That Fuji GFX seems remarkably good…

[ErikKaffehr]

Imaging resource has comparable studio scene shots of the GFX, but also the Phase One IQ3100, the Pentax 645Z and other cameras. Now that Lightroom supports all of those I checked out the GFX and found it to have impressive detail. I tossed in the A7rII in the comparison as it is my presently most used camera.

Below are comparisons to the cameras uprezzed to IQ3100 image height:

• http://echophoto.dnsalias.net/ekr/Articles/FujiGFX/8708res_centre_detail.png
• http://echophoto.dnsalias.net/ekr/Articles/FujiGFX/8708res_corner_detail.png
• http://echophoto.dnsalias.net/ekr/Articles/FujiGFX/8708res_moire_shot.png
• http://echophoto.dnsalias.net/ekr/Articles/FujiGFX/8708res_whibal_shot.png

The next question I asked my self, what happens in a print 40" height. I would say that looking at 180PPI on screen closely corresponds to looking at large prints at close range. So I rescaled the images to 7200 pixels height. That mean upscaling to 7200 pixel for Sony, Pentax and Fuji and downscaling for the IQ3100.

http://echophoto.dnsalias.net/ekr/Articles/FujiGFX/7200res_centre.png
http://echophoto.dnsalias.net/ekr/Articles/FujiGFX/7200res_corner.png

Sharpening was done before scaling i both cases, sharpening after resizing would leave less artefacts.

So, what did I see?

• The Fuji GFX holds up remarkably well.
• It may be that the Schneider lens doesn't keep up with the sensor, but it seems to be well balanced with the sensor showing very little moiré.
• The Pentax 645Z lags behind the Fuji.
• The A7rII has the lowest resolution and it shows

The Schneider Kreuznach 80mm LS f/2.8 has a fairly typical optical design, a 6 element double Gauss without field flattener. That kind of lens is often used for "standard lenses". For large apertures a 7-th focal plane flattening element is often added.


The GFX uses a quiete complex lens design. The front group is still a traditional double Gauss, with a additional ED lens to control chromatic aberration, but the field flattener is a triplet in two groups. The lens focuses using the front group, so it control focal plane curvature using variable air space. I guess Fuji is serious when they say the lens is calculated for 100 MP sensor resolution:


That is not saying that the Fuji lens is better than the Schneider, double Gauss designs can be very sharp, but Fujifilm has invested a lot of design work in that 63 mm medium aperture lens.

Best regards
Erik

 

[tashley]

The next question I asked my self, what happens in a print 40" height. I would say that looking at 180PPI on screen closely corresponds to looking at large prints at close range.

I totally agree, it's a very useful exercise. I have noticed that the new Microsoft Surface Studio has almost exactly that native screen resolution and as soon as they release them in Europe I'm going to get one for exactly this sort of proofing. People so often think that they can judge print resolution from monitors with different resolution.

However, no amount of re-sizing will let you see this if your monitor doesn't have that native resolution. Pixels need to be mapped one to one for pixels - using zoom at more than 100% on a denser pixel monitor or using display scaling isn't a sufficiently useful approximation.

You should also be aware that that imaging resource test has been questioned by some as to focussing accuracy and alignment of image plane. I haven't looked into that but I have seen it discussed.

 

[ErikKaffehr]

Hi Tim,

I found that scaling the picture for 180 PPI print size and viewing at normal distance a 24" inch screen is a pretty good approximation for close view a large print. Computer screens are about 96 DPI, so screen viewed at say 60 cm distance is similar to looking at the print at 30 cm.

Regarding Imaging Resource studio scene, we need to keep in mind that the image plane is tilted, else it would not be possible to bring both flasks and background into focus. IR has arranged a set of resolution targets standing in front of the main subjects. There are two strips of them and they make it possible to judge the focus. Elevation obviously matters. Another good reference point is the "whibal" chart at the top of the image. If that chart is sharp. The plane of focus is the plane going trough the sharpest USAF target in the front.

The great unknown Is that the focal plane is almost never straight, on some lenses it can vary an awful lot (Hi S??y, do you listen?).

If a lens produces moiré that indicates it outresolves the lens, or the subject.

A third area to discuss is aperture choice. Almost any lens I tested was sharpest near the optical axis at f/4-f/5.6. Stopping down a sharp lens from f/5.6 to f/8 has a negative effect on sharpness. It may be that IR tests two stops down from maximal aperture.

