Micro Four Thirds for high depth of field

[Interceptor121]

I have moved from my GH5M2 to a Sony A1 for underwater photography two years ago

However I find myself shooting at f/16 all the times due to dome port optics and lack of depth of field

I recently have acquired a GH7 as I got a second hand housing deal for the GH6 (same dimensions)

I shoot the same canon 8-15mm with both cameras, the GH7 at f/7.1 f/8 and the A1 at f/14 f/16

Despite double the pixels there is very little difference in detail looking by eye indeed I get better edges on the GH7

I therefore started looking at data on optyczne of matrix resolution by aperture

I found out two things:

1. When you stop down a lot the lens the difference in megapixels is reduced as the resolution drops esponentially by fstop

2. When I hit f/16 on full frame equivalent to f/8 on micro four thirds I can't tell the images apart

This is the resuling graph in essence unless I manage to shoot at f/11 my A1 the results are identical

In addition I often use f/11 or even f/16 for tiny macro critters and I just don't have that on full frame.

There is only one scenario my full frame set up is therefore visibly better and this is a surprise

In this plot various cameras with different pixel count as I am doing another study



--
If you like my image I would appreciate if you follow me on social media
instagram http://instagram.com/interceptor121
My flickr sets http://www.flickr.com/photos/interceptor121/
Youtube channel http://www.youtube.com/interceptor121
Underwater Photo and Video Blog http://interceptor121.com
If you want to get in touch don't send me a PM rather contact me directly at my website/social media

 

[Regor250]

Not sure I follow your first statement. Do you mean the difference between the two sensors' total megapixel count is negated by the loss in optical resolution?

--
Roger

 

[Interceptor121]

Not sure I follow your first statement. Do you mean the difference between the two sensors' total megapixel count is negated by the loss in optical resolution?

you can see from the curves that as the aperture stops down the resolution gets close and closer despite the initial higher value

At f4 there is a big difference once you hit f/8 there is less difference by f/16 there is very little difference

I think this is the effect of diffraction of the lens destroying the additional sensor resolution

I have been looking at that too as I have two full frame sensor one 50 and the other 33 megapixels

However when I look at micro four thirds there is not such a remarkable getting close the sensor increase resolution holds up

--
If you like my image I would appreciate if you follow me on social media
instagram http://instagram.com/interceptor121
My flickr sets http://www.flickr.com/photos/interceptor121/
Youtube channel http://www.youtube.com/interceptor121
Underwater Photo and Video Blog http://interceptor121.com
If you want to get in touch don't send me a PM rather contact me directly at my website/social media

 

[ahaslett]

If you look at lens MTF charts from a RAW using a specified body, the diffraction limiting curve is pretty obvious from the centre resolution.

Shooting a 61Mpix FF body for landscape, I’m acutely aware of being boxed in by depth of field, diffraction and resolution.

The good thing is that most FF lenses perform well at f9-12.

I wouldn’t shoot MFT beyond f6.3 for landscape because foliage and skyline detail is lost to the point where you have the same result as shooting wider open and having that part slightly out of the zone of critical focus.

You need to view pretty large before 61Mpix is better than 42Mpix for landscape, actually huge if printed.

My main requirement for lenses is consistent sharpness edge to edge, although there is a minimum IQ expectation. Just enough SA to smooth the focus transition is good, and sun stars are a bonus.

A

 

[Interceptor121]

If you look at lens MTF charts from a RAW using a specified body, the diffraction limiting curve is pretty obvious from the centre resolution.

Shooting a 61Mpix FF body for landscape, I’m acutely aware of being boxed in by depth of field, diffraction and resolution.

The good thing is that most FF lenses perform well at f9-12.

I wouldn’t shoot MFT beyond f6.3 for landscape because foliage and skyline detail is lost to the point where you have the same result as shooting wider open and having that part slightly out of the zone of critical focus.

You need to view pretty large before 61Mpix is better than 42Mpix for landscape, actually huge if printed.

