How does Physical sensor size affect Image Quality ?

[glenerrolrd]

The current debate seems to be about how well the Fuji GFX 100 files compare to the larger Phase IQ4 150 files . Could the smaller and less costly MF replace Phase One Backs as the standard for great image quality ?

The value proposition should be a given ..no need to thrash Phase for being much more expensive ..you know if you can afford to pay a premium .

Lets assume for a minute that the thread on print sizes is a given . If you print large there is always an advantage to having larger files that do not require resizing . But how much of your work would really benefit from being able to print super large ...how much of it do you actually print that large ?

As was pointed out the pixel sizes are the same and since they are both Sony BSI sensors ..probably of similar capability .

My standards are lower ...if it looks great on my iMac 5K ...it will print as large as I might every want 17 x22 . Yet I can easily see the differences in image quality between 37.5 Mp Leica S 007 file and a similar Nikon D810 file . In fact when I compared the Panasonic S1R 47MP files to the Leica S files ..again ..no comparison .

So I have been conditioned by experience that each step up in physical sensor size APS-C to FF to MF to LF ...results in a better (to my eye ) aesthetic ....larger files just look better .

 

[dave.gt]

The current debate seems to be about how well the Fuji GFX 100 files compare to the larger Phase IQ4 150 files . Could the smaller and less costly MF replace Phase One Backs as the standard for great image quality ?

The value proposition should be a given ..no need to thrash Phase for being much more expensive ..you know if you can afford to pay a premium .

Lets assume for a minute that the thread on print sizes is a given . If you print large there is always an advantage to having larger files that do not require resizing . But how much of your work would really benefit from being able to print super large ...how much of it do you actually print that large ?

As was pointed out the pixel sizes are the same and since they are both Sony BSI sensors ..probably of similar capability .

My standards are lower ...if it looks great on my iMac 5K ...it will print as large as I might every want 17 x22 . Yet I can easily see the differences in image quality between 37.5 Mp Leica S 007 file and a similar Nikon D810 file . In fact when I compared the Panasonic S1R 47MP files to the Leica S files ..again ..no comparison .

So I have been conditioned by experience that each step up in physical sensor size APS-C to FF to MF to LF ...results in a better (to my eye ) aesthetic ....larger files just look better .

Hi, Roger,

Great topic so I thought for what it is worth (not much) I would throw in my 2cents before the experts crush my spirit. I am high-tech fatigued anyway lately.

Bottom line for me is that I have seen the quality difference just as you described. That said, the S files to me are much better than the D850 files and my little Leica X1 files look fine up to the size you mentioned. But, the S files are "richer" for lack of a better word and obviously will print much larger.

At the end of the day, I don't seem to see a 1:1 correlation in every situation. To me, it seems to depend on many factors, but I do love the rendering of the S files and as Andreas stated, there is more to it than just MP counts.

This discussion will be interesting...

 

[Paul2660]

It's a good question, and for my work 100MP is plenty. With the modern software tools available for uprez, I can easily get 24 x 36 from the D180, D850 etc. I still prefer native pixel output. The 100MP 3100 was the first sensor I have used, (I may be an exception) to where I was able to crop into a smaller part of the image and still get a good file printed.

My largest prints are 6' by 12', in a Natural resources complex, and they were from a lot earlier cameras. I regularly print up to 36 x 72 on canvas, and 40 x 60 on paper.

The flip side I ran into with the 100MP sensor is working the files in CC as with such large files, you are quickly into 3.3GB and 6 GB with layers, and only have PSB format to save in. Which takes longer and can only be opened by CC, even LR can't open a PSB last time I checked. Most software also tend to choke on the pano's taken with the 3100, due to overall pixel count. I just worked up a 6 shot 3100 shot for a client last week, 6 portrait images into a landscape oriented image, and had to cut the image in half to work it with several software tools I prefer.

Several tools I regularly use in CC also seem to choke when on the larger images. Brush in layers.

C1 sadly also has reached a choking point with the 3100 an 4150 files. With the C1 12 the masking toolset is excellent, IMO the best out there. Wonderful options. But attempting to use these now on the 4150 or 3100 files, on a MacPro late 2016, 6 way, latest software from Apple causes a lot of delays, occasional C1 crashes, and a lot of time watching the spinning color wheel. Windows 10 seems to be a bit faster but also reaches a choke point

Love the print from these cameras, (for web work totally overkill) but the output file sizes and current software to work them are my main hangup.

Paul C

 

[algrove]

Paul
Good post to read

I have the Mac Pro 2013 with OWC 2TB SSD , 64GB Ram and the best vid cards available a few year ago when I bought it as a refurbd unit. I have never had the beach ball with my 100MP files that like you say get up to 3-4GB nor with my IQ4150 files which get even larger at times. C1 is very fast and never slows me down. Am I just lucky or is it another issue between our systems?

 

[Shashin]

So I have been conditioned by experience that each step up in physical sensor size APS-C to FF to MF to LF ...results in a better (to my eye ) aesthetic ....larger files just look better .

This is a really complex question. Ultimately, it come down to the final perception of an image by a viewer, and not just the photographer. If you have a strong composition, great light, compelling subject, and good exposure and focus, the difference among those format diminish. And probably to the point the viewer does not care or notice.

As photographers, we are conditioned to see things others cannot. This also leads to bias as that colors our vision--we know what to look for and use that information to make the judgement, even though those subtleties don't influence the overall image to any meaningful degree.

