A question regarding pixel size and the number of pixel

[JimKasson]

"blowing out of the water" is a phrase that is used too often and many times is an exaggeration.

Not to mention the fact that it has no measurable meaning

My personal pet peeve is every new product seems to be a "game changer", which also has no measurable meaning

Quantitative thinking is harder than qualitative thinking. Insistence on scalar metrics is comforting to some people, but is ultimately subjective, since the choice of the weights is subjective.

I have an idea. We need to create a new unit of measure. We can call the unit a GameChanger.

In use, a measure of 0 gc's would mean no change and a measure of 100 gc's would mean an entirely different device

You could even have negative gc's, meaning the change was regressive

Reminds me of the unit of feminine beauty, the milliHelen, which is equal to the number of ships launched.

 

[Schrodingers cat]

"blowing out of the water" is a phrase that is used too often and many times is an exaggeration.

Not to mention the fact that it has no measurable meaning

My personal pet peeve is every new product seems to be a "game changer", which also has no measurable meaning

Quantitative thinking is harder than qualitative thinking. Insistence on scalar metrics is comforting to some people, but is ultimately subjective, since the choice of the weights is subjective.

I have an idea. We need to create a new unit of measure. We can call the unit a GameChanger.

In use, a measure of 0 gc's would mean no change and a measure of 100 gc's would mean an entirely different device

You could even have negative gc's, meaning the change was regressive

Reminds me of the unit of feminine beauty, the milliHelen, which is equal to the number of ships launched.

I was thinking we could make it quantitive by crowd sourcing data. Like the number of GameChangers is dependent on the number of thumbs up a particular device receives

And I think I have reached the limits of elasticity regarding how far I can stretch this BS before it achieves off topic status

 

[deednets]

I just had a chat with somebody on the Sony FF forum who claimed that the GFX100 would "blow the A7RV out of the water". An easy assumption to make, but I always failed to see the logic. Bear with me for a moment, here is an image shot with the lowly (I think) Fuji X-T1, 10 years ago in Bhaktapur, Nepal, a few days before the big earthquake:



I used the image as an example, where I asked as to how this image could lose dynamic range when you crop this image? Cut the black stuff off, either in post - or by using a par of scissors??

I feel like I would like to put this animal to rest, once and if possible, for all: If the above shot was taken with a GFX100 - and then cropped to the above size - would then dynamic range only exists across the WHOLE photo? And ignore the differences from brightest to darkest spots in the centre?

The way I see it - and where I am possibly be wrong is this:

The pixel pitch of the 26Mpx sensor is identical to the 60Mpx A7RV/Q3/Leica 11 - and the 100Mpx GFX100. Just more pixel, but per-pixel no jump in quality. Is this where I got it wrong?? Somebody here posted some landscape photos taken with his GFX100RF and claimed a tonal quality in his cropped examples. But, but, but??? Would a 100Mpx sensor with the same pixel pitch cropped to 60Mpx or below not have the same DR as the lowly A7RV//Q3 etc??

Anyway I thought the "blowing out of the water", often used on dpreview, when the difference is North of Crass, got me thinking that when people claim the tonality of a crop has the MF tonality, are they actually correct?

Note: I have dabbled with a Hasselblad once, landscapes in Laos, where I couldn't fully verify the superiority, you know, the blow out of the water "thing", but hey, maybe my standards are just not particularly high, a claim I have heard a few times, so must be some truth in it, right??

Thanks for a short answer, if anybody here can. I am aware of Bill Claff's website where FF sensors lose DR when shot in APS-C mode, like the brightest spot MUST be outside the APS-C crop to verify this??

Deed

The assumption is that the images are printed at the same size. So the noise in the cropped image will be magnified more in the print.

I wasn't talking about printing at the same size, that's equivalent to the re-size argument, correct? I was talking about dynamic range.

Comparing dynamic range makes only sense when you compare it at the same output size.

Yes but: pick 2 spots in my image above then see how many stops this covers. Why does this then depend on how large your output size is? You may rememebr those DR graphs with 15 or so stops from clear-white to black.