My main requirement for lenses is consistent sharpness edge to edge, although there is a minimum IQ expectation. Just enough SA to smooth the focus transition is good, and sun stars are a bonus.

A

Thats not what the chart are saying

What they say is that an MFT camera at f/8 resolves the same total LPPH than a 60 megapixel full frame camera at f/16

it is also apparent that the drop does not happen in the same fashion

I have interpolated several cameras the factor in front for the MFT camera is smaller while the exponent is the same which means the chart drops less step than a full frame camera as you stop down like saying that diffraction or whatever phenomena at play os less prominent

As result when you operate at identical depth of field the gap closes does not open up even if MFT has really small pixels

I have no explanation for this but it is consistent across several different sensors







--
If you like my image I would appreciate if you follow me on social media
instagram http://instagram.com/interceptor121
My flickr sets http://www.flickr.com/photos/interceptor121/
Youtube channel http://www.youtube.com/interceptor121
Underwater Photo and Video Blog http://interceptor121.com
If you want to get in touch don't send me a PM rather contact me directly at my website/social media

 

[ahaslett]

If you look at lens MTF charts from a RAW using a specified body, the diffraction limiting curve is pretty obvious from the centre resolution.

Shooting a 61Mpix FF body for landscape, I’m acutely aware of being boxed in by depth of field, diffraction and resolution.

The good thing is that most FF lenses perform well at f9-12.

I wouldn’t shoot MFT beyond f6.3 for landscape because foliage and skyline detail is lost to the point where you have the same result as shooting wider open and having that part slightly out of the zone of critical focus.

You need to view pretty large before 61Mpix is better than 42Mpix for landscape, actually huge if printed.

My main requirement for lenses is consistent sharpness edge to edge, although there is a minimum IQ expectation. Just enough SA to smooth the focus transition is good, and sun stars are a bonus.

A

Thats not what the chart are saying

What they say is that an MFT camera at f/8 resolves the same total LPPH than a 60 megapixel full frame camera at f/16

it is also apparent that the drop does not happen in the same fashion

I have interpolated several cameras the factor in front for the MFT camera is smaller while the exponent is the same which means the chart drops less step than a full frame camera as you stop down like saying that diffraction or whatever phenomena at play os less prominent

As result when you operate at identical depth of field the gap closes does not open up even if MFT has really small pixels

I have no explanation for this but it is consistent across several different sensors

System MTF is the multiplier of lens MTF, camera to RAW MTF and RAW to image. The largest component of camera MTF at moderate apertures is sensor resolution. As you stop down a lens, diffraction limits lens MTF. At constant AoV and subject distance, you get the same depth of field and diffraction at the same aperture at constant image size (ie lpph).

Aperture is f-stop x focal length, so your chart would be easier to read if you plotted f4 for MFT and f8 for FF on the same position on the x-axis.

By f8 MFT and f16 FF, you are seeing diffraction having a dominating effect on lens MTF and therefore reducing the added benefit of sensor resolution. The gap between the four FF curves reduces towards f16.

You would expect the G9ii point at f8/to be close to the A7C point at f16.

Is your issue that the G9ii curve looks anomalous?

The only factor left is RAW processing to an image, unless the G9ii secretly has 60Mpix or we are comparing pixel shift with single shot, …

Andrew

--
Infinite are the arguments of mages. Truth is a jewel with many facets. Ursula K LeGuin
Please feel free to edit any images that I post

 

[Interceptor121]

If you look at lens MTF charts from a RAW using a specified body, the diffraction limiting curve is pretty obvious from the centre resolution.

Shooting a 61Mpix FF body for landscape, I’m acutely aware of being boxed in by depth of field, diffraction and resolution.

The good thing is that most FF lenses perform well at f9-12.

I wouldn’t shoot MFT beyond f6.3 for landscape because foliage and skyline detail is lost to the point where you have the same result as shooting wider open and having that part slightly out of the zone of critical focus.

You need to view pretty large before 61Mpix is better than 42Mpix for landscape, actually huge if printed.