To get more to the point on the Fuji, given the largest printer most will practically use is a 44". Given my experience of using 44" printers, no one will be able to perceive the difference between a Fuji or Phase at this point. And once you get to 44" prints, bigger than that is not a factor either.

 

[glenerrolrd]

Physical size of the sensor does very much affect IQ ....when the sensors are between 36 -50 AND viewed on a iMac 5K . Put up any 47 MP FF file and do a side by side with the Fuji GFX 50 ,the HB X1D or the Leica S 007 . You an see the differences at normal viewing distance AND to my eye its substantial . Better color fidelity and tone separation come to mind .

Plenty has been written about the MF advantage (33x44 ) over FF ....and its not about printing large .

What I can not follow is the insistence that the only difference between the Fuji GFX 100 (33x44) and the much large Phase IQ4 150 is the ability to print BIG . Thats what I was reading on the thread about printing large .

Are we just hitting a point of diminishing returns ? Is 100MP on a 33x44 enough for most any application?

 

[Shashin]

Physical size of the sensor does very much affect IQ...

If you believe that, then you have your answer. The larger size of the Phase sensor will make a difference.

 

[Paul2660]

Paul
Good post to read

I have the Mac Pro 2013 with OWC 2TB SSD , 64GB Ram and the best vid cards available a few year ago when I bought it as a refurbd unit. I have never had the beach ball with my 100MP files that like you say get up to 3-4GB nor with my IQ4150 files which get even larger at times. Ci is very fast and never slows me down. Am I just lucky or is it another issue between our systems?

Not sure. Where I see the slow down is with the more advanced masking tools in a layer. The latest version of auto mask is excellent but can bog down for me at times. And when you combine it with a Luma fine tune it can get slow. But the results are worth it.

The layering C1 offers now is really vastly improved. And I tend to use it with all my images.

Normal rendering and or processing of files is fine.

Paul C

 

[glenerrolrd]

If you believe that, then you have your answer. The larger size of the Phase sensor will make a difference.

Will

What is wrong with having a point of view ? We all have different experience ,requirements ,perspectives to draw upon . What is lost in these discussions ..is that they are no longer discussions ....rather just a blabbering of “my requirements “ . No effort to try to see things from a different set of requirements .

What a nasty snide comment .

I am done with posting here ........

 

[Shashin]

Will

What is wrong with having a point of view ? We all have different experience ,requirements ,perspectives to draw upon . What is lost in these discussions ..is that they are no longer discussions ....rather just a blabbering of “my requirements “ . No effort to try to see things from a different set of requirements .

What a nasty snide comment .

I am done with posting here ........

Roger, I don't understand. You came here asking a question. I gave a thoughtful reply, which you seemed to pass over and simply reiterate your position. If you believe that sensor size is that important, and that is fine with me, and following your experience, then the larger Phase sensor will result in better quality. How is that "nasty"? I would encourage you to try the sensor.

And yes, we all have different experiences and perspectives. If you ask them to be shared, then allow for that diversity.

 

[PhiloFarmer]

For a veritable trove of information on sensor-size/pixel-pitch/aperture-diameter....see www.ClarkVision.com.

This dude knows his stuff...regardless of computer-monitor/computer-software/computer-OS..

Cheers,

Hank

 

[DougDolde]

But its easier to keep everything in focus with a smaller sensor there is greater depth of field. If sharpness all over is your goal it may be more difficult to attain with a larger sensor.

 

[flash]

I know I can make fabulous prints on my 24" printer with about 45MP. Doesn't matter where those MP come from. The difference between the sensor sizes is how much pushing and prodding I can do to a file before it starts to fall apart. At the same resolution files from a larger sensor are more robust. I can push my miniMF files a huge amount in post and still get pleasing results. I also think there's better colour retention at higher ISO's on the bigger sensor cameras I own (645Z, X1D and S007) versus the smaller format sensors I own (A7R3, S1R and Z7). I also like long exposure photography (usually between 30 seconds and 8 minutes). The miniMF sensor in my 645Z and X1D does spectacular long exposures while the A7R3 and S1R files are far inferior without LENR and somewhat so with it. It also helps that the X1D has the best long exposure implementation of any camera I've ever used.

For me and the way I shoot I appreciate the small advantages a larger sensor offers.

Gordon

 

[edouard]

... and don't forget the main quality factor: the analogical system before the sensor: the lens and its "projected" image over the "sampling" area (the sensor).
With a larger area you have larger details = you are using lower spatial frequencies that are easier to resolve (higher transmitted contrast) by the lens = better micro-contrast (before even being "sampled" by the sensor!)

in few words:
SHIFT TO LOWER SPATIAL FREQUENCIES

p.s. larger format lenses, as having to cover a larger image circle are more difficult to build and have likely less good transmitted contrast at equal spatial frequencies (that's why in my schema I shifted the blue - larger format lens - to the left compared to the pink - smaller format), but as the used spatial frequencies are lower the end result is better overall transmission for the larger format (as long as the reduced transmission by absolute frequency performance is not bigger than the gain from using lower frequencies).
So maybe a 44x33 with a bad lens is not better than 24x36 with a good lens, but as long you're using good lenses and a larger enough format, the image being sampled is better to begin with. Will this matter in the "final" image, that's another question ('depends on the image stytle)...

p.s. don't confuse my (absolute)-frequencies-vs-transmitted-contrast-mtf-schema with vendor mtf schema (distance-from-center-vs transmitted-contrast for different spacial frequencies and measurement directions) :