This was shown, all 15 or whatever, stops in a 200x800 image. Or whatever the size of that thing was ...

If you then turn this on it's head and say that a 3000x3000px crop, say, from your GFX100RF cannot depict 15 stops within the same image as the image output needs to be larger?

What am I missing? WEre those DR charts showing those shades of grey incorrect? Or do those shots taken with the GFX100RF lose dynamic range because the guy cropped the original image?

Dunno ... but I am clearly not getting it.

Deed

Tthis new article by Jim Kasson should clarify what DR is and how it should be observed, compared, and measured:

https://blog.kasson.com/the-last-word/noise-dynamic-range-and-print-size/

Thank you! I read it, but think I am not following the emphasis on "print size"?? What if you look at an image at 100% on a screen? Does the overall sensor size then matter less when you can see a maximum of 2560x1600px on a 30" DELL IPS? If then the pixel size is the same, is the Fuji X-Pro3 26Mpx sensor indistinguishable from the GFX100 when all you look at is the 2560x1600px?

What if you then print the pixels you can see to a 30" (diagonal) print?

Thanks anyway ...

Deed

I'm guessing your confusion is occurring because you are not taking into account the effect of noise of dynamic range. Jim explained this perfectly, but let me have a go in a more layman style, maybe that'll will help.

The highlight end of dynamic range is limited by the clipping point: you can't see tonality in clipped highlights. The shadow end of dynamic range is in a sense also limited by clipping in that you can no longer see tonality once the shadows are clipped to black.Taken at face value it would seem the clipping points determine dynamic range and ought to the same for all print sizes.

But that doesn't really reflect the real world situation. The shadows tend to be limited by noise long before they get clipped to black: once noise rises above a certain threshold level (subjectively determined, everyone has different tolerances), we can no longer accept it and are no longer prepared to lift shadows to reveal further tones. The normal way we deal with excessive noise in the shadows is to keep them dark to hide it. Artificially clipping blacks to disguise noise reduces dynamic range.

The amount of noise we can tolerate in the shadows is very much dependent on print size: the bigger the print, the more obvious the noise. This means that with big noisy prints we have to darken the shadows more to cover it up, reducing overall dynamic range. Thus bigger prints have their dynamic range more limited by noise than smaller prints where the reduced enlargement factor and noise averaging disguises the noise. We perceive this as the smaller print having both less noise and greater dynamic range.

Thank you will read this later! I am aware of the noise, but was a bit hung up on the notion that a GFX100 would "blow an A7RV out of the water" ... no questions asked. My "internal" debate then questioned this by asking somewhat childish questions, like if you go into any photo at 100% on a 30" screen, would you then see this "blow out of the water" thing, if the pixel size was exactly the same?

And from there on it went past names calling, etc so I thought I would ask here.

I have used MFT, APS-C, FF and also MF, plus use an Osmo Pocket 3, 1" sensor so have some idea as to how size, regarding printing matters.

Thanks again!

Deed

 

[deednets]

I just had a chat with somebody on the Sony FF forum who claimed that the GFX100 would "blow the A7RV out of the water". An easy assumption to make, but I always failed to see the logic. Bear with me for a moment, here is an image shot with the lowly (I think) Fuji X-T1, 10 years ago in Bhaktapur, Nepal, a few days before the big earthquake:



I used the image as an example, where I asked as to how this image could lose dynamic range when you crop this image? Cut the black stuff off, either in post - or by using a par of scissors??

I feel like I would like to put this animal to rest, once and if possible, for all: If the above shot was taken with a GFX100 - and then cropped to the above size - would then dynamic range only exists across the WHOLE photo? And ignore the differences from brightest to darkest spots in the centre?

The way I see it - and where I am possibly be wrong is this:

The pixel pitch of the 26Mpx sensor is identical to the 60Mpx A7RV/Q3/Leica 11 - and the 100Mpx GFX100. Just more pixel, but per-pixel no jump in quality. Is this where I got it wrong?? Somebody here posted some landscape photos taken with his GFX100RF and claimed a tonal quality in his cropped examples. But, but, but??? Would a 100Mpx sensor with the same pixel pitch cropped to 60Mpx or below not have the same DR as the lowly A7RV//Q3 etc??