My main requirement for lenses is consistent sharpness edge to edge, although there is a minimum IQ expectation. Just enough SA to smooth the focus transition is good, and sun stars are a bonus.

A

Thats not what the chart are saying

What they say is that an MFT camera at f/8 resolves the same total LPPH than a 60 megapixel full frame camera at f/16

it is also apparent that the drop does not happen in the same fashion

I have interpolated several cameras the factor in front for the MFT camera is smaller while the exponent is the same which means the chart drops less step than a full frame camera as you stop down like saying that diffraction or whatever phenomena at play os less prominent

As result when you operate at identical depth of field the gap closes does not open up even if MFT has really small pixels

I have no explanation for this but it is consistent across several different sensors

System MTF is the multiplier of lens MTF, camera to RAW MTF and RAW to image. The largest component of camera MTF at moderate apertures is sensor resolution. As you stop down a lens, diffraction limits lens MTF. At constant AoV and subject distance, you get the same depth of field and diffraction at the same aperture at constant image size (ie lpph).

Aperture is f-stop x focal length, so your chart would be easier to read if you plotted f4 for MFT and f8 for FF on the same position on the x-axis.

By f8 MFT and f16 FF, you are seeing diffraction having a dominating effect on lens MTF and therefore reducing the added benefit of sensor resolution. The gap between the four FF curves reduces towards f16.

You would expect the G9ii point at f8/to be close to the A7C point at f16.

Is your issue that the G9ii curve looks anomalous?

The only factor left is RAW processing to an image, unless the G9ii secretly has 60Mpix or we are comparing pixel shift with single shot, …

Andrew

The first chart in my original post has the full resolution in LPPH all the values are normalised and comparable at equivalent depth of field

in the range of f/8 to f/16 equivalent aperture both the A7C and A7 IV resolve same or less pixels than the G9M2 while the G9 original matches them at f/16

I think what seems to be happening is that once a pixel is got smaller than a certain amount getting it smaller no longer has side effects but the resolution goes up

If you compare the G9 with the G9M2 the latter holds the advantage as you stop down the curves do not get that much closer like it happens to full frame

Why? I do not know but thats what the measures show

In practical terms you know those people that say if this camera diffraction limit kicks in at f/10 then with smaller pixel it may be f/9 well that does not seem to be the case at all for smaller pixels. While the full frame curves are much steeper as resolution increases the MFT ones are practially paraller. If you look at the trendline the difference in the equation are 0.071 to 0.065 for MFT when you go from 25 to 16 megapixels

However the same trendline exponent goes from 0.064 to 0.051 so bigger variation between 50 and 33 megapixels on full frame

No explanation just what the data set say obviously one should test with an identical lens to see exactly what is happening but on the other hand you do not shoot the same lens across systems

--
If you like my image I would appreciate if you follow me on social media
instagram http://instagram.com/interceptor121
My flickr sets http://www.flickr.com/photos/interceptor121/
Youtube channel http://www.youtube.com/interceptor121
Underwater Photo and Video Blog http://interceptor121.com
If you want to get in touch don't send me a PM rather contact me directly at my website/social media

 

[ahaslett]

Are these processed from RAW?

Have you checked the EXIF to make sure you weren’t using a Hi-Res mode on the G9ii by accident (per Sherlock Holmes)?

What ISO were the images taken at?

Don’t see how the difference in resolution and pixel size between the G9 and G9ii could explain these results.

A

 

[Interceptor121]

Are these processed from RAW?

Have you checked the EXIF to make sure you weren’t using a Hi-Res mode on the G9ii by accident (per Sherlock Holmes)?

What ISO were the images taken at?
Don’t see how the difference in resolution and pixel size between the G9 and G9ii could explain these results.

A

The are the results from optyczne always using the same identical process at whatever the base ISO of the camera is

I shoot underwater at f/16 with my A1 a lot. I recently took the GH7 on a trip I use the canon 8-15 with both cameras using metabones. I compare f/8 on the GH7 with f/16 on the A1 using the same dome port and almost identical setting and subject

Despite the megapixels are double when I look 100 at the little details on the fish the A1 resolves nothing more.