It is a bit interesting that we are so obsessed with sharpness. Total sharpness may only exist in one single plane of focus and that plane is mostly not flat, except say landscape shots architecture shot at relatively large distances. DoF and diffraction may be the main limitation in many cases…

Best regards
Erik

I totally agree, it's a very useful exercise. I have noticed that the new Microsoft Surface Studio has almost exactly that native screen resolution and as soon as they release them in Europe I'm going to get one for exactly this sort of proofing. People so often think that they can judge print resolution from monitors with different resolution.

However, no amount of re-sizing will let you see this if your monitor doesn't have that native resolution. Pixels need to be mapped one to one for pixels - using zoom at more than 100% on a denser pixel monitor or using display scaling isn't a sufficiently useful approximation.

You should also be aware that that imaging resource test has been questioned by some as to focussing accuracy and alignment of image plane. I haven't looked into that but I have seen it discussed.

 

[tashley]

Hi Tim,
Computer screens are about 96 DPI, so screen viewed at say 60 cm distance is similar to looking at the print at 30 cm.

But sadly they aren't - they are all different these days. My MacBook Pro is about 220, my main 4K monitor is about 140. There is one Dell monitor that does 184 DPI but I don't like their colour space. The range is enormous.

Regarding Imaging Resource studio scene, we need to keep in mind that the image plane is tilted, else it would not be possible to bring both flasks and background into focus. IR has arranged a set of resolution targets standing in front of the main subjects. There are two strips of them and they make it possible to judge the focus. Elevation obviously matters. Another good reference point is the "whibal" chart at the top of the image. If that chart is sharp. The plane of focus is the plane going trough the sharpest USAF target in the front.

The great unknown Is that the focal plane is almost never straight, on some lenses it can vary an awful lot (Hi S??y, do you listen?).

If a lens produces moiré that indicates it outresolves the lens, or the subject.

A third area to discuss is aperture choice. Almost any lens I tested was sharpest near the optical axis at f/4-f/5.6. Stopping down a sharp lens from f/5.6 to f/8 has a negative effect on sharpness. It may be that IR tests two stops down from maximal aperture.

It is a bit interesting that we are so obsessed with sharpness. Total sharpness may only exist in one single plane of focus and that plane is mostly not flat, except say landscape shots architecture shot at relatively large distances. DoF and diffraction may be the main limitation in many cases…

Best regards
Erik

All agreed, thank you!

 

[D&A]

Tim Wrote: >>>"You should also be aware that that imaging resource test has been questioned by some as to focusing accuracy and alignment of image plane. I haven't looked into that but I have seen it discussed."<<<

I was one of the initial individuals that questioned the alignment of the focusing plane of Imaging Resources (IR) shot with the Pentax 645Z. I explained my observations in a different thread here on Getdpi as to why this was so and it appeared to me fairly obvious. In a nutshell, the lower portion of the IR test shot image used, it clearly showed that sharpness/resolution between the GFX and Pentax 645z was extremely close....yet the comparitive sharpness of upper regions of the same test image were dramatically different, where the Pentax 645z was quite soft and relatively unsharp. Whether this was due to misalignment of the Pentax 645z/lens in relationship to the test image or something else, its hard to say at this point and I would be speculating.

I echo Tim's assessment, great exercise Erik and very pertinent thoughts regarding sharpness and the pitfalls one can fall into along with the myriad of parameters one can alter/use in assessing sharpness of one camera/lens system vs. another. This is especially so in the assessment of a particular image file for determining and achieving the necessary resolution and sharpness required for a given sized large format print and unfortunately the monitor(s) used (especially their resolution as Tim pointed out), are sometimes or often the weak link. This is especially so if the resolution of the monitor is insufficient or incorrect ( again as Tim pointed out) in relationship to what's required for accurate assessment. I too have been interested in the Microsoft Surface Studio for the very reasons outlined, but with the desire to keep up with and upgrade all the bits and pieces in the entire chain of image reproduction, often makes this difficult at times.

Excellent thread including the info and comments contained within. Thanks!

Dave (D&A)

 

[DB5]

It truly is remarkably good. I'm very excited to see how the system develops. I think all other manufacturers must be feeling the heat at the moment.

Is it likely we will see bigger sensors from Fuji?