Anyway I thought the "blowing out of the water", often used on dpreview, when the difference is North of Crass, got me thinking that when people claim the tonality of a crop has the MF tonality, are they actually correct?

Note: I have dabbled with a Hasselblad once, landscapes in Laos, where I couldn't fully verify the superiority, you know, the blow out of the water "thing", but hey, maybe my standards are just not particularly high, a claim I have heard a few times, so must be some truth in it, right??

Thanks for a short answer, if anybody here can. I am aware of Bill Claff's website where FF sensors lose DR when shot in APS-C mode, like the brightest spot MUST be outside the APS-C crop to verify this??

Deed

The assumption is that the images are printed at the same size. So the noise in the cropped image will be magnified more in the print.

I wasn't talking about printing at the same size, that's equivalent to the re-size argument, correct? I was talking about dynamic range.

Comparing dynamic range makes only sense when you compare it at the same output size.

Yes but: pick 2 spots in my image above then see how many stops this covers. Why does this then depend on how large your output size is? You may rememebr those DR graphs with 15 or so stops from clear-white to black.

This was shown, all 15 or whatever, stops in a 200x800 image. Or whatever the size of that thing was ...

If you then turn this on it's head and say that a 3000x3000px crop, say, from your GFX100RF cannot depict 15 stops within the same image as the image output needs to be larger?

What am I missing? WEre those DR charts showing those shades of grey incorrect? Or do those shots taken with the GFX100RF lose dynamic range because the guy cropped the original image?

Dunno ... but I am clearly not getting it.

Deed

Tthis new article by Jim Kasson should clarify what DR is and how it should be observed, compared, and measured:

https://blog.kasson.com/the-last-word/noise-dynamic-range-and-print-size/

Thank you! I read it, but think I am not following the emphasis on "print size"?? What if you look at an image at 100% on a screen? Does the overall sensor size then matter less when you can see a maximum of 2560x1600px on a 30" DELL IPS? If then the pixel size is the same, is the Fuji X-Pro3 26Mpx sensor indistinguishable from the GFX100 when all you look at is the 2560x1600px?

What if you then print the pixels you can see to a 30" (diagonal) print?

Thanks anyway ...

Deed

At 2560x1600px you're essentially comparing two 4MP identical images.

The fact that one came from a crop of a 26MP APS-C sensor and the other one from a larger 100MP 33x44mm sensor is irrelevant.

BUT, you are NOT seeing the full images, hence you are NOT comparing their respective dynamic ranges. You are instead comparing the DR of two 4MP images, in which case, they're the same (because you've thrown away more of the larger image).

Yes, I am aware of that! But, say, at base ISO, if you take 2 images, using a sensor that uses identically sized pixels, how was it possible then that MF is inherently better, regardless of the print size? e.g. your UK landscape photos, where the author claimed that the tonality was superior, even when cropped:

GFX 100RF in the UK Lake District

Hi, Some photos below from a memorable day in the Lake District on Friday just gone, going up and around the Fairfield Horseshoe. Set the alarm for before 5am to make sure I was up on the fells for sunrise, and was rewarded with some beautiful light and lingering mist in the valleys below...

www.dpreview.com

My naive question was then: would the cropped images "dynamic range to die for" then always be better than FF or APS-C, even when the pixel size is identical?

I noted that the answers were mainly about noise, print/output media, but not about the pixel size? Does that mean it doesn't matter?

Deed

 

[marco1974]

I just had a chat with somebody on the Sony FF forum who claimed that the GFX100 would "blow the A7RV out of the water". An easy assumption to make, but I always failed to see the logic. Bear with me for a moment, here is an image shot with the lowly (I think) Fuji X-T1, 10 years ago in Bhaktapur, Nepal, a few days before the big earthquake:



I used the image as an example, where I asked as to how this image could lose dynamic range when you crop this image? Cut the black stuff off, either in post - or by using a par of scissors??

