Thought experiment on size, format and resolution for stabilisation.

[dbateman]

First assumption you print at 8x10.

Second assumption, tripods are not allowed.

Third assumption is camera body new or used must be below $3000.

So we are are handheld only, but print size is low so can get away with down sizing the image.

Now what are are arguments for best camera body.

I haven't seen the 135 format 45Mpixel or 60Mpixel cameras used yet below $3000. Maybe a Sigma fpl, but no IBIS. Also for me at least, (not everyone) a certain camera size balances better to be more stable. Like the 50S with grip or Gfx100 body. I always have the grips on the Olympus Em1mk1 and Em5mk2 cameras. My Gm5 is small, but really hard to hold still, without a IS lens. So camera sizes larger seem better to hold still.

Now for the thought experiment. Would a 24Mpixel 135 format camera or smaller sensor be better than the 100Mpixel gfx100, for this print size at say 300dpi, for sharpness, contrast, dynamic range, given that you can significantly down size the Gfx100 vs the other 20-33Mpixel options?

 

[Rob de Loe]

Funnily enough, I just adopted a Ricoh GR IIIx, so I've been thinking a lot about this.

The GR IIIx offers 3 stops of IBIS, and is exceptionally tiny. I can easily hand hold 1/15th of a second reliably, and sometimes even slower. I haven't had nearly as much luck hand holding a 100S and relying on IBIS (but to be fair it's not something I do much so practice might lead to improvements).

The lens on the GR IIIx is exceptionally sharp. The camera outputs 4,000 x 6,000 pixel DNG files. At 8x10 print size, I seriously doubt anyone can tell the difference between a GFX 100S print and a print from the GR IIIx at that size.


100% view of same scene, different cameras. Both are down sampled to 2,400 ppi on the short edge. One is a GFX 100S image and one is Ricoh GR IIIx (using roughly equivalent aperture and field of view).

Of course, the GFX 100S file does much better as print size increases. At 17" on the short edge, it takes a skilled observer to pick out the GFX 100S version. Larger than that it gets easier as print size increases because the APS-C version is really being pushed. Areas of continuous tone (like clear blue skies) hold up much better in the 100S file.

I feel like what you're actually asking here is whether the GFX 100RF is a better choice than a camera with a smaller sensor. I thought about that and chose the GR IIIx because I explicitly wanted a "180 degree opposite" camera from what I normally use (digital view camera with GFX 100S). The 26mm focal length lens on the GR IIIx is perfect for me -- about the same as 55mm (after cropping to 4:3 on the Ricoh) on my GFX.

The GR IIIx is also much smaller and lighter, and much cheaper than a GFX 100RF. For me, it's the perfect "180 degree" camera that (I hope!) will let me do the kind of photography I want to do in a different way some of the time.

 

[rbf]

Funnily enough, I just adopted a Ricoh GR IIIx, so I've been thinking a lot about this.

The GR IIIx offers 3 stops of IBIS, and is exceptionally tiny. I can easily hand hold 1/15th of a second reliably, and sometimes even slower. I haven't had nearly as much luck hand holding a 100S and relying on IBIS (but to be fair it's not something I do much so practice might lead to improvements).

The lens on the GR IIIx is exceptionally sharp. The camera outputs 4,000 x 6,000 pixel DNG files. At 8x10 print size, I seriously doubt anyone can tell the difference between a GFX 100S print and a print from the GR IIIx at that size.


100% view of same scene, different cameras. Both are down sampled to 2,400 ppi on the short edge. One is a GFX 100S image and one is Ricoh GR IIIx (using roughly equivalent aperture and field of view).

Of course, the GFX 100S file does much better as print size increases. At 17" on the short edge, it takes a skilled observer to pick out the GFX 100S version. Larger than that it gets easier as print size increases because the APS-C version is really being pushed. Areas of continuous tone (like clear blue skies) hold up much better in the 100S file.

I feel like what you're actually asking here is whether the GFX 100RF is a better choice than a camera with a smaller sensor. I thought about that and chose the GR IIIx because I explicitly wanted a "180 degree opposite" camera from what I normally use (digital view camera with GFX 100S). The 26mm focal length lens on the GR IIIx is perfect for me -- about the same as 55mm (after cropping to 4:3 on the Ricoh) on my GFX.

The GR IIIx is also much smaller and lighter, and much cheaper than a GFX 100RF. For me, it's the perfect "180 degree" camera that (I hope!) will let me do the kind of photography I want to do in a different way some of the time.

My XT5 is what I use for my "180 degree" camera, usually with some small primes. I was entertaining the idea of a GFX 100RF but I'll be sticking with the XT5.

 

[JimKasson]

First assumption you print at 8x10.

Second assumption, tripods are not allowed.

Third assumption is camera body new or used must be below $3000.

