[UPDATE] Fujifilm GFX 100 added to studio test scene

Mechanical first curtain causes shutter shock. Mechanical rear curtain (EFCS) causes a very very slight amount shake, probably from its initiation. Electronic shutter gives the sharpest results.

Jun 19, 2019permalink

sobermeier

I don't understand why dpreview has released the test shots by the fact they are out of focus. This is not good for Fuji as well as dpreview - So this article is nothing more than clickbait.

Jun 19, 2019permalink

They didn't know they were sub-optimally focused when they published the article. To be fair, how would they know? They presumably use their standardized methods and published it ahead of the review as per normal. The only way to be sure that focus is sub-optimal is to produce a result which is better, which clearly they didn't manage to do in the first iteration of testing. It's only when you compare them carefully to samples from other cameras that it starts to look likely that there was a problem. Having realised the results are sub-optimal, they have acknowledged that and committed to re-testing. Absolutely fair enough as far as I'm concerned.

Jun 19, 2019permalink

Of course you never make mistakes or dont stand by rhem?
Oh and sure you paid a million per post so you get to expect perfection every time.
It is diligent of dpr to stand by a mistake made and correcting them

Jun 19, 2019permalink

sobermeier

I would simply remove it from the page. I bet 90% of the audience here are pixel fetishists and it was obvious when you see the results, that something went wrong here - And a 100mpx MF camera is all about detail and nothing else. I noticed the lag of detail at first sight and I'm not a pixel peeper.

And you know how the internet works - some idiot will download the file, up it into any photographers forum with the subline "look how bad the Fuji is" simple as that.

Jun 19, 2019permalink

https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=daylight&attr13_0=fujifilm_gfx100&attr13_1=fujifilm_gfx50r&attr13_2=panasonic_dcs1r&attr13_3=nikon_z7&attr15_0=raw&attr15_1=raw&attr15_2=raw&attr15_3=raw&attr16_0=100&attr16_1=100&attr16_2=100&attr16_3=100&attr126_0=4&attr126_1=4&attr126_2=2&attr126_3=4&normalization=full&widget=687&x=0.3192994857564471&y=-0.07887191732732605

It’s fixed - no need to let ‘the internet’ hold em down

Jun 19, 2019permalink

Eric Calabros

With such high resolution even a small error in focus will be noticeable.

Jun 19, 2019permalink

vscd

Just check the DP1 Merrill in the green grass... with 15 MPixel and only $400 bucks it's sharper and has somewhat more detail than this large CMOS sensor.

Jun 19, 2019permalink

Flashback

Good point, interesting. . .

Jun 19, 2019permalink

I owned a DP2 and never should have sold it.
Should come with a tripod though

Jun 19, 2019permalink

vscd

I still have my DP3M... and I'm shocked how well this cam works in the studio. Useless with anything else than ISO100, but if the light is there you get stunning results. (The green/purple tint sometimes gives problems... but turn it to b/w and you get fine art quality)

Jun 20, 2019permalink

left eye

I'd like to see results using the EFCS.

Jun 19, 2019*permalink

They are ever so slightly softer at the longer shutter speeds. Not enough to make much of a real world impact, I would guess (though we haven't tested all shutter speeds to check for the difference between EFCS and ES).

Jun 19, 2019permalink

GoodKnight

Considering the focus issue these look very nice.

Jun 19, 2019permalink

fPrime

My money goes to the Phase One IQ180 here. It's vastly superior color over the GFX 100, S1R, and 50R trump any slight loss in resolution.

Jun 19, 2019permalink

brendon1000

It used to be slightly more expensive I believe. :)

Jun 19, 2019permalink

fPrime

Good point. But looking at eBay now it seems that mint used copies of the IQ180 are actually less expensive than a new GFX 100. Tempting.

Jun 19, 2019*permalink

brendon1000

Where did you find a phase IQ 180 on ebay? I did a search and came up with zero results. Were you searching on a local ebay?

Jun 20, 2019permalink

km25

To be honest. The Fuji does very well, even at high ISO. All of the cameras did well, I compared A7r111, D850 , Sr1R and the Fuji. Looking at the camera and the lens and how you would use these camera is also very important.