 

[ErikKaffehr]

Hi,

I don't think so. It doesn't make a lot of sense to design a lens system for a "small" sensor and than do it again for a "large" sensor. But Fuji has stated the lenses are designed for 100 MP, and I am pretty confident we are going to 70-100 MP sensor in the 44x33 mm form factor. And those pixel will be affordable…

Best regards
Erik

It truly is remarkably good. I'm very excited to see how the system develops. I think all other manufacturers must be feeling the heat at the moment.

Is it likely we will see bigger sensors from Fuji?

 

[D&A]

Hi,

I don't think so. It doesn't make a lot of sense to design a lens system for a "small" sensor and than do it again for a "large" sensor. But Fuji has stated the lenses are designed for 100 MP, and I am pretty confident we are going to 70-100 MP sensor in the 44x33 mm form factor. And those pixel will be affordable…

Best regards
Erik

I agree Erik but that's apparently what some manufacturers have done with regards to full frame 35mm vs. APS. Pentax for years until the digital SLR age had only full frame 35mm lenses and then with the introduction of their APS DSLR's at the time introduced a whole new line of lenses that for the most part (although not all), were designed with an image circle for APS. Then with their first entry of a full frame 35mm DSLR, it's apparent that now full frame lenses will be introduced.

Even with the Pentax digital 645 bodies with the 44x33 sensor, Pentax had a whole line of legacy full frame 645 lenses but decided to introduce new lenses with both a mix of image circles primarily for the 44x33 sensor size as well as other lenses that can be used on full frame medium format 645 sensors as well as 645 film. It's sometimes hard to know what these companies are thinking with regards to future development. Something tells me they more often than not only plan for the here and now (maybe not in the case of resolving power of new GFX lenses) and will think of future sensor sizes and their use and what's required when and if they get there. Maybe I am wrong but sometimes it seems its that way.

Dave (D&A)

 

[DougDolde]

Not sure I get this thing about 180 dpi on a monitor looking iike a real print.

If you have your screen resolution set in Photoshop, viewing at "Print Size" makes more sense to me.

Also most labs I've talked to say anything down to 150 dpi will make a good print on Fujiflex. Not sure about inkjet though my 36x48 prints from the IQ180 print great around 210 dpi

 

[Shashin]

Not sure I get this thing about 180 dpi on a monitor looking iike a real print.

If you have your screen resolution set in Photoshop, viewing at "Print Size" makes more sense to me.

Also most labs I've talked to say anything down to 150 dpi will make a good print on Fujiflex. Not sure about inkjet though my 36x48 prints from the IQ180 print great around 210 dpi

I have made 60" prints from native 24mp files (and scans from 35mm film) for museums on ink jet printers. They were beautiful. Also, from making 176" x 165" exhibition prints for museum exhibitions, I would say that pixel resolution is not a limit to print size, which is good news, because the science behind it says the same thing.

I would also say trying to judge print size from monitors is not very good as printers don't really print the pixels as displayed. Best to simply print a full width test strip.

 

[mark1958]

Today I had a chance to play with the macro lens. I shot some images at very high F stops and I was amazed that the degree of diffraction and image degradation is much less noticeable compared to what I would get with the Sony A7RII or Nikon D810. This is a real plus..

 

[ErikKaffehr]

Hi Doug,

It is quite a bit based on my own experience. In the old days, before we got 4K most monitors had a pixel pitch of 1/100 inch. That is a bit to coarse for near viewing, at least in theory. Human vision resolves about 180 PPI at 50cm / 20", you can find this information in a number of books, like Bruse Fraser's and Jeff Schewe's book on image sharpening. A good viewing distance for a 100 PPI screen would thus be 50 * 180 / 100 = 90 cm. So a 24" screen viewed at 90 cm would correspond to viewing an 180 PPI print at 50 cm.

This corresponds a bit to my experience.

Image A	Image B

If you check the crops above, both correspond to a 180PPI 33"x47" print. I would say that Image A is not very sharp while Image B is pretty OK. I have not made 33"x47" prints, but made 50% (linear) crops and printed in A2. Four A2 images would correspond to an A0 print.

So, what I have seen was that viewed close, say 50 cm I saw what I saw on screen Image B was much better than Image A. Moving back to around 1m looking distance the difference has disappeared, and it may be the Image A was perceived sharper. I have about 20/20 vision with corrective lenses, but vision obviously matters. I am near sighted, so taking of my glasses allows me to look a bit closer.

But all this applies to 24" screen with 1920x1200 resolution.