I feel like I would like to put this animal to rest, once and if possible, for all: If the above shot was taken with a GFX100 - and then cropped to the above size - would then dynamic range only exists across the WHOLE photo? And ignore the differences from brightest to darkest spots in the centre?

The way I see it - and where I am possibly be wrong is this:

The pixel pitch of the 26Mpx sensor is identical to the 60Mpx A7RV/Q3/Leica 11 - and the 100Mpx GFX100. Just more pixel, but per-pixel no jump in quality. Is this where I got it wrong?? Somebody here posted some landscape photos taken with his GFX100RF and claimed a tonal quality in his cropped examples. But, but, but??? Would a 100Mpx sensor with the same pixel pitch cropped to 60Mpx or below not have the same DR as the lowly A7RV//Q3 etc??

Anyway I thought the "blowing out of the water", often used on dpreview, when the difference is North of Crass, got me thinking that when people claim the tonality of a crop has the MF tonality, are they actually correct?

Note: I have dabbled with a Hasselblad once, landscapes in Laos, where I couldn't fully verify the superiority, you know, the blow out of the water "thing", but hey, maybe my standards are just not particularly high, a claim I have heard a few times, so must be some truth in it, right??

Thanks for a short answer, if anybody here can. I am aware of Bill Claff's website where FF sensors lose DR when shot in APS-C mode, like the brightest spot MUST be outside the APS-C crop to verify this??

Deed

The assumption is that the images are printed at the same size. So the noise in the cropped image will be magnified more in the print.

I wasn't talking about printing at the same size, that's equivalent to the re-size argument, correct? I was talking about dynamic range.

Comparing dynamic range makes only sense when you compare it at the same output size.

Yes but: pick 2 spots in my image above then see how many stops this covers. Why does this then depend on how large your output size is? You may rememebr those DR graphs with 15 or so stops from clear-white to black.

This was shown, all 15 or whatever, stops in a 200x800 image. Or whatever the size of that thing was ...

If you then turn this on it's head and say that a 3000x3000px crop, say, from your GFX100RF cannot depict 15 stops within the same image as the image output needs to be larger?

What am I missing? WEre those DR charts showing those shades of grey incorrect? Or do those shots taken with the GFX100RF lose dynamic range because the guy cropped the original image?

Dunno ... but I am clearly not getting it.

Deed

Tthis new article by Jim Kasson should clarify what DR is and how it should be observed, compared, and measured:

https://blog.kasson.com/the-last-word/noise-dynamic-range-and-print-size/

Thank you! I read it, but think I am not following the emphasis on "print size"?? What if you look at an image at 100% on a screen? Does the overall sensor size then matter less when you can see a maximum of 2560x1600px on a 30" DELL IPS? If then the pixel size is the same, is the Fuji X-Pro3 26Mpx sensor indistinguishable from the GFX100 when all you look at is the 2560x1600px?

What if you then print the pixels you can see to a 30" (diagonal) print?

Thanks anyway ...

Deed

At 2560x1600px you're essentially comparing two 4MP identical images.

The fact that one came from a crop of a 26MP APS-C sensor and the other one from a larger 100MP 33x44mm sensor is irrelevant.

BUT, you are NOT seeing the full images, hence you are NOT comparing their respective dynamic ranges. You are instead comparing the DR of two 4MP images, in which case, they're the same (because you've thrown away more of the larger image).

Yes, I am aware of that! But, say, at base ISO, if you take 2 images, using a sensor that uses identically sized pixels, how was it possible then that MF is inherently better, regardless of the print size? e.g. your UK landscape photos, where the author claimed that the tonality was superior, even when cropped:

https://www.dpreview.com/forums/post/68456969

"even when cropped" depends on how much one crops. If you crop the MF image down to APS-C size, then the tonality will end up being no better or worse than that of an APS-C camera that uses a sensor with identical pixels.

My naive question was then: would the cropped images "dynamic range to die for" then always be better than FF or APS-C, even when the pixel size is identical?

No.