Jun 19, 2019permalink

sharpbokeh

never thought i'd be asking this of fuji, but does this mean the lens is the limiting factor (when compared with the g50*)

Jun 19, 2019permalink

No, it's the slight miss-focus, as DPR says in the big green note under the comparison :)

Jun 19, 2019permalink

left eye

and BSI is known to be slightly softer due to cross-talk.

With traditional CMOS the photodiodes are deeper in a 'well', surrounded by wiring, in effect acting as a lens hood. With the 'wiring hood' set below, BSI allows more light to reach the diodes, but also stray light from neighbouring microlenses.

Jun 19, 2019permalink

@left eye ... yes, at least in theory, but that is not what we see here.
Look at cameras that have added BSI and there in not a notable difference even at much smaller pixels. For example compare RX100 and RX100 mk2 (where BSI is introduced) and mk4 (where stacked BSI is added) - i struggle to see any difference - if anything the newer models are a tiny bit sharper.

Jun 19, 2019*permalink

sharpbokeh

ohhh - did i have to read stuff? geez what's next

Jun 19, 2019permalink

Kim Yee

Less moire than the 50r, did they stop using those smaller pixel microlens? It feel smoother, but not as crisp or sharp.

Jun 19, 2019permalink

left eye

With the smaller/cramped pixel-pitch I suspect the micro-lenses are now full-size (or near), to gather as much light as possible. Yes, the GFX 50S/R perceptually resolves higher than 51MP due to the reduced fill micro-lenses (not just from the studio scene here but from my daily work with the GFX 50S) - the 50S/R is quite prone to moire colour-artefacts as a result though.

Jun 19, 2019permalink

Kim Yee

Yes, Initially I thought that was a good idea until I notice the amount of moire It generate quite a lot of moire in city shots etc... But ohh it feel sharp.... really really sharp....

Jun 19, 2019permalink

Vahab

I hope my professor comes up with such a follow-up regarding my grades.

Jun 19, 2019permalink

fmian

Medium format bayer is fInally catching up to the APSc Sigma cameras. Excellent.

Jun 19, 2019permalink

GoodKnight

Got to wonder what a medium format Foveon would look like

Jun 19, 2019permalink

Mr Bolton

In what way do you mean? (Not snark, I'm genuinely curious..)

Jun 19, 2019permalink

Twigwater

Yeah, amazing performance for APSC.

Jun 19, 2019permalink

Yes, that Foveon sensors are damn amazing... but unfortunately for the tech only at base ISO. However, imagine a 44x33 foveon sensor :)) ... that would be a landscaper's dream. Nothing could touch it for a while...

Jun 19, 2019permalink

fmian

@Mr Bolton - clarity, edge contrast and resolving power lends to give the images more depth. At least that's what I'm seeing. Most of the test chart/display is flat, but have a look at the 3 dimensional objects like the bottles on the bottom and the green thick leaves here and there. Test shots from the foveon just seem more realistic.
Although yes, the colour accuracy and ISO performance is lacking.

Jun 19, 2019*permalink

sh10453

Knowing that tons of your readers never read disclaimers, even in bright green, evident by lots of the comment, it probably was not a good idea to publish this report.
There is still time, and nothing is wrong in taking it out; you already admitted some errors, and I think it would be prudent, and brave, to totally remove this unfair report.
I'm sure Fuji is not thrilled.

Jun 19, 2019permalink

adegroot

I second that.

An out of focus shot is not representative;
and they should have also shot the target with autofocus and then they could have compared the results.

To manually focus a 100MP camera, you have to be super accurate, which apparently is difficult to achieve. Makes you think twice about using this camera with manual focus.

Jun 19, 2019permalink

adegroot: "Makes you think twice about using this camera with manual focus." It shouldn't because the error may be nothing to do with the camera. The most likely cause of error with manual focus is human error. All the camera does is to provide a set of tools and controls to help the user make a choice. Despite this, DPR still choose manual focus over AF for this job because it has proven to be more reliable. So using AF won't help. Don't forget that in the real world, most subjects are three-dimensional and this tiny amount of error will just mean that a slightly different part of your subject will be perfectly in focus, which in many cases won't matter. Getting perfect focus reliably with any camera is difficult - and the higher the resolution, the harder it becomes because inaccuracies are more visible. Does that mean that it's not worth having higher resolution cameras? No.