Another way to see it…

The size of my 24" monitor is about 20"x12.5" and it has 1920x1200 pixels. An IQ3 50MP image is 6208 pixels high, so using the vertical dimensions I can see 1200 / 6208 -> 0.1933 -> 19.33% of the image. The full image corresponding to an actual pixel view would be 64" x 103". So an actual pixel view corresponds to a very large print.

Best regards
Erik

Not sure I get this thing about 180 dpi on a monitor looking iike a real print.

If you have your screen resolution set in Photoshop, viewing at "Print Size" makes more sense to me.

Also most labs I've talked to say anything down to 150 dpi will make a good print on Fujiflex. Not sure about inkjet though my 36x48 prints from the IQ180 print great around 210 dpi

 

[tashley]

I have made 60" prints from native 24mp files (and scans from 35mm film) for museums on ink jet printers. They were beautiful. Also, from making 176" x 165" exhibition prints for museum exhibitions, I would say that pixel resolution is not a limit to print size, which is good news, because the science behind it says the same thing.

I would also say trying to judge print size from monitors is not very good as printers don't really print the pixels as displayed. Best to simply print a full width test strip.

I agree with the test strip but:

1) you can of course print to lower resolutions than 180 but in practice, for me, unless the original file has been well up-resed and then had further work done, 180 is the lowest I would go in order to keep the impression of high quality at close range. iPhone billboards are made from 12mp captures which are processed very carefully and viewed from quite a distance but for a fine art print in a gallery context, 180 is the limit for me - unless the subject matter is naturally hazy, blurry or otherwise intentionally indistinct - in which case it can look gorgeous. High resolution is not by any means always integral to the finished work but sometimes it is. So...

2) choice of resolution is part of the final work itself - part of the artisit's creative intent. Ming Thein makes 'ultra prints' at well over 1000DPI and they will have a different feel, which is integral to the work.

3) all scaled screen modes other than 100% will involve re-mapping and / or interpolation. 100% screen size is the best if not the only way to see the closest you can see to what your print will look like if you print to that same resolution. It's not the end of the world if your screen is 220 PPI and you print to 360 or 180 DPI but for judging sharpness, detail and acuity you can't beat printing to the same resolution as your monitor and proofing at 100% - unless you make a test strip. But you can't pan around a test strip.

4) the above 180DPI guideline assumes a combination of lens, sensor and capture technique that creates a capture with a COC no worse than about twice pixel pitch. If you want a sharp print and the most detailed part of your frame doesn't match that criteria, you'll need to print smaller unless you're creatively happy with a slightly fuzzy look.

5) finally (phew!) there's been a whole lot of work done that strongly implies that printers, because of the way they dither, are better at up-resing than computers are for those images where you want to print larger. In my experience that's true but some people feel otherwise.

 

[D&A]

Tim, well said (your post "above") and it closely matches my experience, especially when it comes to accurately assessing a files sharpness prior to printing large.

Dave (D&A)

 

[Shashin]

Tim and Dave, naturally you can apply any personal criteria to your work. Print size, or more accurately, viewing distance, does change perception of an image. But to take personal criteria and state that a print has to be x, y, or z is as meaningful as saying a photograph should have a specific contrast, sharpness, or color. My recommendation is print the size you want and take it from there. Having good printing skills or a good printer helps. I have had so many artist and curators come to me and say an image can only be x larger. I ask them to humor me and let me try a much larger print, only to have them reevaluate their criteria.

I think we are essentially in agreement.

 

[Shashin]

Hi Doug,

It is quite a bit based on my own experience. In the old days, before we got 4K most monitors had a pixel pitch of 1/100 inch. That is a bit to coarse for near viewing, at least in theory. Human vision resolves about 180 PPI at 50cm / 20", you can find this information in a number of books, like Bruse Fraser's and Jeff Schewe's book on image sharpening. A good viewing distance for a 100 PPI screen would thus be 50 * 180 / 100 = 90 cm. So a 24" screen viewed at 90 cm would correspond to viewing an 180 PPI print at 50 cm.

This corresponds a bit to my experience.

Image A	Image B

If you check the crops above, both correspond to a 180PPI 33"x47" print. I would say that Image A is not very sharp while Image B is pretty OK. I have not made 33"x47" prints, but made 50% (linear) crops and printed in A2. Four A2 images would correspond to an A0 print.