So we are are handheld only, but print size is low so can get away with down sizing the image.

Now what are are arguments for best camera body.

I haven't seen the 135 format 45Mpixel or 60Mpixel cameras used yet below $3000. Maybe a Sigma fpl, but no IBIS. Also for me at least, (not everyone) a certain camera size balances better to be more stable. Like the 50S with grip or Gfx100 body. I always have the grips on the Olympus Em1mk1 and Em5mk2 cameras. My Gm5 is small, but really hard to hold still, without a IS lens. So camera sizes larger seem better to hold still.

Now for the thought experiment. Would a 24Mpixel 135 format camera or smaller sensor be better than the 100Mpixel gfx100, for this print size at say 300dpi, for sharpness, contrast, dynamic range, given that you can significantly down size the Gfx100 vs the other 20-33Mpixel options?

Why would the smaller camera be better for sharpness, contrast, and/or DR? And what does better contrast mean: more contrast? By that standard, Kodalith is the best film.

 

[dbateman]

First assumption you print at 8x10.

Second assumption, tripods are not allowed.

Third assumption is camera body new or used must be below $3000.

So we are are handheld only, but print size is low so can get away with down sizing the image.

Now what are are arguments for best camera body.

I haven't seen the 135 format 45Mpixel or 60Mpixel cameras used yet below $3000. Maybe a Sigma fpl, but no IBIS. Also for me at least, (not everyone) a certain camera size balances better to be more stable. Like the 50S with grip or Gfx100 body. I always have the grips on the Olympus Em1mk1 and Em5mk2 cameras. My Gm5 is small, but really hard to hold still, without a IS lens. So camera sizes larger seem better to hold still.

Now for the thought experiment. Would a 24Mpixel 135 format camera or smaller sensor be better than the 100Mpixel gfx100, for this print size at say 300dpi, for sharpness, contrast, dynamic range, given that you can significantly down size the Gfx100 vs the other 20-33Mpixel options?

Why would the smaller camera be better for sharpness, contrast, and/or DR? And what does better contrast mean: more contrast? By that standard, Kodalith is the best film.

 

[Rob de Loe]

I am genuinely curious what arguments people make either way. Rob seems to be defending his recent purchase.

Heck no. We are on a photography forum where purchases need not be defended or justified!

You asked, "Would a 24Mpixel 135 format camera or smaller sensor be better than the 100Mpixel gfx100, for this print size at say 300dpi, for sharpness, contrast, dynamic range, given that you can significantly down size the Gfx100 vs the other 20-33Mpixel options?"

I argued that for an 8x10 print, it's a wash. Use a 100 MP sensor if that's what you have handy, but I don't think you'll necessarily see a better print than the one made with the 24 MP GR IIIx if post processing is allowed and if the photographer knows how to make a photograph. This finding does not apply to every 24 MP sensor (I hope that's obvious).

Even more simply: I don't think 8x10 prints offer the best business case for a 100 MP sensor.

 

[dbateman]

I am genuinely curious what arguments people make either way. Rob seems to be defending his recent purchase.

Heck no. We are on a photography forum where purchases need not be defended or justified!

You asked, "Would a 24Mpixel 135 format camera or smaller sensor be better than the 100Mpixel gfx100, for this print size at say 300dpi, for sharpness, contrast, dynamic range, given that you can significantly down size the Gfx100 vs the other 20-33Mpixel options?"

I argued that for an 8x10 print, it's a wash. Use a 100 MP sensor if that's what you have handy, but I don't think you'll necessarily see a better print than the one made with the 24 MP GR IIIx if post processing is allowed and if the photographer knows how to make a photograph. This finding does not apply to every 24 MP sensor (I hope that's obvious).

Even more simply: I don't think 8x10 prints offer the best business case for a 100 MP sensor.

True. Also you did state the Gr IIIx image stabilisation is better in your hand than the Gfx100S. That is a major factor.

An argument I was thinking is that 8x10 at 300dpi is basically only 7.2 Mpixels of perfect in focus sharp, stable image. So how much blur (shaken pixels) can 24, 33 or 100Mpixels allow to reach this?

We have a higher chance of less blur with the 100 than the 24. But maybe the 24 has better IBIS than that of the 100, which off sets that advantage?

I know this is multiple variable problem and maybe an argument for a gfx camera over the 135 format camera.

I was just wanting to get people thinking, especially after Jim's recent posts about Dof, equivalence and format size.

 

[Doppler9000]

61 MP, 5-axis IBIS…

 

[Rob de Loe]

I am genuinely curious what arguments people make either way. Rob seems to be defending his recent purchase.

Heck no. We are on a photography forum where purchases need not be defended or justified!

You asked, "Would a 24Mpixel 135 format camera or smaller sensor be better than the 100Mpixel gfx100, for this print size at say 300dpi, for sharpness, contrast, dynamic range, given that you can significantly down size the Gfx100 vs the other 20-33Mpixel options?"