Jun 19, 2019permalink

"Don't forget that in the real world, most subjects are three-dimensional and this tiny amount of error will just mean that a slightly different part of your subject will be perfectly in focus, which in many cases won't matter."

This.

Jun 19, 2019permalink

sh10453

I understand, but still, why's the rush to add a non-production camera to the studio test, especially a flagship camera like this?

Jun 19, 2019permalink

LoneTree1

Looks very good, but Panasonic's high resolution mode is the top of the heap right now.

Jun 19, 2019permalink

Doesn't look good, DPReview just wrote the GFX 100's studio scene shot is blurry because they didn't focus well... Wait the have the studio scene redone...

By the way, the GFX 100 will also get its hi-res mode

Jun 19, 2019permalink

LoneTree1

Do it backwards. Make a 5mp FF capable of extreme low-light shooting and have a hand-held high resolution mode of 20mp if needed. When global sensors appear.

Jun 19, 2019permalink

Mr Bolton

Olympus' M43 secret weapon ;-)

Jun 19, 2019permalink

Tony Bonanno

Why would you publish this studio test scene knowing that the camera had a focus issue ?? A more responsible and ethical decision on the part of your editors would be to wait until you had a production version. By your own admission, this camera was not representative of the camera's IQ.

Jun 18, 2019*permalink

Terrible Photographer

RIGHT??

Jun 19, 2019permalink

Scott Everett

To be clear, the camera didn't have a focusing issue, we did. We do technical testing on pre-production cameras with the permission from manufacturers only when they advise us that technical performance will not change with final firmware. In this case, because we are about to receive a production camera imminently, we are planning to re-shoot the test with that camera.

Jun 19, 2019permalink

Tony Bonanno

OK, thanks for clarifying.. I still wouldn't have posted it.. but hopefully you'll have a new test scene soon.

Jun 19, 2019*permalink

Scott Everett

Agreed, it was a mistake that we missed, hence the update to the story.

Jun 19, 2019permalink

Jim PDX

So what mistake did you make? Is that at least news? I would hate to have this conversation with Fuji right now if this wasn't remotely their fault...

(Disclaimer probably should be the first thing you read...)

Jun 19, 2019permalink

@Scott Get another copy of the 120mm lens for the new studio scene shots.... It's not normal to have some parts of the scene in focus (the bearded man) and others not...

Jun 19, 2019*permalink

The angry ppl on the internet. Geez

Jun 19, 2019permalink

ClosePhoto

So the S1R with pixel shift looks better for static scenes....hmmm.

Jun 18, 2019permalink

A Owens

When the Fuji is mis-focused. Did you clock the update?

Jun 19, 2019permalink

There are regions of the images in good focus for the GFX100 (see the bearded man drawing). And the S1R pixel-shift shows more resolution, but also seems a bit less "natural" ... maybe a bit of over-sharpening in merging the images or something, but anyway, clearly more resolution.

However, while interesting for academic discussion or for some specific work, comparing a pixel-shift mode (that requires use of tripod) with a single exposure is unfair. After all, with any camera you can take more shots and combine them in post to obtain better IQ.

Jun 19, 2019permalink

What happened with this studio scene? I think the lens fell, got damaged and therefore we have some parts in focus (the bearded man) and others not...

This studio scene should not have been posted

Jun 18, 2019*permalink

John Crowe

Colour is way off. Oh, someone already said that. I would pull these down until the actual camera comes out, since they are not in focus anyway. Seems to be doing the camera a disservice since I know IQ will be unmatchable.

Jun 18, 2019permalink

chrissf

As a sony user myself I say the GFX 100 has better iso at pretty much any setting.
It looks like 1 stop better than GFX 50 and Sony A9 and Sony A7 III
Finally the GFX is able to utulize its bigger sensor.

Comparison at normal print size iso 6400 vs 3200 on the other cameras
https://www.dpreview.com/reviews/image-comparison/fullscreen?attr18=daylight&attr13_0=fujifilm_gfx100&attr13_1=fujifilm_gfx50r&attr13_2=sony_a9&attr13_3=sony_a7iii&attr15_0=raw&attr15_1=raw&attr15_2=raw&attr15_3=raw&attr16_0=6400&attr16_1=3200&attr16_2=3200&attr16_3=3200&attr126_0=4&attr126_1=4&attr126_2=4&attr126_3=4&normalization=print&widget=687&x=0.309523156050887&y=-0.05177076289319863

Jun 18, 2019permalink

Rootbeer Man

🤔 🤔

Jun 18, 2019permalink

sneakyracer

Incredible that the PhaseOne IQ180, released 8 years ago is still tops in color and holds its won in resolution. Wow.

