Cropping in the GFX 100RF and the effects on dynamic range

Fuji is recommending cropping as a method to ameliorate the limitations of a fixed lens in the GFX 100RF. That will affect dynamic range. I've calculated that, using BIll Claff's Photographic Dynamic Range measurements from the GFX 100S II.

Here it is:

3be65bdef58341e1913b43829c6fea12.jpg.png

The vertical axis is PDR in stops. The horizontal axis is ISO setting. The equivalent focal length for a 33x44mm sensor is shown for each line.

I will be happy to take questions and comments in this thread.
I would expect 70mm (2x from 35mm) to "cost" two stops of PDR instead of only one. When we compare FF with m43 PDRs (2x difference), we see a difference of two stops of PDR.

Where is my thinking wrong?
I now realize that my thinking was simplistic. It's not possible to compute the curves precisely without going back to the photon transfer curves and recalculating from there, which I'm too lazy to do at this point. So I've redone the curves with a heuristic that I derived from Bill's data.

Here's the modified formula:

7e11fb371b5e47a59dc3d93920e209e5.jpg.png

The heuristic is 1.1. I also added a column for a focal length of 48mm, which gives the FF pixel height, and checked it against BIll's numbers.

I also changed the 60mm line to a 63mm line, to match the Fuji crop selection.

Here's the corrected graph.

45fa5e5043f043fe96db7759b886ee29.jpg.png

I can't justify the heuristic on theoretical grounds, so take this with a grain of NaCl.

To deal with your question about the slopes, the SNR slope in read noise limited regions is different than that in the photon noise limited regions. PDR is affected by both kinds of noise.
It would be nice if we had measurements to confirm your analysis. It could then apply also to Leica Q's digital crop.
I may have to fire up my camera simulator, if I can remember how to use it.

--
 
Does DR improve if you crop and then run the file through gigapixel Ai?
Any thing that reduces noise increases DR. But that is post processing. The graphs apply to raw files.
 
Talk about a hole in my understanding of how this all works. I have no idea how to conceptualize why DR would change if I crop my image. To a layperson like me, that seems completely nonsensical. I’m not sure I could understand it, even if someone tried to explain it to me. This is a comment about me, and how I’m sometimes surprised by my lack of ability to understand, not about being dubious of the concept itself.

I also had some fun looking at the web page with all the DR’s of various cameras, and was then surprised to see that two cameras I own (X-T5 and X-H2) that have identical sensors and processors have different DR. Is that because the testing has a margin of error? If so, then the variation in a lot of the cameras at “x” DR seem very much equivalent, and that to “rank” them based on these numbers seems next to worthless. I also notice that the GFX 100 and GFX 100 II have different DR. And I thought I understood that this “claim” by Fuji was more smoke and mirrors than real difference.

Arrrgh….. LOL

Sure glad I can still make nice photographs w/o understanding any of this. 😎

Rand
 
Talk about a hole in my understanding of how this all works. I have no idea how to conceptualize why DR would change if I crop my image. To a layperson like me, that seems completely nonsensical. I’m not sure I could understand it, even if someone tried to explain it to me. This is a comment about me, and how I’m sometimes surprised by my lack of ability to understand, not about being dubious of the concept itself.

I also had some fun looking at the web page with all the DR’s of various cameras, and was then surprised to see that two cameras I own (X-T5 and X-H2) that have identical sensors and processors have different DR. Is that because the testing has a margin of error? If so, then the variation in a lot of the cameras at “x” DR seem very much equivalent, and that to “rank” them based on these numbers seems next to worthless. I also notice that the GFX 100 and GFX 100 II have different DR. And I thought I understood that this “claim” by Fuji was more smoke and mirrors than real difference.

Arrrgh….. LOL

Sure glad I can still make nice photographs w/o understanding any of this. 😎

Rand
I think there are a couple things going on, one that is easy to understand, and then whatever Jim is addressing. ;)

The easy one... imagine cropping this. If DR is the range between darkest and lightest, crops from this image will reduce dynamic range (assuming the crop at least leaves out the first or last rows).

c909c1ec867244308905d02fe977c019.jpg
 
Last edited:
Yes, at base ISO..

Now add the sliw lens and the missing IBIS, and in the real world you will crank up the ISO quite often
Fyi, in my tests (at 35mm) ibis was fully effective only about 1 stop - so much for the 8 stop advantage of the 100S ii.

