Fuji GFX100 DPReview

[algrove]

100MP @5fps and IBIS and decent autofocus is a street shooters weapon par excellence .

You are welcome to do that with the 100MP GFX.

However, the 50R is about the max for this street shooter who normally uses Leica M & Q gear in that mode.

I would post a couple of images recently taken in Venice here but most are Leica Q images and do not belong in this Fuji thread.

 

[GrahamWelland]

I agree - 100mp handheld even with IBIS just sounds like beating yourself with your own camera. Stick that sucker on a tripod to get the full Dante of resolution.

Graham

You are welcome to do that with the 100MP GFX.

However, the 50R is about the max for this street shooter who normally uses Leica M & Q gear in that mode.

I would post a couple of images recently taken in Venice here but most are Leica Q images and do not belong in this Fuji thread.

 

[Shashin]

However, the 50R is about the max for this street shooter who normally uses Leica M & Q gear in that mode..

Yeah, I am not that interested in more pixels, just better ones. But even that is starting to have diminishing returns. I printed with a 42" printer and 50MP files are just idling at that size. Naturally, that will just up the pot on Dante's forum--you are just going to have to buy a bigger house to display all those large prints.

 

[Shashin]

I agree - 100mp handheld even with IBIS just sounds like beating yourself with your own camera. Stick that sucker on a tripod to get the full Dante of resolution.

Graham

Only if if it has pixel shift, otherwise you are just playing...

:toocool:

How big is a 400MP 16-bit file?

 

[Gerd]

..........

How big is a 400MP 16-bit file?

estimated - about ~ 600 MB

Greeting Gerd

 

[GrahamWelland]

Naturally, that will just up the pot on Dante's forum--you are just going to have to buy a bigger house to display all those large prints.

You're going to need a bigger boat ... oops ... house! (Thank you Jaws!).

I agree I bought an image/print shelving system to handle this. However, even this has its limits!

In the end, as I've found with my IQ3100, it's all about better (finer) tonality from more pixels than just pure resolution. The GFX 50S delivers great pixels and I have no reason to believe that the GFX100 won't also deliver just as well. :thumbup:

 

[Boinger]

You're going to need a bigger boat ... oops ... house! (Thank you Jaws!).

I agree I bought an image/print shelving system to handle this. However, even this has its limits!

In the end, as I've found with my IQ3100, it's all about better (finer) tonality from more pixels than just pure resolution. The GFX 50S delivers great pixels and I have no reason to believe that the GFX100 won't also deliver just as well. :thumbup:

Well one of the big perks of more resolution is more color data. That is why I think you see the tonality difference.

It is why the foveon sensor from sigma looks so much different from traditional bayer cameras.

In a sense there is less interpretation being done.

Even on a smallish print like an 8x10 my wife somehow liked the stuff I shot on medium format better than the ones I shot on small format. She didn't know what was shot on what she just thought some images had better depth. Take that for what you will, but to me it seems to make a difference.

 

[PeterA]

You are welcome to do that with the 100MP GFX.

However, the 50R is about the max for this street shooter who normally uses Leica M & Q gear in that mode.

I would post a couple of images recently taken in Venice here but most are Leica Q images and do not belong in this Fuji thread.

I dont care what camera is used to make good images....and what thread they are posted in..it is the photographer not the camera - every single time...
I've explained my rationale for 100MP and in particular the way Fuji (and the XID does btw) allows for different aspect ratios in other threads- no need ot repeat the obvious here...

everyone is entitled to their view on anything re what os 'appropriate' but a lot of people cant walk the talk - very funny.

 

[GrahamWelland]

Well one of the big perks of more resolution is more color data. That is why I think you see the tonality difference.

It is why the foveon sensor from sigma looks so much different from traditional bayer cameras.

In a sense there is less interpretation being done.

Even on a smallish print like an 8x10 my wife somehow liked the stuff I shot on medium format better than the ones I shot on small format. She didn't know what was was shot on what she just thought some images had better depth. Take that for what you will, but to me it seems to make a difference.

I absolutely agree.

 

[mristuccia]

Well one of the big perks of more resolution is more color data. That is why I think you see the tonality difference.