Jun 18, 2019permalink

Wait until DPReview redo the studio scene...

Jun 18, 2019permalink

raymondg

I have had the pleasure of using the Phase One XF/IQ180 as well as the 100MP Back and all I can say is that it totally spoiled my opinion on cameras/sensors. There is something quite unique about the Phase One systems that I have not observed with any others. The photographers I personally know who use Phase all produce stunning photos, in part to their skill, patience, creativity, but also to the gear.

Jun 19, 2019permalink

s1oth1ovechunk

Except vs the Panasonic hi res mode...

Jun 19, 2019permalink

The GFX 100 will also get its hi-res (or pixel shift) mode

Jun 19, 2019permalink

s1oth1ovechunk

In your dreams for sure but only a remote possibility in reality.

Jun 19, 2019permalink

They wrote that in "Fujirumors" (they usually are right), and as the camera has IBIS it's not really a "remote possibility"...

Jun 19, 2019permalink

Azimuth Jones

The most interesting thing about this to me is that if shooting product shots or still lifes like this one, the Panasonic G9 in high-resolution mode (a $1200 camera costing an eighth of the Fuji 100MP camera) gets you in the very same ballpark. For macro or still lifes (and constant lighting) the Panasonic seems to be the price-performance champion.

Jun 18, 2019permalink

Kevin Barrett

Maybe the parking lot of the ballpark? The G9 is a great little camera that pushes the envelope of what m43 can do, while the GFX 100 redefines the envelope for MF

Jun 18, 2019*permalink

ClosePhoto

To me it's that the S1R is better than the GFX 100 for static shots.

Jun 18, 2019permalink

Terrible Photographer

S1R is better, G9 is in the same county as that ballpark, GFX definitely does not "redefine the envelope for MF"

Jun 19, 2019permalink

The GFX 100 will also get its hi-res (or pixel shift) mode

Jun 19, 2019permalink

Terrible Photographer

Like the Imacons from 15 years ago? Still not “redefining” medium format

Jun 19, 2019permalink

Laybourne

You people are in desperate need of some AI/software to automatically iterate the focus until it is sharpest.

Isn't such a thing available?

Jun 18, 2019permalink

A better feature is called Autofocus, you select the desired focus point and the camera focuses automatically

Jun 18, 2019permalink

Except that AF isn't reliable when you get down to the tolerances necessary for this purpose.

Jun 19, 2019permalink

MikeRan

DPR, why don’t you autofocus and choose th sharpest of, say 3-5 images? If it’s so hard to manual focus a static scene, we shouldn’t expect better results than what we see here. If the AF is inconsistent, and MF aids stink, then why does anyone need 100MP?

Jun 18, 2019permalink

In the past we've compared AF to MF, and have repeatedly found that, for the most part, we get better results with manual focus.

Since we shoot hundreds of test images for the studio scene and dynamic range widgets, manually focusing 3 (or more) times becomes very laborious. It's a tough balance to strike.

Jun 18, 2019permalink

Barry Benowitz

Silly question but is the camera stand on cement floor or wooden, on the ground floor of a building or raised floor or second story. On a city street where trucks frequently pass by?

Jun 19, 2019permalink

@MikeRan. As Rishi says, AF can not be relied upon to be absolutely optimum - in any camera. Actuating AF several times doesn't help as it may have a tendancy to back- or front- focus. The only way to be sure is to manually bracket focus in small increments, starting from a point which is definitely front-focused to a point which is definitely back-focused - ideally using a macro rail rather than the lens ring. Totally get that this may increase the burden beyond what is feasible if you have to do this manually, but it's the only way to be sure I think. You could speed up the process with an automated rail and streamline systems for choosing the sharpest shot however.

Jun 19, 2019permalink

MikeRan

Only way to be sure, except when it doesn’t work... anyway mirrorless is here. And back/front focusing due to calibration problems are a thing of the past for us...