It is a drawback for sure but I would still take compact medium format without IBIS instead of compact full frame or crop format with ibis/ois. Totally personal preference and kind of besides the point:)
In my experience IBIS is much more effective and i usually get 3, 4 stop of improvement in lowlight handheld situations. I don't know why you only get 1 stop better performances
I will go back do some more testing with different focal lengths but at 35mm, my findings, for my technique and the way I viewed the images here , were pretty consistent. Perhaps IBIS is more effective on speeds under 1/30, which is the slowest I measured, but probably (though not hopefully) much lower than 50% effective.
Your results are so far away from my own experience that I can only imagine you need to improve your handholding technique. And I'm very particular about sharpness, mostly using IBIS only to guarantee sharp files using shutter speeds equivalent to or slightly slower than focal length (in DMF context).
 
Talk about a hole in my understanding of how this all works. I have no idea how to conceptualize why DR would change if I crop my image. To a layperson like me, that seems completely nonsensical. I’m not sure I could understand it, even if someone tried to explain it to me. This is a comment about me, and how I’m sometimes surprised by my lack of ability to understand, not about being dubious of the concept itself.

I also had some fun looking at the web page with all the DR’s of various cameras, and was then surprised to see that two cameras I own (X-T5 and X-H2) that have identical sensors and processors have different DR. Is that because the testing has a margin of error? If so, then the variation in a lot of the cameras at “x” DR seem very much equivalent, and that to “rank” them based on these numbers seems next to worthless. I also notice that the GFX 100 and GFX 100 II have different DR. And I thought I understood that this “claim” by Fuji was more smoke and mirrors than real difference.

Arrrgh….. LOL

Sure glad I can still make nice photographs w/o understanding any of this. 😎

Rand
I think there are a couple things going on, one that is easy to understand, and then whatever Jim is addressing. ;)

The easy one... imagine cropping this. If DR is the range between darkest and lightest, crops from this that leave out the first row or the last row will reduce dynamic range.

c909c1ec867244308905d02fe977c019.jpg
Nice try! LOL BUT isn’t any one square capable of being (recording) either the lightest value or the darkest value, or anywhere in between the extremes of the range? Isn’t that how “any one section” of a sensor works? There is obviously a huge conceptual hole in my understanding of this.

Rand
 
Yes, at base ISO..

Now add the sliw lens and the missing IBIS, and in the real world you will crank up the ISO quite often
Fyi, in my tests (at 35mm) ibis was fully effective only about 1 stop - so much for the 8 stop advantage of the 100S ii.

It is a drawback for sure but I would still take compact medium format without IBIS instead of compact full frame or crop format with ibis/ois. Totally personal preference and kind of besides the point:)
In my experience IBIS is much more effective and i usually get 3, 4 stop of improvement in lowlight handheld situations. I don't know why you only get 1 stop better performances
I will go back do some more testing with different focal lengths but at 35mm, my findings, for my technique and the way I viewed the images here , were pretty consistent. Perhaps IBIS is more effective on speeds under 1/30, which is the slowest I measured, but probably (though not hopefully) much lower than 50% effective.
Your results are so far away from my own experience that I can only imagine you need to improve your handholding technique. And I'm very particular about sharpness, mostly using IBIS only to guarantee sharp files using shutter speeds equivalent to or slightly slower than focal length (in DMF context).
 
Yes, at base ISO..

Now add the sliw lens and the missing IBIS, and in the real world you will crank up the ISO quite often
Fyi, in my tests (at 35mm) ibis was fully effective only about 1 stop - so much for the 8 stop advantage of the 100S ii.

It is a drawback for sure but I would still take compact medium format without IBIS instead of compact full frame or crop format with ibis/ois. Totally personal preference and kind of besides the point:)
In my experience IBIS is much more effective and i usually get 3, 4 stop of improvement in lowlight handheld situations. I don't know why you only get 1 stop better performances
I will go back do some more testing with different focal lengths but at 35mm, my findings, for my technique and the way I viewed the images here , were pretty consistent. Perhaps IBIS is more effective on speeds under 1/30, which is the slowest I measured, but probably (though not hopefully) much lower than 50% effective.
Your results are so far away from my own experience that I can only imagine you need to improve your handholding technique. And I'm very particular about sharpness, mostly using IBIS only to guarantee sharp files using shutter speeds equivalent to or slightly slower than focal length (in DMF context).
Can you please advise why improving my technique would benefit ibis more than non ibis? I would expect the opposite. How many stops of clear advantage do you get with ibis on vs. non ibis?
You aren't making that many frames so I'm assuming consistency must be playing a part, otherwise I have no idea.