It is why the foveon sensor from sigma looks so much different from traditional bayer cameras.

In a sense there is less interpretation being done.

Even on a smallish print like an 8x10 my wife somehow liked the stuff I shot on medium format better than the ones I shot on small format. She didn't know what was shot on what she just thought some images had better depth. Take that for what you will, but to me it seems to make a difference.

Sorry I don't get this. We are always in presence of a bayer CFA here, meaning that for each pixel two of the three color components are still artificially constructed by the demosaiking algorithm.
I would rather associate color richness to fat pixels than to more megapixels. Of course both conditions can happen together, but this is not the case of the new Fuji GFX100, which will have more pixels for the same sensor size.

 

[Stuart Richardson]

More megapixels means more "guesses", and the more efficiently the bayer matrix can resolve the image. So for any given area of the image, you have a finer ability to resolve detail and color...more megapixels means better color differentiation, less moire and so on. Of course, fat pixels mean more light gathering efficiency and a higher signal to noise ratio, but all else being equal, more megapixels also means better color discrimination and therefore better tonality.

Think of it this way. If you only had four giant pixels, a photo would always be a single color. The more pixels you add, the more nuanced and accurate the picture could be.

 

[Gerd]

Sorry I don't get this. We are always in presence of a bayer CFA here, meaning that for each pixel two of the three color components are still artificially constructed by the demosaiking algorithm.
I would rather associate color richness to fat pixels rather than to more megapixels. Of course both conditions can happen together, but this is not the case of the new Fuji GFX100, wich will have more pixels for the same sensor size.

The same applies to the IQ4 150. The IQ4 150 has exactly the same pixelpitch as the GFX 100S of 3.74 micron. The IQ3 100 has 4.6 micron on the same surface as IQ4 (54x40mm). Take a look at the differences between IQ3100 (Tr.) and IQ4 150. And maybe they get an idea how it works.

Greeting Gerd

 

[Boinger]

Sorry I don't get this. We are always in presence of a bayer CFA here, meaning that for each pixel two of the three color components are still artificially constructed by the demosaiking algorithm.
I would rather associate color richness to fat pixels than to more megapixels. Of course both conditions can happen together, but this is not the case of the new Fuji GFX100, which will have more pixels for the same sensor size.

You are correct, we are always in the presence of the Bayer filter. But as the pixels get smaller it does improve the color response.

Think of it like this:

Instead of thinking of pixels think of the image as an area so each small area of a picture now is getting more sample rgb values of light. More sampling equals more color accuracy.

The difference is subtle but it is there.

 

[mristuccia]

More megapixels means more "guesses", and the more efficiently the bayer matrix can resolve the image. So for any given area of the image, you have a finer ability to resolve detail and color...more megapixels means better color differentiation, less moire and so on. Of course, fat pixels mean more light gathering efficiency and a higher signal to noise ratio, but all else being equal, more megapixels also means better color discrimination and therefore better tonality.

Think of it this way. If you only had four giant pixels, a photo would always be a single color. The more pixels you add, the more nuanced and accurate the picture could be.

Thank you for the explanation Stuart,

now I can better understand what you meant before. That's about more color nuances in the smallest details, which means more color resolution.

The foveon sensor on the other side tries to reach color accuracy the right way: by not having to interpolate (aka invent) colors at all for any pixel.

So you are basically saying that in order to reach the color purity of a (reference) foveon sensor with a bayer one the latter must heavily outresolve the former (by four times I would say). Something like this?

 

[Stuart Richardson]

Thank you for the explanation Stuart,

now I can better understand what you meant before. That's about more color nuances in the smallest details, which means more color resolution.

The foveon sensor on the other side tries to reach color accuracy the right way: by not having to interpolate (aka invent) colors at all for any pixel.

So you are basically saying that in order to reach the color purity of a (reference) foveon sensor with a bayer one the latter must heavily outresolve the former (by four times I would say). Something like this?