I noted that the answers were mainly about noise, print/output media, but not about the pixel size?

Because the assumption is that images would be compared at their respective native formats, not when cropped to identical pixel sizes.

Does that mean it doesn't matter?

It no longer matters if you compare at 100% on screen, or after cropping the larger image down to the same pixel size ad the smaller one.



--
Marco

Galleries by Marco Raugei

pbase.com

 

[Stan Disbrow]

Hi,

Can we measure the speed of those ships in Furlongs per Fortnight? You know, FF.

Stan

 

[JimKasson]

Yes, I am aware of that! But, say, at base ISO, if you take 2 images, using a sensor that uses identically sized pixels, how was it possible then that MF is inherently better, regardless of the print size?

Not everybody makes that assertion.

e.g. your UK landscape photos, where the author claimed that the tonality was superior, even when cropped:

https://www.dpreview.com/forums/post/68456969

I think that point of view is bogus. If you set the GFX 100x to 35mm crop, and use the same lenses that you'd use on a Sony a7RV, you're going to get substantially identical images.

My naive question was then: would the cropped images "dynamic range to die for" then always be better than FF or APS-C, even when the pixel size is identical?

I noted that the answers were mainly about noise,

DR and noise are two sides to the same coin.

print/output media, but not about the pixel size? Does that mean it doesn't matter?

See above.

Deed

 

[JimKasson]

Hi,

Can we measure the speed of those ships in Furlongs per Fortnight? You know, FF.

Stan

Ah, the venerable FFS system.

 

[Stan Disbrow]

Hi,

Yes. If you shoot a 100 MP MF and then crop from 4:3 to 3:2 then you get exactly what you would from the 60 MP FF sensor. Kind of a handy happenstance, actually.

Stan

 

[deednets]

"blowing out of the water" is a phrase that is used too often and many times is an exaggeration.

Not to mention the fact that it has no measurable meaning

My personal pet peeve is every new product seems to be a "game changer", which also has no measurable meaning

That's a new idea. How about we create a new unit of measure called the GameChanger. A measurement of 0 gc' would be no change while a measurement of 100 gc's would be a totally different device

Didn't want to say, but I North of Cringe, when I read about a game changer ... ;-)

 

[deednets]

Hi,

Can we measure the speed of those ships in Furlongs per Fortnight? You know, FF.

Stan

Ah, the venerable FFS system.

Speaking of which: I noticed that the only lens you appear to own, is the one hard-mounted to the Q2?? Any non-linear measurement for this kind of shooting experience??

Zen??

 

[SrMi]

Hi,

Can we measure the speed of those ships in Furlongs per Fortnight? You know, FF.

Stan

Ah, the venerable FFS system.

Speaking of which: I noticed that the only lens you appear to own, is the one hard-mounted to the Q2?? Any non-linear measurement for this kind of shooting experience??

Zen??

Jim owning only one lens? ROFL! Did you check out other articles on his blog?

 

[deednets]

Hi,

Can we measure the speed of those ships in Furlongs per Fortnight? You know, FF.

Stan

Ah, the venerable FFS system.

Speaking of which: I noticed that the only lens you appear to own, is the one hard-mounted to the Q2?? Any non-linear measurement for this kind of shooting experience??

Zen??

Jim owning only one lens? ROFL! Did you check out other articles on his blog?

Just look at his gear list here

 

[SrMi]

Hi,

Can we measure the speed of those ships in Furlongs per Fortnight? You know, FF.

Stan

Ah, the venerable FFS system.

Speaking of which: I noticed that the only lens you appear to own, is the one hard-mounted to the Q2?? Any non-linear measurement for this kind of shooting experience??

Zen??

Jim owning only one lens? ROFL! Did you check out other articles on his blog?

Just look at his gear list here

I do not update my gear list here, nor do I consider it up-to-date for any DPR member.

 

[Stan Disbrow]

Hi,

I used to have my gear list loaded and kept it up to date. Many items in it which weren't in the site tables and so hand entered. Stuff I have and stuff I used to have, going back to my beginnings.

