloudy.

But whatever the processing was affected all the images, since I used the f/4 shots for the series as the starting place for the Matlab processing.

My instinct is to suggest that the order of the blur and sharpening operations could matter, but after playing around in photoshop I can't demonstrate that in any noticeable way.  Sharpening unlikely to be the issue here.

Of course, I didn't have image of what the lens was putting down on the sensor at 600 nm sampling resolution. I just had the image captured at 3.76 um sampling resolution. I did enlarge that to 900% before doing the CoC simulation, and I used bilinear interpolation for that enlargement. Maybe I lost some sharpness there.

Maybe, seems pretty reasonable to me though and unlikely to cause a significant loss in image sharpness.

-- hide signature --

Stay Calm and Carry Cameras
https://thecentercolumn.com/

JimKasson
OP MOD JimKasson Forum Pro • Posts: 33,941
Re: At f/2.8...

Hamiltionian wrote:

JimKasson wrote:

Hamiltionian wrote:

JimKasson wrote:

...DOF is still greater than the calculators say it should be, but it is more symmetric.

https://blog.kasson.com/gfx-100s/are-dof-calculators-useful-part-2/

Interesting. Makes sense that we are seeing more symmetry, but I am still somewhat puzzled that we are seeing less dof than the calculators predict. At the narrower aperture I would have expected it to be pretty close.

I didn't see how the images were processed. Sharpening? Something funny going on in the demosaicing?

Lr with defaults except for Sharpness amount 20, radius 1, detail 0, and WB Cloudy.

But whatever the processing was affected all the images, since I used the f/4 shots for the series as the starting place for the Matlab processing.

My instinct is to suggest that the order of the blur and sharpening operations could matter, but after playing around in photoshop I can't demonstrate that in any noticeable way. Sharpening unlikely to be the issue here.

The order in not important in linear convolution, just like in multiplication. In fact, convolution in the space domain is equivalent to multiplication in the frequency domain. However, the demosaicing process as performed by Lr is nonlinear. I’ll do some tests.

Of course, I didn't have image of what the lens was putting down on the sensor at 600 nm sampling resolution. I just had the image captured at 3.76 um sampling resolution. I did enlarge that to 900% before doing the CoC simulation, and I used bilinear interpolation for that enlargement. Maybe I lost some sharpness there.

Maybe, seems pretty reasonable to me though and unlikely to cause a significant loss in image sharpness.

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
MOD Hamiltionian Contributing Member • Posts: 929
Re: At f/2.8...
1

My instinct is to suggest that the order of the blur and sharpening operations could matter, but after playing around in photoshop I can't demonstrate that in any noticeable way. Sharpening unlikely to be the issue here.

The order in not important in linear convolution, just like in multiplication. In fact, convolution in the space domain is equivalent to multiplication in the frequency domain. However, the demosaicing process as performed by Lr is nonlinear. I’ll do some tests.

Gotcha thanks. Haven't spent time on this before, which now seems an oversight. It wasn't immediately obvious to me that sharpening was a linear convolution. And then yes, linear convolutions commute. For others following along:

https://en.wikipedia.org/wiki/Kernel_(image_processing)

https://en.wikipedia.org/wiki/Convolution

Of course, I didn't have image of what the lens was putting down on the sensor at 600 nm sampling resolution. I just had the image captured at 3.76 um sampling resolution. I did enlarge that to 900% before doing the CoC simulation, and I used bilinear interpolation for that enlargement. Maybe I lost some sharpness there.

Maybe, seems pretty reasonable to me though and unlikely to cause a significant loss in image sharpness.

-- hide signature --

Stay Calm and Carry Cameras
https://thecentercolumn.com/

JimKasson
OP MOD JimKasson Forum Pro • Posts: 33,941
Round trip test
2

JimKasson wrote:

The order in not important in linear convolution, just like in multiplication. In fact, convolution in the space domain is equivalent to multiplication in the frequency domain. However, the demosaicing process as performed by Lr is nonlinear. I’ll do some tests.

I've rejiggered the code and run a round trip test. If you set the CoC very small, the output image is the same as the input image.

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
Keyboard shortcuts:
FForum MMy threads