Do we have a metric that explains what "clear advantage" is? With the 45 f2.8 I have no issues achieving critical sharpness as slow as 1/20 on most attempts. I have a decent hit rate that slow with the 120 f4 (the combination of OIS and IBIS may factor).
 
Yes, at base ISO..

Now add the sliw lens and the missing IBIS, and in the real world you will crank up the ISO quite often
Fyi, in my tests (at 35mm) ibis was fully effective only about 1 stop - so much for the 8 stop advantage of the 100S ii.

It is a drawback for sure but I would still take compact medium format without IBIS instead of compact full frame or crop format with ibis/ois. Totally personal preference and kind of besides the point:)
In my experience IBIS is much more effective and i usually get 3, 4 stop of improvement in lowlight handheld situations. I don't know why you only get 1 stop better performances
I will go back do some more testing with different focal lengths but at 35mm, my findings, for my technique and the way I viewed the images here , were pretty consistent. Perhaps IBIS is more effective on speeds under 1/30, which is the slowest I measured, but probably (though not hopefully) much lower than 50% effective.
Your results are so far away from my own experience that I can only imagine you need to improve your handholding technique. And I'm very particular about sharpness, mostly using IBIS only to guarantee sharp files using shutter speeds equivalent to or slightly slower than focal length (in DMF context).
Can you please advise why improving my technique would benefit ibis more than non ibis? I would expect the opposite. How many stops of clear advantage do you get with ibis on vs. non ibis?
If camera motion causes the IBIS travel limits to be exceeded, it will be ineffective.
 
Yes, at base ISO..

Now add the sliw lens and the missing IBIS, and in the real world you will crank up the ISO quite often
Fyi, in my tests (at 35mm) ibis was fully effective only about 1 stop - so much for the 8 stop advantage of the 100S ii.

It is a drawback for sure but I would still take compact medium format without IBIS instead of compact full frame or crop format with ibis/ois. Totally personal preference and kind of besides the point:)
In my experience IBIS is much more effective and i usually get 3, 4 stop of improvement in lowlight handheld situations. I don't know why you only get 1 stop better performances
I will go back do some more testing with different focal lengths but at 35mm, my findings, for my technique and the way I viewed the images here , were pretty consistent. Perhaps IBIS is more effective on speeds under 1/30, which is the slowest I measured, but probably (though not hopefully) much lower than 50% effective.
Your results are so far away from my own experience that I can only imagine you need to improve your handholding technique. And I'm very particular about sharpness, mostly using IBIS only to guarantee sharp files using shutter speeds equivalent to or slightly slower than focal length (in DMF context).
Can you please advise why improving my technique would benefit ibis more than non ibis? I would expect the opposite. How many stops of clear advantage do you get with ibis on vs. non ibis?
You aren't making that many frames so I'm assuming consistency must be playing a part, otherwise I have no idea.
My ibis shots present very little to no variation. But even if that wasn't the case, the fact I am picking the best ibis shots helps ibis. Unless I didn't understand this point.
Do we have a metric that explains what "clear advantage" is? With the 45 f2.8 I have no issues achieving critical sharpness as slow as 1/20 on most attempts. I have a decent hit rate that slow with the 120 f4 (the combination of OIS and IBIS may factor).
As SrMi phrased it, at what min speed you get 100% sharpness with vs. without ibis. Is that 1, 2 or more stop difference for you?
 
Yes, at base ISO..

Now add the sliw lens and the missing IBIS, and in the real world you will crank up the ISO quite often
Fyi, in my tests (at 35mm) ibis was fully effective only about 1 stop - so much for the 8 stop advantage of the 100S ii.

It is a drawback for sure but I would still take compact medium format without IBIS instead of compact full frame or crop format with ibis/ois. Totally personal preference and kind of besides the point:)
In my experience IBIS is much more effective and i usually get 3, 4 stop of improvement in lowlight handheld situations. I don't know why you only get 1 stop better performances
I will go back do some more testing with different focal lengths but at 35mm, my findings, for my technique and the way I viewed the images here , were pretty consistent. Perhaps IBIS is more effective on speeds under 1/30, which is the slowest I measured, but probably (though not hopefully) much lower than 50% effective.
Your results are so far away from my own experience that I can only imagine you need to improve your handholding technique. And I'm very particular about sharpness, mostly using IBIS only to guarantee sharp files using shutter speeds equivalent to or slightly slower than focal length (in DMF context).
Can you please advise why improving my technique would benefit ibis more than non ibis? I would expect the opposite. How many stops of clear advantage do you get with ibis on vs. non ibis?
If camera motion causes the IBIS travel limits to be exceeded, it will be ineffective.
 
