DoF may not exist...

Started 5 months ago | Discussions
JimKasson
MOD JimKasson Forum Pro • Posts: 36,080
Re: If you don't contribute to a discussion you don't need to participate

JimKasson wrote:

This looks very natural, filmic. But to my eye at 100% magnification, it looks like it was taken on 50mp full frame camera and sized up to 100mp with AI software. I much prefer the eye-watering sharpness of the 50S/R at 100%.

Could be atmospherics or bad PP. But I like that shot.

I don't think there's anything wrong with how sharp that shot is. There's not a lot of important super-high-frequency detail anyway. I'd probably sharpen it a bit before I printed it, depending on how big the print was.

It does need some work with tone curves to get some drama into it. It also looks to me like the horizon is tilted, but I'm not sure about that.

It's a nice shot. Here's a quick cut at an edit, which may give you some ideas.

Greg, if you don't want me to post an edited version of your shot, let me know and I'll take it down.

Here's the recipe:

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
highdesertmesa
highdesertmesa Contributing Member • Posts: 805
Re: If you don't contribute to a discussion you don't need to participate

DMillier wrote:

It's an amazing piece of scenery, like so much in Iceland and this is a decent photo. But how to make a great photo rather just a merely good photo is the question.

I keep staring at this picture and thinking it's marvelous but "how would I make that shot" and I don't really know.There's so much going on, it kind of defies picking a subject and building supporting lines around it. I think it was Michael Reichmann in one of his old video diaries who said something along the lines of "sometimes you just have to stop and and drink it in because there is no way to turn it into the shot you want it to be".

A scene like that is just begging for the GF 250 with and without the 1.4x.

Technically what makes that scene problematic is the main area of the scene is filtered daylight, but there are very bright slices of sun on the far mountains.

Jared Willson Senior Member • Posts: 1,363
Re: Done and dusted

JimKasson wrote:

Greg7579 wrote:

JimKasson wrote:

Greg7579 wrote:

JimKasson wrote:

You asked -- a long time ago, not just today. I answered. I did a fair amount of work to verify my answer (which, by the way, works for the a7RIV, too). I asked you to see if you agree with my answer. You haven't done that. The ball is in your court.

If it works for the alphasevenarefour then I am suspicious because that camera is a smaller format that we used to call Full-Frame that has a lot more DOF than the GFX. 😁

Greg, you are quick to disparage the conclusions about the GFX of those that don't shoot the GFX. Yet, you have the temerity to opine with superlative certainty about the a7RIV, which I believe you have never used. Have you used the a7RIV in secret?

I have both GFX 100x cameras and an a7RIV. When you do the kind of examination of the image at 1:1 that you're talking about, the size of the sensor doesn't matter. In both the case of the a7RIV and thew GFX 100x, the vast majority of the pixels in the image will be off the screen, and won't influence what you're seeing. What will influence what you're seeing as to sharpness vs focal plane distance is the pitch of the sensor -- which is the same for all three cameras, the size of the microlenses -- which is the same for all three cameras, the lens, the aperture, the subject distance, and where the lens is focused.

Take the same lens, set it to an f-stop, focus it at a certain distance, aim it at a subject, take the picture, develop it, and pixel peep at 100% magnification. You'll see the same sharpness vs distance for both the a7RIV and the GFX 100x. So the same CoC diameter is appropriate for prior knowledge of the visual DOF.

Why is this so hard to understand?

Jim

You are right about that alphasevenarefour bit. I haven't shot it. I have shot the Q2 and Canon FF DSLRs for years. And of coursecI shot Fuji X a bazillion times with every lens they made but the 200.

None of those cameras have 3.76 um pitch, right?

Not sure which Fuji X series cameras Greg used in the past, but the current X-T4, X-E4, and X-S10 all have the same 3.76u pixels as in the A7R4 and the GFX 100(s). Different color filter, though… X-Trans rather than Bayer, so there could be slight differences from demosaicing.

 Jared Willson's gear list:Jared Willson's gear list
Leica Q2 Leica SL2 Hasselblad X1D II 50C Leica SL 90-280mm F2.8–4 Hasselblad XCD 30mm F3.5 +8 more
Erik Kaffehr
OP Erik Kaffehr Veteran Member • Posts: 6,291
Another approach...

We can see that mst good lenses reach optimum sharpness at f/4. The only factor reducing sharpness when stopping down is diffraction.