I have also taken plenty of fish shots at f/11 with the Panasonic 45/2.8 the detail holds up. If I shoot my 90mm macro on the A1 it is a total soup at f/22. I have no idea why but it is consistent with those charts

It is obvious that real life phenomena are not linear nor simple so when you make an assumption that one thing and another are related exactly the same way perhaps other factors kick in

An additional point is I shoot my A1 at ISO 500 underwater wide angle while I shoot the GH7 at ISO 100. The SNR of the GH7 is actually higher and I believe the resolution of the A1 has dropped a little as I am no longer at base ISO again another consideration of equivalence that one does not account for when you shoot at base ISO on all systems

 

[ahaslett]

Are these processed from RAW?

Have you checked the EXIF to make sure you weren’t using a Hi-Res mode on the G9ii by accident (per Sherlock Holmes)?

What ISO were the images taken at?
Don’t see how the difference in resolution and pixel size between the G9 and G9ii could explain these results.

A

The are the results from optyczne always using the same identical process at whatever the base ISO of the camera is

I shoot underwater at f/16 with my A1 a lot. I recently took the GH7 on a trip I use the canon 8-15 with both cameras using metabones. I compare f/8 on the GH7 with f/16 on the A1 using the same dome port and almost identical setting and subject

Despite the megapixels are double when I look 100 at the little details on the fish the A1 resolves nothing more.

I have also taken plenty of fish shots at f/11 with the Panasonic 45/2.8 the detail holds up. If I shoot my 90mm macro on the A1 it is a total soup at f/22. I have no idea why but it is consistent with those charts

It is obvious that real life phenomena are not linear nor simple so when you make an assumption that one thing and another are related exactly the same way perhaps other factors kick in

An additional point is I shoot my A1 at ISO 500 underwater wide angle while I shoot the GH7 at ISO 100. The SNR of the GH7 is actually higher and I believe the resolution of the A1 has dropped a little as I am no longer at base ISO again another consideration of equivalence that one does not account for when you shoot at base ISO on all systems

Shooting into diffraction territory with equivalent settings is a great leveller of everything.

Voigtlaender don’t let you close the aperture beyond f16 on their lenses. Why waste engineering taking you somewhere no customer who appreciates a decent lens would want to go. I guess that’s their thought process - so f8 on MFT.

My favourite is the 12-45/4, which peaks across the frame at f5.6, so almost no wasted lens. You just shoot f4-5.6 and compose accordingly.

I try and shoot landscape with the A7Riv at f5.6-8. Makes you really think about composition.

A

 

[dgrogers]

I have moved from my GH5M2 to a Sony A1 for underwater photography two years ago

However I find myself shooting at f/16 all the times due to dome port optics and lack of depth of field

I recently have acquired a GH7 as I got a second hand housing deal for the GH6 (same dimensions)

I shoot the same canon 8-15mm with both cameras, the GH7 at f/7.1 f/8 and the A1 at f/14 f/16

Despite double the pixels there is very little difference in detail looking by eye indeed I get better edges on the GH7

I therefore started looking at data on optyczne of matrix resolution by aperture

I found out two things:

1. When you stop down a lot the lens the difference in megapixels is reduced as the resolution drops esponentially by fstop

2. When I hit f/16 on full frame equivalent to f/8 on micro four thirds I can't tell the images apart

This is the resuling graph in essence unless I manage to shoot at f/11 my A1 the results are identical

In addition I often use f/11 or even f/16 for tiny macro critters and I just don't have that on full frame.

There is only one scenario my full frame set up is therefore visibly better and this is a surprise

In this plot various cameras with different pixel count as I am doing another study

Increased aperture means increased ISO, which negates some of the advantage of full frame. Your images, minus the increased resolution on some full frame cameras and potentially better edge sharpness on m43 cameras when using the same lens, should be similar. This may be part of the problem.