I argued that for an 8x10 print, it's a wash. Use a 100 MP sensor if that's what you have handy, but I don't think you'll necessarily see a better print than the one made with the 24 MP GR IIIx if post processing is allowed and if the photographer knows how to make a photograph. This finding does not apply to every 24 MP sensor (I hope that's obvious).

Even more simply: I don't think 8x10 prints offer the best business case for a 100 MP sensor.

True. Also you did state the Gr IIIx image stabilisation is better in your hand than the Gfx100S. That is a major factor.

An argument I was thinking is that 8x10 at 300dpi is basically only 7.2 Mpixels of perfect in focus sharp, stable image. So how much blur (shaken pixels) can 24, 33 or 100Mpixels allow to reach this?

We have a higher chance of less blur with the 100 than the 24.

Do we though? If the sensor moves the same distance during the exposure because I can't hold it steady, how does more pixels produce less blur?

But maybe the 24 has better IBIS than that of the 100, which off sets that advantage?

I don't know that it does because I have limited experience with IBIS on the 100S. I'm such a tripod guy. All I can say is IBIS on the tiny Ricoh works very well. It certainly doesn't hurt that it's a featherweight in comparison to the GFX 100S with even the lightest lens. Some people claim that a heavier camera is easier to hold steady; that's not the case for me.

I know this is multiple variable problem and maybe an argument for a gfx camera over the 135 format camera.

I was just wanting to get people thinking, especially after Jim's recent posts about Dof, equivalence and format size.

 

[SeeingSomeThingsMissingMany]

I don't know that it does because I have limited experience with IBIS on the 100S. I'm such a tripod guy. All I can say is IBIS on the tiny Ricoh works very well. It certainly doesn't hurt that it's a featherweight in comparison to the GFX 100S with even the lightest lens. Some people claim that a heavier camera is easier to hold steady; that's not the case for me.

My Ricoh Gx8 no ibis compact 1/1.8" CCD sensor perhaps my favourite digital compact ever. Photographed several of my favorite dearest even photographs with it 2008 2009. Could keep it steady handheld down to 1/4 - 1/15 if I recall.

https://www.dpreview.com/articles/8645107435/ricoh-gx8

It's a weirrrd one re : weight. For whatever reason I found a 800g-1400g dslr body no ibis easier to keep steadier than say a 400g mirrorless for which ibis mos def helps me.

--
Photography after all is interplay of light alongside perspective.

 

[NAwlins Contrarian]

First assumption you print at 8x10.

Second assumption, tripods are not allowed.

Third assumption is camera body new or used must be below $3000.

So we are are handheld only, but print size is low so can get away with down sizing the image.

Now what are are arguments for best camera body.

I haven't seen the 135 format 45Mpixel or 60Mpixel cameras used yet below $3000. Maybe a Sigma fpl, but no IBIS. Also for me at least, (not everyone) a certain camera size balances better to be more stable. Like the 50S with grip or Gfx100 body. I always have the grips on the Olympus Em1mk1 and Em5mk2 cameras. My Gm5 is small, but really hard to hold still, without a IS lens. So camera sizes larger seem better to hold still.

Now for the thought experiment. Would a 24Mpixel 135 format camera or smaller sensor be better than the 100Mpixel gfx100, for this print size at say 300dpi, for sharpness, contrast, dynamic range, given that you can significantly down size the Gfx100 vs the other 20-33Mpixel options?

You are talking about large degrees of downsampling and dynamic range compression, so within reason, the software / downsampling and sharpening algorithms, and the dynamic range squeezing techniques, probably matter more than the camera to the print's sharpness, contrast, and dynamic range.

If you're printing 8x10" at 300 ppi, that's 2400x3000 pixels. The ubiquitous 24 MP FF and APS-C cameras therefore need a 1.67x linear downsampling. With e.g. a 45 MP Nikon Z7 II, it's a 2.29:1 downsampling. With a Sony A7C R (as Doppler9000 pointed out, a 60 MP FF camera with pretty good IBIS available new for $2998) it's 2.64:1. And with one of the GFX 100 models, it's 3.64:1. With first-rate lenses and technique, I strongly suspect that any of these cameras can capture essentially all the detail that's visible in the specified print.

The smaller formats have modest deficits in capture dynamic range. Whether that matters at all depends on the subject. And of course, at ISO 100 or whatever, all of them can capture far more dynamic range than a print can display, so again there's a potentially-large effect from the way whatever dynamic range is captured gets mapped to the print.

IME, if the goal is an 8x10" print, then my little Canon S110 (a 2012 model with a 12 MP 7.44x5.58mm CMOS sensor) leaves only a modest amount of room for improvement at the wider end of its zoom range (before diffraction gets bad) when there's enough light and its autofocus and lack of a viewfinder aren't major limitations.