So, what I have seen was that viewed close, say 50 cm I saw what I saw on screen Image B was much better than Image A. Moving back to around 1m looking distance the difference has disappeared, and it may be the Image A was perceived sharper. I have about 20/20 vision with corrective lenses, but vision obviously matters. I am near sighted, so taking of my glasses allows me to look a bit closer.

But all this applies to 24" screen with 1920x1200 resolution.

Another way to see it…

The size of my 24" monitor is about 20"x12.5" and it has 1920x1200 pixels. An IQ3 50MP image is 6208 pixels high, so using the vertical dimensions I can see 1200 / 6208 -> 0.1933 -> 19.33% of the image. The full image corresponding to an actual pixel view would be 64" x 103". So an actual pixel view corresponds to a very large print.

Best regards
Erik

Crops can really be a horrible way to evaluate print sharpness as there is no context. At least work from a full width crop.

 

[D&A]

Tim and Dave, naturally you can apply any personal criteria to your work. Print size, or more accurately, viewing distance, does change perception of an image. But to take personal criteria and state that a print has to be x, y, or z is as meaningful as saying a photograph should have a specific contrast, sharpness, or color. My recommendation is print the size you want and take it from there. Having good printing skills or a good printer helps. I have had so many artist and curators come to me and say an image can only be x larger. I ask them to humor me and let me try a much larger print, only to have them reevaluate their criteria.

I think we are essentially in agreement.

Just some thoughts. A image and a subsequent print of that inage is very subjective, both from the standpoint of the emotional inpact the print makes as well as the technical aspects. Opinions will of course vary and additionally on the technical side, whats acceptable to one may not be acceptable to another. Of course when an image has emotional impact for a viewer, regardless of the subject matter, it generally tends to superceed whatever technical deficiencies the image might have.

In my opinion, what Erik and Tim have expressed with regards to effectively judge and inage file on a monitor with regards to judging sharpess thats sufficent to print at a given size, is a good guideline from my experience but as you say, its not the be all to end all....simply a relatively good place (and a fairly accurate one) to start with. In my case this is especially so when ink costs are high and I would prefer to get relatively close on the 1st full test print as opposed to simply judge by repetitive printing at that substatial size.

Good thoughts all around how there are many ways to achieve a similar objective.

Dave (D&A)

 

[Vivek]

That Fuji GFX seems remarkably good…

Not that good enough to place an order immediately? :grin:

 

[Shashin]

Just some thoughts. A image and a subsequent print of that inage is very subjective, both from the standpoint of the emotional inpact the print makes as well as the technical aspects. Opinions will of course vary and additionally on the technical side, whats acceptable to one may not be acceptable to another. Of course when an image has emotional impact for a viewer, regardless of the subject matter, it generally tends to superceed whatever technical deficiencies the image might have.

In my opinion, what Erik and Tim have expressed with regards to effectively judge and inage file on a monitor with regards to judging sharpess thats sufficent to print at a given size, is a good guideline from my experience but as you say, its not the be all to end all....simply a relatively good place (and a fairly accurate one) to start with. In my case this is especially so when ink costs are high and I would prefer to get relatively close on the 1st full test print as opposed to simply judge by repetitive printing at that substatial size.

Good thoughts all around how there are many ways to achieve a similar objective.

Dave (D&A)

Dave, nothing really to disagree with. But I would like to add experience. The comment about Erik's crop is also one of experience. Those seeking guidance are also those with the least experience. They may not realize the cognitive limitation to using a crop to judge quality. I used to teach students large-format printing. Thinking a crop would save a lot of ink and paper (especially since one student was making full 40" test prints and blew through nearly $1,000 worth of paper using sRGB as the printer profile and was having a quality issue, go figure), I taught them to do that. I soon discovered that they were not making great judgements. I ended up teaching full-width crops and that solved lots of problems. It also saved lots of paper.

 

[ErikKaffehr]

Hi Will,

Thanks for an interesting comment!

I have A2 as standard print size, larger prints are less convenient.

Best regards
Erik

Dave, nothing really to disagree with. But I would like to add experience. The comment about Erik's crop is also one of experience. Those seeking guidance are also those with the least experience. They may not realize the cognitive limitation to using a crop to judge quality. I used to teach students large-format printing. Thinking a crop would save a lot of ink and paper (especially since one student was making full 40" test prints and blew through nearly $1,000 worth of paper using sRGB as the printer profile and was having a quality issue, go figure), I taught them to do that. I soon discovered that they were not making great judgements. I ended up teaching full-width crops and that solved lots of problems. It also saved lots of paper.