Jun 19, 2019permalink

David610

I suppose this question has been asked a million times (although as a regular on this site since 2001 I would have thought I had seen the answer) ... Why can't all the images be presented at the same magnification? I suppose the answer is the aspect ratios of the sensor. It makes comparisons totally meaningless. I have felt the tool is past it's use by date for some time.

Jun 18, 2019permalink

They can be - just click 'Compare' at the upper right of the widget.

Jun 18, 2019permalink

David610

Thanks I have never clicked on the Comp button. It might be nice if when you hover the mouse pointer over it it indicates it's function.

Jun 18, 2019permalink

tm8

There's a button for that. Nor is comparing on a pixel level meaningless if you're interested in the noise performance at a 1:1 crop.

Jun 18, 2019permalink

DarnGoodPhotos

David, I get a tooltip on mouse over; however, the description could be better.

Jun 18, 2019permalink

Eric Calabros

I see why I don't need higher resolution than 40mp.

Jun 18, 2019permalink

Barry Benowitz

you don't
your correct
but I do!

Jun 18, 2019permalink

tangbunna

is that money limit?

Jun 19, 2019permalink

Marcelobtp

GFX 50 has better colors in raw and jpeg, what has Fuji done? looks horrendous.

Jun 18, 2019permalink

The JPEGs look similar to me, but Adobe's color profile for the GFX 100 seems to be off. We've contacted Adobe, thanks for bringing this to our attention.

Jun 18, 2019permalink

Pixel8888

I would rather use Capture One with Fujis, Xtrans or Bayer.

Jun 19, 2019permalink

Marcelobtp

@Rishi Sanyal Jpegs are less different but still, seems Fuji has some counter algorithm to remove the color cast, from people's faces, a problem that has been plaguing Sony sensors for a long time, but the problem it's too agressive and it's removing too much causing faces to look pale and plasticky most of the time.
Here is some video from Max Yurev that makes it clearer.
https://www.youtube.com/watch?v=sMtPHdjxvNE&t=548s

Jun 19, 2019permalink

Mssimo

DC-SR1 is not defaulting to pixel shift image.

Jun 18, 2019permalink

Fixed.

Jun 18, 2019permalink

Dennis

Not bad for casual use, but the IQ180 is the obvious choice for anything you care about ...

Jun 18, 2019permalink

Clint Dunn

Such a ridiculous statement for so many reasons.

Jun 18, 2019permalink

Tomasz_Wk

« We are planning to re-shoot the scene as soon as a production camera arrives and would like to apologize for any misconception these images may have furthered. »

Thank you DPR! You’re the best!

Jun 18, 2019permalink

Thanks to DPR for acknowledging this. Looking forward to the re-shot samples. In the interest of transparency, and because we are moving into such ultra-high resolution territory, maybe DPR should share a bit more of their methodology for shooting the studio scene. I think most people appreciate that it's becoming a pretty difficult thing to do in a reliable and exacting manner.

Jun 18, 2019permalink

cosinaphile

waiting for this ... thanks ....

Jun 18, 2019permalink

(unknown member)

"We are planning to re-shoot"
Don't! We need real-world manual focus comparisons, not comparisons where the focus is adjusted with a scanning tunneling microscope.

Jun 18, 2019permalink

No please do. It's always good to see what a system is capable of in ideal conditions.

Jun 18, 2019permalink

winkalman

Some of us would like to know what the camera is capable of rather than whether or not the DPR staff is good at manual focusing.

Jun 18, 2019*permalink

OhWeh

Thanks for this update. This explains the blurry images.

Jun 18, 2019permalink

pentaust

Miss-focus in unlikely on a test chart and I guess the tester checked the focus before exposure. EXIF data indicates shutter speed 1/40th s., a value lower than the max flash sync speed on the gfx100, indicating possible presence of shutter blur. So the value of the gfx100 would be more for the higher frames rate, 4k video and phase detect autofocus, not so much for the 100Mpixels.

Jun 9, 2019*permalink

Shutter is not the issue, as we shot fully electronic shutter. MTF analyses of the separate color channels indicate focus could be improved.