The airy radius is 1.22 * N * Lambda. Assuming Lambda being 0.55 micron, we would get the airy radius where diffraction cames into role to be 1.22 * 4 * 0.55 -> 2.7 microns.

The corresponding diameter is 5.4 microns.

So, it may make some sense to use around 5-6 microns as CoC good lenses at optimal aperture.

But, that would yield a pretty short DoF.

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

Greg7579
Greg7579 Forum Pro • Posts: 14,041
Re: If you don't contribute to a discussion you don't need to participate
1

highdesertmesa wrote:

DMillier wrote:

It's an amazing piece of scenery, like so much in Iceland and this is a decent photo. But how to make a great photo rather just a merely good photo is the question.

I keep staring at this picture and thinking it's marvelous but "how would I make that shot" and I don't really know.There's so much going on, it kind of defies picking a subject and building supporting lines around it. I think it was Michael Reichmann in one of his old video diaries who said something along the lines of "sometimes you just have to stop and and drink it in because there is no way to turn it into the shot you want it to be".

A scene like that is just begging for the GF 250 with and without the 1.4x.

Technically what makes that scene problematic is the main area of the scene is filtered daylight, but there are very bright slices of sun on the far mountains.

I agree!  I was looking at it and wishing I had the 250, which was back in the car a mile away.  That was a two mile walk (one up - one back) to get to that lookout.  I only took the 45-100 on the camera around my neck and a monopod.  But remember, I did very little post on this and there was some haze above that ice.  I have several more that may be better but I just threw this one out to see what Erik had to say about the DOF from front to back that far away.  It all should be in focus but I don't know if it is yet.  Plus, I did shoot some at F8 and even one at F5.6.  I was standing there on a monopod just shooting.  Plus, that shot could have had some camera movement.  The wind was blowing and the monopod helps but is not a tripod.

There are all kinds of variables.

As I have said many times, I shoot at F10 and 11 a lot and I don't really see the diffraction, but I know it is there.  I'm sure Jim and Erik I'm sure would advise that far away shot at F5.6 or 8.  There is a lot of depth (a mile or more) to that scene, but that first ice wall is far away.  The glacier behind it runs a mile up the mountain or more.

-- hide signature --
 Greg7579's gear list:Greg7579's gear list
Leica Q2 Fujifilm GFX 50R Fujifilm GFX 100 Fujifilm GF 32-64mm F4 Fujifilm 120mm F4 Macro +8 more
DMillier Forum Pro • Posts: 23,017
Re: If you don't contribute to a discussion you don't need to participate

I reckon what you really need for this shot is hundred foot tall legs. Then you could look down on it a bit and get the line of the glacier as the primary visual element.

The scale is so immense and the detail so complex and overwhelming, it's really difficult to make a human sized picture of it. Great detail for pixel peeping though!

-- hide signature --
jonas ar
jonas ar Contributing Member • Posts: 915
Re: Another approach...

Erik Kaffehr wrote:

We can see that mst good lenses reach optimum sharpness at f/4. The only factor reducing sharpness when stopping down is diffraction.

The airy radius is 1.22 * N * Lambda. Assuming Lambda being 0.55 micron, we would get the airy radius where diffraction cames into role to be 1.22 * 4 * 0.55 -> 2.7 microns.

The corresponding diameter is 5.4 microns.

So, it may make some sense to use around 5-6 microns as CoC good lenses at optimal aperture.

But, that would yield a pretty short DoF.

Best regards

Erik

How does color sampling affect your choice?

Erik Kaffehr
OP Erik Kaffehr Veteran Member • Posts: 6,291
Re: Another approach...

jonas ar wrote:

Erik Kaffehr wrote:

We can see that mst good lenses reach optimum sharpness at f/4. The only factor reducing sharpness when stopping down is diffraction.

The airy radius is 1.22 * N * Lambda. Assuming Lambda being 0.55 micron, we would get the airy radius where diffraction cames into role to be 1.22 * 4 * 0.55 -> 2.7 microns.

The corresponding diameter is 5.4 microns.

So, it may make some sense to use around 5-6 microns as CoC good lenses at optimal aperture.

But, that would yield a pretty short DoF.

Best regards

Erik

How does color sampling affect your choice?

I don't think it comes into play...

The question is how small CoC we need to achieve maximum sharpness.

We can just look at MTF. MTF is the absolute part of the MTF of the PSF. We know from Jim's data and other sources that most really good lenses peak around f/4 or f/5.6.
Now, why do they peak at those apertures? Because diffraction reduces sharpness when stopping down more. The shape of the diffraction blur is well know.