The other part, as mentioned elsewhere, is diffraction, and it seems as if we are headed to a point where all systems are diffraction limited and increased resolution won't matter.

 

[SonyX]

Gh5 +12-35/2.8 at 35mm f/8
A7rII +24-70/2.8 at 70mm f/16

RAW files to download

https://drive.google.com/drive/folders/1wspbgb7aQLV5b8CBWDKozC2C62CRYkwT?usp=sharing

ps. as I'm using A7rII primarily with 85/1.4 at f/1.4, first time I see so much dust on a sensor :-O

 

[Gary from Seattle]

If you look at lens MTF charts from a RAW using a specified body, the diffraction limiting curve is pretty obvious from the centre resolution.

Shooting a 61Mpix FF body for landscape, I’m acutely aware of being boxed in by depth of field, diffraction and resolution.

The good thing is that most FF lenses perform well at f9-12.

I wouldn’t shoot MFT beyond f6.3 for landscape because foliage and skyline detail is lost to the point where you have the same result as shooting wider open and having that part slightly out of the zone of critical focus.

I routinely shoot m4/3 @F7.1 because that is what I need when there is a critical FG element in the image; that would be about F16 FF. I shoot a ton of mountain landscape images.

You need to view pretty large before 61Mpix is better than 42Mpix for landscape, actually huge if printed.

My main requirement for lenses is consistent sharpness edge to edge, although there is a minimum IQ expectation. Just enough SA to smooth the focus transition is good, and sun stars are a bonus.

A

 

[JosephScha]

Welcome back to Micro 4/3! Interesting graph indeed.

 

[ahaslett]

If you look at lens MTF charts from a RAW using a specified body, the diffraction limiting curve is pretty obvious from the centre resolution.

Shooting a 61Mpix FF body for landscape, I’m acutely aware of being boxed in by depth of field, diffraction and resolution.

The good thing is that most FF lenses perform well at f9-12.

I wouldn’t shoot MFT beyond f6.3 for landscape because foliage and skyline detail is lost to the point where you have the same result as shooting wider open and having that part slightly out of the zone of critical focus.

I routinely shoot m4/3 @F7.1 because that is what I need when there is a critical FG element in the image; that would be about F16 FF. I shoot a ton of mountain landscape images.

You need to view pretty large before 61Mpix is better than 42Mpix for landscape, actually huge if printed.

My main requirement for lenses is consistent sharpness edge to edge, although there is a minimum IQ expectation. Just enough SA to smooth the focus transition is good, and sun stars are a bonus.

A

I do shoot up to f10 on MFT very occasionally. I can see f7.1 where you have a detailed FG subject and the background is hazy. That’s f14.2 on FF.

The OP charts show how much resolution at the point of critical focus is traded for sharper foregrounds and backgrounds.

A

 

[Jens_ac]

I have moved from my GH5M2 to a Sony A1 for underwater photography two years ago

However I find myself shooting at f/16 all the times due to dome port optics and lack of depth of field

I recently have acquired a GH7 as I got a second hand housing deal for the GH6 (same dimensions)

I shoot the same canon 8-15mm with both cameras, the GH7 at f/7.1 f/8 and the A1 at f/14 f/16

Despite double the pixels there is very little difference in detail looking by eye indeed I get better edges on the GH7

I therefore started looking at data on optyczne of matrix resolution by aperture

I found out two things:

1. When you stop down a lot the lens the difference in megapixels is reduced as the resolution drops esponentially by fstop
2. When I hit f/16 on full frame equivalent to f/8 on micro four thirds I can't tell the images apart

This is the resuling graph in essence unless I manage to shoot at f/11 my A1 the results are identical

In addition I often use f/11 or even f/16 for tiny macro critters and I just don't have that on full frame.

There is only one scenario my full frame set up is therefore visibly better and this is a surprise

In this plot various cameras with different pixel count as I am doing another study

 

[tko]

You can prove that under DOF limited conditions and everything else being equal, you will have approximately the same resolution for any format. Of course, everything else typically isn't equal, but that's another topic.