Fuji is recommending cropping as a method to ameliorate the limitations of a fixed lens in the GFX 100RF. That will affect dynamic range. I've calculated that, using BIll Claff's Photographic Dynamic Range measurements from the GFX 100S II.

Here it is:

3be65bdef58341e1913b43829c6fea12.jpg.png

The vertical axis is PDR in stops. The horizontal axis is ISO setting. The equivalent focal length for a 33x44mm sensor is shown for each line.

I will be happy to take questions and comments in this thread.
First off thanks for this. It is appreciated and is about what I expected.

That is a scary looking chart if you are working with f4 and fast shutter speeds, because you are going to need to compensate with iso if you don't have flash. I mean who wants to lose approximately 5 to 6 stops of light by cropping in camera and using higher ISO?



This is why I learn from these charts but take it with a grain of salt when it comes to the final output. Having taken shots with the sensor, I know that what ever the DR is at 12,800 iso I can work with it.
 
Thank you Jim.

I tried to add some lines that show the crops Fuji implemented.

If I read the graphs correctly, then cropping to has a similar DR as if you raise ISO from ISO 100 to:
  • between ISO 125 and ISO 160 for cropping to 45mm
  • between ISO 200 and ISO 320 for cropping to 63mm
  • between ISO 320 and ISO 400 for cropping to 80mm
Correct?

Can someone calculate exact number for the DR losses?

e5223518b8aa4743b8b13afe02bf0e9b.jpg.png
Thanks for this! Never thought to interpret the graph this way.

I think while it will have an effect we are discussing a relatively small range in performance (about 1-3 stops between the uncropped at iso 100 and the other crop values).
 
Nice try! LOL BUT isn’t any one square capable of being (recording) either the lightest value or the darkest value, or anywhere in between the extremes of the range? Isn’t that how “any one section” of a sensor works? There is obviously a huge conceptual hole in my understanding of this.

Rand
Someone is going to come along soon and correct me if I'm wrong, but my understanding is that Dynamic Range in a photograph is nothing more than the range of values between lightest and darkest: Dmax to Dmin.

In that graphic, it goes from pure white to pure black (so maximum dynamic range). If you crop in (e.g., you leave behind the blackest black and the whitest white), then you have reduced the dynamic range of this specific image.

I'm probably muddying the water because the dynamic range of an image is different from the dynamic range of a sensor.
 
Last edited:
I don't know how people view things, but I will pretty much always crop for a number of reasons:
  • I do not see in either 3x2 or 4x3
  • As I print my images I am always going to crop as minimum to just fit the aspect ratio of the paper, either 1.414:1 for A papers or 1.468:1 for A3+
  • I am going to crop based on my artistic interpretation of my image
  • I am not going to worry about what cropping does to dynamic range, I don't begin to understand the discussion and frankly little to no interest in it either
  • There is plenty of DR in images produced by cameras regardless of cropping
This is the kind of image I like taking

6422843e8e364422860417ca9a89923f.jpg

I've no idea what the DR range is in this image, it is cropped to 1x1 and was taken nearly an hour before sunrise, and looks like DPR has done something to it, but you get the point

All of which means I'm sure the GFX100RF with its cropping will produce some amazing images as will my GFX50s ii

Oh, and IBIS would not have helped here either :-)
 
Andy Gordon's marvelous photograph prompted

Perhaps a thread : show us your wide dynamic range photos.

Perhaps helps us see how often to how little we utilise wide dynamic range.

(Not for me to start such a thread as who am I? A nobody 😹).

--
Photography after all is interplay of light alongside perspective.
 
Last edited:
Nice try! LOL BUT isn’t any one square capable of being (recording) either the lightest value or the darkest value, or anywhere in between the extremes of the range? Isn’t that how “any one section” of a sensor works? There is obviously a huge conceptual hole in my understanding of this.

Rand
Someone is going to come along soon and correct me if I'm wrong, but my understanding is that Dynamic Range in a photograph is nothing more than the range of values between lightest and darkest: Dmax to Dmin.

In that graphic, it goes from pure white to pure black (so maximum dynamic range). If you crop in (e.g., you leave behind the blackest black and the whitest white), then you have reduced the dynamic range of this specific image.

I'm probably muddying the water because the dynamic range of an image is different from the dynamic range of a sensor.
This is my intuition on this. To avoid clipping blacks when light is limited, you need to raise ISO to the min level required so you avoid any loss of black details. By doing so however, i.e. raising ISO, some of the highlights start getting clipped because the sensor's max highlight capabilities, of course, haven't changed (increased). Does this make sense?!