How comes demosaic into this?

Each element of the imaging chain causes some blur, which we could describe with a Point Spread Function (PSF). It would be said that the original image is convolved with the PSF.
Now, in the frequency plane, convolution is just a multiplication. Therefore we can break down blur into components and just multiply them.

So, lens aberrations have an MTF, that increases when stopping down. All lenses are also affected by diffraction. The MTF of diffraction reduces when stopping down.

Further down the line, the pixels also blur the image, as they have finite resolution, so the pixel has an MTF. Bayer deconvolution also add blur so it has an MTF.

But, the nice thing that in the frequency plane all MTFs multiply. So all PSF have an effekt, except when one component goes to zero, in that case MTF for the system is zero.

So, demosaic does affect system MTF, but it is just a component of the MTF.

Now, once we stop down, diffraction will add more blur. So it probably makes sense to base CoC in the DoF calculation on diffraction.
One of the reasons I look into this is that I use Lumariver DoF calculator. Lumariver uses max of  (2X pixel pitch, Airy) as CoC as default. That will give larger DoF than setting say pixel pitch as CoC.

That approach makes sense. We could use a CoC of say 5 microns, but stopping down would yield much more blur than 5 microns.

Sorry for not being able to provide a better answer.

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
MOD JimKasson Forum Pro • Posts: 36,080
Re: Another approach...

Erik Kaffehr wrote:

We can just look at MTF. MTF is the absolute part of the MTF of the PSF.

MTF is the normalized absolute value of the Fourier transform of the PSF.

Without normalization, it can be computer in one line of Matlab code:

MTF = abs(fftshift(fft2(psf)));

Jim

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
JimKasson
MOD JimKasson Forum Pro • Posts: 36,080
Re: Another approach...

Erik Kaffehr wrote:

But, the nice thing that in the frequency plane all MTFs multiply. So all PSF have an effekt, except when one component goes to zero, in that case MTF for the system is zero.

This is only an approximation, and one that doesn't work very well in estimating the effects of various kinds of lens aberrations. It works better with diffraction and pixel aperture, though. We have dropped the phase plane to get from PSF to MTF, and for accurate estimations of combined effects, we'd have to multiply the Fourier transforms of the component PSFs, not the absolute value of those components.

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
JimKasson
MOD JimKasson Forum Pro • Posts: 36,080
Re: Another approach...

Erik Kaffehr wrote:

We can see that mst good lenses reach optimum sharpness at f/4. The only factor reducing sharpness when stopping down is diffraction.

The airy radius is 1.22 * N * Lambda. Assuming Lambda being 0.55 micron, we would get the airy radius where diffraction cames into role to be 1.22 * 4 * 0.55 -> 2.7 microns.

The corresponding diameter is 5.4 microns.

So, it may make some sense to use around 5-6 microns as CoC good lenses at optimal aperture.

I consider that to pessimistic.

https://blog.kasson.com/the-last-word/blur-circle-size-estimation-with-encircled-energy/

http://web.ipac.caltech.edu/staff/fmasci/home/astro_refs/PSFtheory.pdf

So your 5.4 um diffraction diameter would be about 3.9 um for EED70.

-- hide signature --
 JimKasson's gear list:JimKasson's gear list
Nikon Z7 Fujifilm GFX 100 Sony a7R IV Sony a9 II +1 more
Erik Kaffehr
OP Erik Kaffehr Veteran Member • Posts: 6,291
Re: Another approach...

JimKasson wrote:

Erik Kaffehr wrote:

We can see that mst good lenses reach optimum sharpness at f/4. The only factor reducing sharpness when stopping down is diffraction.

The airy radius is 1.22 * N * Lambda. Assuming Lambda being 0.55 micron, we would get the airy radius where diffraction cames into role to be 1.22 * 4 * 0.55 -> 2.7 microns.

The corresponding diameter is 5.4 microns.

So, it may make some sense to use around 5-6 microns as CoC good lenses at optimal aperture.

I consider that to pessimistic.

https://blog.kasson.com/the-last-word/blur-circle-size-estimation-with-encircled-energy/

http://web.ipac.caltech.edu/staff/fmasci/home/astro_refs/PSFtheory.pdf

So your 5.4 um diffraction diameter would be about 3.9 um for EED70.

Jim,
Thanks for your efforts. I am on travel right now, but I will check the references.