Jun 18, 2019permalink

vscd

The most impressive cam for resultion and what you get out of it, is still the Sigma Merrill. For 15 MPixel you easily get something like 40+... and yes, I know the backdraws. I have one ;)

Jun 6, 2019permalink

stevevelvia50

Impressive how the Olympus EM1 X, possibly the most versatile camera on the planet, when set to high res mode, can hold its own against the big boys here at such high magnification of the test scene. Even the noise characteristics at higher iso values are similar, and this is true pixel peeping.

Jun 6, 2019permalink

otto k

EM1X has exactly the same image as EM12. G9 is actually better. ISO makes no sense as these images are merged from 8 images so maybe cut the ISO by three stops?

Jun 18, 2019permalink

DarnGoodPhotos

Unless there is movement in the frame, at which point a single, high-res shot is better.

Jun 18, 2019permalink

larkhon

In what way is G9 better?

Jun 19, 2019permalink

otto k

It produces more detail in high resolution pixel shift photos.

Jun 19, 2019permalink

kristian1

Miss focus...

As you can see with hair , botle.. those are way softer and behind is sharper, so the image is back focused. Sensor / lens combo looks to be very sharp just fine tunning focus is needed.

Jun 5, 2019permalink

DPReview reported that they shot that image at f5.6 and with a different lens, while the GFX 50r image was at f8.0 on a different lens. They did this because they felt they were getting diffraction issues at f8.0 on the GFX 100 shot. They also agreed that depth of field was likely the cause of the rapid focus fall off at f5.6.

Jun 6, 2019permalink

Using f/5.6 of f/8 will not make much difference if you did not focus well or have a bad lens copy...

Jun 6, 2019permalink

kinoworks

Moving the goal posts in changing tests to suit the camera?

As an aside, anyone know the source on the sensor?

Jun 5, 2019permalink

Not at all. What is happening here is that people report discrepancies in the test images, then DPReview, much to their credit, acknowledges the issues and recognizes that the game is changing for high resolution and medium format sensors. I understand that some people might prefer that tests show no advantage for expensive cameras, but I would prefer an accurate representation of camera differences, rather than making them all look alike. Times have changed, the testing needs to evolve as well. If you look at the images as I suggested, you will see the differences for yourself. You won't need me or anyone else to convince you.

Jun 6, 2019permalink

The sensor is Sony's IMX461

Jun 6, 2019permalink

Important Note: To their credit, DPReview is acknowledging some issues with this test. The thread at https://www.dpreview.com/forums/thread/4399061#forum-post-62750010 has the full dialog, however, if you want to see the GFX 100 at its best in this particular test: Do not bother with the Daylight view, switch to the low light view and switch to jpeg! Look at the cracks in the clear coat on the paintbrushes at the lower right corner, and the texture of the paper in the playing cards on the upper right side. DPReview is looking into revamping their sharpening protocols on the raw files, the lenses and f-stops used, and are likely to re-shoot this test. For now, use the low light image and choose jpeg for your comparison. Feel free to switch the Panasonic frame into Raw and the high resolution, just for fun, but leave the others in jpeg.

Jun 5, 2019permalink

pentaust

Why would we need to adjust test condition for the GFX100? Why not adjust condition for every camera model then? The goal of a comparison is precisely to see how images compare in the same conditions. Even on the JPEGs we see some smear that looks like shutter shock blur, JPEG sharpening improved appearance a bit not does not eliminate the blur completely.

Jun 5, 2019*permalink

This is not just about the GFX100, it applies to all of the high resolution sensors currently on offer, as well as those to come. As the test stands now, people are saying they see no real difference between the cameras, when in fact, by switching to the low light frames and jpeg, suddenly real differences are apparent. It is time to revamp that test to account for new standards of resolution and depth of field fall-off. Perhaps a new test for 30 plus megapixel cameras? That will take a while to put together, in the meantime, some readjustment of this test might be in order. How about adding a daylight balanced flash test frame that has more dramatic/specular sidelight such as the low light test rather than just the flat light designed to hide subtle detail like skin pores. :)

Jun 5, 2019permalink

The problem here isn't about sharpening, it's about focusing or maybe a lens problem... Some parts of the GFX 100's Daylight RAW Studio Test Image are sharp, e.g. the bearded man drawing... If the issue were about sharpening, the whole image would be soft...

Jun 6, 2019*permalink