--
Apollon
 
Yes, at base ISO..

Now add the sliw lens and the missing IBIS, and in the real world you will crank up the ISO quite often
Fyi, in my tests (at 35mm) ibis was fully effective only about 1 stop - so much for the 8 stop advantage of the 100S ii.

It is a drawback for sure but I would still take compact medium format without IBIS instead of compact full frame or crop format with ibis/ois. Totally personal preference and kind of besides the point:)
In my experience IBIS is much more effective and i usually get 3, 4 stop of improvement in lowlight handheld situations. I don't know why you only get 1 stop better performances
I will go back do some more testing with different focal lengths but at 35mm, my findings, for my technique and the way I viewed the images here , were pretty consistent. Perhaps IBIS is more effective on speeds under 1/30, which is the slowest I measured, but probably (though not hopefully) much lower than 50% effective.
Your results are so far away from my own experience that I can only imagine you need to improve your handholding technique. And I'm very particular about sharpness, mostly using IBIS only to guarantee sharp files using shutter speeds equivalent to or slightly slower than focal length (in DMF context).
Can you please advise why improving my technique would benefit ibis more than non ibis? I would expect the opposite. How many stops of clear advantage do you get with ibis on vs. non ibis?
You aren't making that many frames so I'm assuming consistency must be playing a part, otherwise I have no idea.
My ibis shots present very little to no variation. But even if that wasn't the case, the fact I am picking the best ibis shots helps ibis. Unless I didn't understand this point.
You do not pick the best shot. Instead you shoot 16 images at a given shutter speed and reject that shutter speed for being too slow if one is not sharp among them.

Do we have a metric that explains what "clear advantage" is? With the 45 f2.8 I have no issues achieving critical sharpness as slow as 1/20 on most attempts. I have a decent hit rate that slow with the 120 f4 (the combination of OIS and IBIS may factor).
As SrMi phrased it, at what min speed you get 100% sharpness with vs. without ibis. Is that 1, 2 or more stop difference for you?
--
Apollon
http://www.flickr.com/photos/apollonas/
 
Yes, at base ISO..

Now add the sliw lens and the missing IBIS, and in the real world you will crank up the ISO quite often
Fyi, in my tests (at 35mm) ibis was fully effective only about 1 stop - so much for the 8 stop advantage of the 100S ii.

It is a drawback for sure but I would still take compact medium format without IBIS instead of compact full frame or crop format with ibis/ois. Totally personal preference and kind of besides the point:)
In my experience IBIS is much more effective and i usually get 3, 4 stop of improvement in lowlight handheld situations. I don't know why you only get 1 stop better performances
I will go back do some more testing with different focal lengths but at 35mm, my findings, for my technique and the way I viewed the images here , were pretty consistent. Perhaps IBIS is more effective on speeds under 1/30, which is the slowest I measured, but probably (though not hopefully) much lower than 50% effective.
Your results are so far away from my own experience that I can only imagine you need to improve your handholding technique. And I'm very particular about sharpness, mostly using IBIS only to guarantee sharp files using shutter speeds equivalent to or slightly slower than focal length (in DMF context).
Can you please advise why improving my technique would benefit ibis more than non ibis? I would expect the opposite. How many stops of clear advantage do you get with ibis on vs. non ibis?
You aren't making that many frames so I'm assuming consistency must be playing a part, otherwise I have no idea.
My ibis shots present very little to no variation. But even if that wasn't the case, the fact I am picking the best ibis shots helps ibis. Unless I didn't understand this point.
You do not pick the best shot. Instead you shoot 16 images at a given shutter speed and reject that shutter speed for being too slow if one is not sharp among them.
That's a matter of objective. My objective was to compare the best-case scenario among both approaches (ibis vs. non ibis) and emulate closer a case whereby I would pay max attention to my shots or a case where I would indeed make a few shots to ensure I obtain a very sharp one. The above objective is about eliminating the risk of coming back home with a blurry shot. I don't see however how either objective favors either approach.
Do we have a metric that explains what "clear advantage" is? With the 45 f2.8 I have no issues achieving critical sharpness as slow as 1/20 on most attempts. I have a decent hit rate that slow with the 120 f4 (the combination of OIS and IBIS may factor).
As SrMi phrased it, at what min speed you get 100% sharpness with vs. without ibis. Is that 1, 2 or more stop difference for you?
--
Apollon
http://www.flickr.com/photos/apollonas/
 
Last edited:

Keyboard shortcuts

Back
Top