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

Erik Kaffehr
OP Erik Kaffehr Veteran Member • Posts: 6,291
Re: Another approach...

JimKasson wrote:

Erik Kaffehr wrote:

We can see that mst good lenses reach optimum sharpness at f/4. The only factor reducing sharpness when stopping down is diffraction.

The airy radius is 1.22 * N * Lambda. Assuming Lambda being 0.55 micron, we would get the airy radius where diffraction cames into role to be 1.22 * 4 * 0.55 -> 2.7 microns.

The corresponding diameter is 5.4 microns.

So, it may make some sense to use around 5-6 microns as CoC good lenses at optimal aperture.

I consider that to pessimistic.

https://blog.kasson.com/the-last-word/blur-circle-size-estimation-with-encircled-energy/

http://web.ipac.caltech.edu/staff/fmasci/home/astro_refs/PSFtheory.pdf

So your 5.4 um diffraction diameter would be about 3.9 um for EED70.

Thanks for that info! 
Just to say, I considered EED70, but didn't know how to calculate it.

Anyway, my starting point was that there is a point where diffraction comes into play and that is about the point where sharpness peaks.

So EED70 (as you suggest) may be a good starting point for CoC, if maximum image quality is strived for.

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

Erik Kaffehr
OP Erik Kaffehr Veteran Member • Posts: 6,291
Re: Another approach...

JimKasson wrote:

Erik Kaffehr wrote:

But, the nice thing that in the frequency plane all MTFs multiply. So all PSF have an effekt, except when one component goes to zero, in that case MTF for the system is zero.

This is only an approximation, and one that doesn't work very well in estimating the effects of various kinds of lens aberrations. It works better with diffraction and pixel aperture, though. We have dropped the phase plane to get from PSF to MTF, and for accurate estimations of combined effects, we'd have to multiply the Fourier transforms of the component PSFs, not the absolute value of those components.

Tanks for making those points. I am losely aware of the definitions, but I didn't want to go into to much detail, as my posting was on the complex side anyway.

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

Greg7579
Greg7579 Forum Pro • Posts: 14,041
Re: If you don't contribute to a discussion you don't need to participate

highdesertmesa wrote:

Greg7579 wrote:

Erik Kaffehr wrote:

Greg7579 wrote:

JimKasson wrote:

Greg7579 wrote:

I'm not interested in conducting tests for you Jim.

You've changed your mind so quickly from what you said here ?

I read your pleas and I like you and respect you and learn from you so I decide to do it. Then I get people warning me not to. They say, "Greg just shoot. Don't get sucked into their testing world or they will eat you alive." Let them do the tests. All good. You just shoot and tell us what you see and what your experiences are. We can weigh it all on our own. Testing and shooting. All good.

Anyway, It's no big deal. Really.

I would leave it alone, but you guys never will. There will be 20 more testing threads on DOF all because I said what I know I see.

Hi Greg,
This thread is about the workings of DoF, explaining how it works. Your statements are a bit extraordinary. So they require extraordinary evidence.
You have not provided any such evidence.
Best regards

Erik

Hi Erik,

Evidence. LOL.

Go have some fun and shoot.

There is nothing extraordinary about anything I have said about GFX DOF.

I see the evidence every day when I shoot.

Go shoot some GFX and you will see.

But no worries. Carry on Sir.

Hey ... here is some GFX DOF. I took this a short time ago on a hike by our hotel.

Check out that beautiful glacier. Svinafellsjokul. The aiming point was on the close ice face, that was very far away. I took another one with the aiming point on the back mountains. Guess what happened? And what do you think the tables say?

This looks very natural, filmic. But to my eye at 100% magnification, it looks like it was taken on 50mp full frame camera and sized up to 100mp with AI software. I much prefer the eye-watering sharpness of the 50S/R at 100%.

Aghhh yes.  Maybe.  But what are we really seeing on DPR when we view posted jpegs at "100%"?  Are you seeing anything like what I'm seeing on the exported jpeg from LR that is on my monitor or viewing the raf file in LR?  Note:  I post this a week later after the ongoing discussion about jpegs on DPR.  😁

-- hide signature --
 Greg7579's gear list:Greg7579's gear list
Leica Q2 Fujifilm GFX 50R Fujifilm GFX 100 Fujifilm GF 32-64mm F4 Fujifilm 120mm F4 Macro +8 more
Keyboard shortcuts:
FForum MMy threads