Well, yes, the Foveon theoretically has this advantage, but the bayer matrix demosaicing algorithms are really really good by now, and there are so many pixels...think about it, for every given photograph, divide that photo into 24-150 million samples. If you take any given tiny area of a photo, you will have many different samples. The bayer works by looking at all the neighboring pixels' responses, and then coordinating. So it is not like a green pixel only reads green. If a green pixel is totally surrounded by red pixels, the algorithm will turn that green pixel red. The key here is the math and the sample size. A sensor like a Foveon would have a huge advantage in a low resolution environment, but with the sheer brute force number of samples in bayer algorithms, along with the sophistication of the demosaicing math, you wind up with very high accuracy. If Foveon really was that much better, it would have saturated the market by now. This is not to say it does not have advantages, but overall, I think the color accuracy is likely more on the camera maker and conversion software than on the underlaying sensor technology.

 

[mristuccia]

Well, yes, the Foveon theoretically has this advantage, but the bayer matrix demosaicing algorithms are really really good by now, and there are so many pixels...think about it, for every given photograph, divide that photo into 24-150 million samples. If you take any given tiny area of a photo, you will have many different samples. The bayer works by looking at all the neighboring pixels' responses, and then coordinating. So it is not like a green pixel only reads green. If a green pixel is totally surrounded by red pixels, the algorithm will turn that green pixel red. The key here is the math and the sample size. A sensor like a Foveon would have a huge advantage in a low resolution environment, but with the sheer brute force number of samples in bayer algorithms, along with the sophistication of the demosaicing math, you wind up with very high accuracy. If Foveon really was that much better, it would have saturated the market by now. This is not to say it does not have advantages, but overall, I think the color accuracy is likely more on the camera maker and conversion software than on the underlaying sensor technology.

Yes, I'm aware of the foveon problems and drawbacks. I took it as reference only because you mentioned it as well in your original post.
Moreover, I know how a demosaicing algorithm works, I'm also developing one for my specific purposes.

I'm simply more inclined to think of color accuracy in terms of avoiding demosaicing at all rather than stepping up the spacial sampling frequency (aka more mpxl). But I understand your point well, and it seams reasonable.

Thanks.

 

[Boinger]

Yes, I'm aware of the foveon problems and drawbacks. I took it as reference only because you mentioned it as well in your original post.
Moreover, I know how a demosaicing algorithm works, I'm also developing one for my specific purposes.

I'm simply more inclined to think of color accuracy in terms of avoiding demosaicing at all rather than stepping up the spacial sampling frequency (aka more mpxl). But I understand your point well, and it seams reasonable.

Thanks.

Well you could also think of it this way once you reach a high enough pixel density you would in theory not even need a Bayer sensor as the rgb array could in theory produce the color tones much like our computer display does. {this would work if there were equal rgb diodes on the sensor}

That is in essence what is happening with high np counts even in the 35mm world as pixel densities increase the differences are becoming smaller.

Foveon never caught on due to licensing and sigma owning the patents. I don't think they played well with others. But capturing a full color signal is the best way to truly emulate what film did.

I do think eventually we will get tech that is able to capture a full color signal thus eliminating the Bayer array.

 

[Satrycon]

....and more bluetooth

https://launchstudio.bluetooth.com/ListingDetails/79253

 

[SrMphoto]

....and more bluetooth

https://launchstudio.bluetooth.com/ListingDetails/79253

This means that we'll be able to efficiently geotag images in the GFX100 (which we cannot with GFX50S).

 

[Quentin_Bargate]

Well I'm here in Miami at the Yacht Show, with my GFX 50S and two lenses, the 32-64mm F4 and a Canon 24mm TS-E which I used to take the attached shot, 5 images stitched, all about 10 secs exposure, then blended using PTGui, all quickly done as I am on a laptop in a hotel room (I will reprocess when back home in tomorrow). Point is I am more than happy to lug the 50S around with two lenses in a camera bag, and wonder if I will do the same with a GFX 100.

I note the 100 has a two battery grip built in which is great for studio work, but why impose such a weight and size burden on everyone? I would have preferred a removable battery grip as on the 50S.


Of course I am very interested in the GFX100 and will almost certainly want to buy one when it comes out, but I doubt I will feel so sanguine about lugging that beast to Miami or anywhere else if weight is an issue (which it is).