 

[dbateman]

If you're printing 8x10" at 300 ppi, that's 2400x3000 pixels. The ubiquitous 24 MP FF and APS-C cameras therefore need a 1.67x linear downsampling. With e.g. a 45 MP Nikon Z7 II, it's a 2.29:1 downsampling. With a Sony A7C R (as Doppler9000 pointed out, a 60 MP FF camera with pretty good IBIS available new for $2998) it's 2.64:1. And with one of the GFX 100 models, it's 3.64:1. With first-rate lenses and technique, I strongly suspect that any of these cameras can capture essentially all the detail that's visible in the specified print.

Ok so useing your linear down sampling values you gain a whole point of movement using the Gfx100 over the Sony A7CR, ignoring any effects from IBIS. The pixel pitch is the same roughly 3.78um. I can hold the Gfx100 steadier than the smaller Sony body, but I will ignore that as its hard to quantify. I don't know if I can easily quantify the IBIS units, but seems like the Gfx100 between these two cameras can be shaken more and result in sharper handheld images, under my constraints.

We have a greater factor over the 24Mpixel models, but we must remember the pixel pitch is roughly 5.94um vs 3.78um. So 2.17 fold difference in linear downsampling, but a 1.57 fold difference in pixel pitch. So ignoring IBIS and general holding of the camera, you can shake the Gfx100 more than a 24Mpixel camera. But now this case might not pan out if there is excellent IBIS in the camera body.

 

[NAwlins Contrarian]

If you're printing 8x10" at 300 ppi, that's 2400x3000 pixels. The ubiquitous 24 MP FF and APS-C cameras therefore need a 1.67x linear downsampling. With e.g. a 45 MP Nikon Z7 II, it's a 2.29:1 downsampling. With a Sony A7C R (as Doppler9000 pointed out, a 60 MP FF camera with pretty good IBIS available new for $2998) it's 2.64:1. And with one of the GFX 100 models, it's 3.64:1. With first-rate lenses and technique, I strongly suspect that any of these cameras can capture essentially all the detail that's visible in the specified print.

Ok so useing your linear down sampling values you gain a whole point of movement using the Gfx100 over the Sony A7CR, ignoring any effects from IBIS. The pixel pitch is the same roughly 3.78um. I can hold the Gfx100 steadier than the smaller Sony body, but I will ignore that as its hard to quantify. I don't know if I can easily quantify the IBIS units, but seems like the Gfx100 between these two cameras can be shaken more and result in sharper handheld images, under my constraints.

We have a greater factor over the 24Mpixel models, but we must remember the pixel pitch is roughly 5.94um vs 3.78um. So 2.17 fold difference in linear downsampling, but a 1.57 fold difference in pixel pitch. So ignoring IBIS and general holding of the camera, you can shake the Gfx100 more than a 24Mpixel camera. But now this case might not pan out if there is excellent IBIS in the camera body.

IMOPO, you're over-complicating this. This is how I see it:

If the photo-taking conditions are such that camera shake net of IBIS and/or stabilized lenses becomes a substantial factor relative to an 8x10" print, then most of your shots won't capture 20 MP of real detail / sharpness, to say nothing of 102 MP.

On the other hand, if the combination of lens angle of view, shutter speed, IBIS, lens OS, and photographer stability is such that camera shake is not a substantial factor, then any of these cameras can capture plenty of detail / provide plenty of sharpness relative to an 8x10" print. And the actual degree of detail and sharpness of the actual print depends mostly on downsampling and sharpening procedures and algorithms and the paper onto which you print.

Then we get to the question of how the camera (and the lenses you use with it) affects whether the photo-taking conditions are such that camera shake becomes a substantial factor. Lens OS and then IBIS have been revolutionary. Comparisons aren't perfect, but there is a standard CIPA method for comparing it. Relative to your $3000 price point, e.g. the GFX100S (original) has IBIS rated for 6 EV, and the A7C R has IBIS rated for 7 EV. Other factors affect hand-holdabilty. Mirror slap was an issue with SLRs, but not with MILCs. Focal-plane shutters create some movement. Fully-electronic shutters help there; the 102 MP sensor's is very slow (1/6 s, per Horshack) relative to many uses, but the A7C R's is slow too (1/10 s). Body shape, weight, and grip matter, but those are pretty personal. I think viewfinder behavior (blackout etc.) matters, but again that's personal. And what's best when you're fresh in the morning might not be best if you've been hiking all day or shooting all night.

But in the end, I keep circling back to the same point, where practical experience of many people leads me to the same conclusion regarding modern CMOS sensors and processing software: if the final output is an 8x10" print, even an excellent one, then whether the sensor is four-thirds or APS-C or FF or 44x33 or 54x40 is less important than a bunch of other things.