It got gone back when the site was shutting down. When it kept going, I simply never went to the trouble of entering it all again.

Stan

 

[JimKasson]

Hi,

Can we measure the speed of those ships in Furlongs per Fortnight? You know, FF.

Stan

Ah, the venerable FFS system.

Speaking of which: I noticed that the only lens you appear to own, is the one hard-mounted to the Q2?? Any non-linear measurement for this kind of shooting experience??

Zen??

What makes you say that? I own far too many lenses.

 

[deednets]

Hi,

Can we measure the speed of those ships in Furlongs per Fortnight? You know, FF.

Stan

Ah, the venerable FFS system.

Speaking of which: I noticed that the only lens you appear to own, is the one hard-mounted to the Q2?? Any non-linear measurement for this kind of shooting experience??

Zen??

What makes you say that? I own far too many lenses.

 

[JimKasson]

Hi,

Can we measure the speed of those ships in Furlongs per Fortnight? You know, FF.

Stan

Ah, the venerable FFS system.

Speaking of which: I noticed that the only lens you appear to own, is the one hard-mounted to the Q2?? Any non-linear measurement for this kind of shooting experience??

Zen??

What makes you say that? I own far too many lenses.

Lighten up a bit??

i was trying to correct your misconception.

Anyway, thanks for trying to explain, but, you know, if it takes 3 A4 pages to explain what should somehow be obvious e.g. DR of one blowing another one out of the water, then it cannot be an easy to see or follow issue.

I am not trying to explain anything about dr blowing anything out of the water. I would not use such a construction.

Or, bluntly put: 2 images at full resolution should have settled this. Be obvious. Visible. Without an instruction booklet.

How could I explain a quantitative concept like this one with two images?

You asked a question. I answered it in some detail. Now you're objecting to my answer being too long.

--
https://blog.kasson.com

 

[deednets]

Hi,

Can we measure the speed of those ships in Furlongs per Fortnight? You know, FF.

Stan

Ah, the venerable FFS system.

Speaking of which: I noticed that the only lens you appear to own, is the one hard-mounted to the Q2?? Any non-linear measurement for this kind of shooting experience??

Zen??

What makes you say that? I own far too many lenses.

Lighten up a bit??

i was trying to correct your misconception.

Anyway, thanks for trying to explain, but, you know, if it takes 3 A4 pages to explain what should somehow be obvious e.g. DR of one blowing another one out of the water, then it cannot be an easy to see or follow issue.

I am not trying to explain anything about dr blowing anything out of the water. I would not use such a construction.

Or, bluntly put: 2 images at full resolution should have settled this. Be obvious. Visible. Without an instruction booklet.

How could I explain a quantitative concept like this one with two images?

You asked a question. I answered it in some detail. Now you're objecting to my answer being too long.

NO! I am arguing that if a solution is simple, it should not take long to explain!

But I am not here to stir up conflict so will leave it at that.

 

[JimKasson]

Hi,

Can we measure the speed of those ships in Furlongs per Fortnight? You know, FF.

Stan

Ah, the venerable FFS system.

Speaking of which: I noticed that the only lens you appear to own, is the one hard-mounted to the Q2?? Any non-linear measurement for this kind of shooting experience??

Zen??

What makes you say that? I own far too many lenses.

Lighten up a bit??

i was trying to correct your misconception.

Anyway, thanks for trying to explain, but, you know, if it takes 3 A4 pages to explain what should somehow be obvious e.g. DR of one blowing another one out of the water, then it cannot be an easy to see or follow issue.

I am not trying to explain anything about dr blowing anything out of the water. I would not use such a construction.

Or, bluntly put: 2 images at full resolution should have settled this. Be obvious. Visible. Without an instruction booklet.

How could I explain a quantitative concept like this one with two images?

You asked a question. I answered it in some detail. Now you're objecting to my answer being too long.

NO! I am arguing that if a solution is simple, it should not take long to explain!

But I am not here to stir up conflict so will leave it at that.

If you now understand the explanation, feel free to summarize it in a few words.

So far, you have not indicated that you understand the explanation.