When Resolution Is Not Enough | Featuring the Fujifilm X-H1, X-H2, GFX50R & GFX100S 

The problem is with expectations and math. When comparing resolution we are dealing with area math rather than linear math. Double the pixels does not equal double the resolution, it is 25% increase for double the pixels. Also, you need to resolve line pairs not just pixel dots. I am not surprised that you do not see a clear advantage. especially if you are relying on image stabilization and the zoom lens. That lens is, by all reports, a very good lens, but some lenses in the line are better for resolution I think. So, try a good prime with a tripod and studio flash, then you will get the most out of it, but don't expect earth shaking improvements. I believe 25% is what you can expect. This is why you did not see a bug jump in the APSC cameras when you double the pixels. Again, the math says 25%, but losses due to camera shake or lens quality will detract from the math based results. Very helpful test for the general enthusiast.

A quick thought about resolution doubling and effect on detail... If you have a square 10 pixels by 10 pixels, that's 100 pixels. If you double the pixels in each axis it would increase the area to 400 pixels. A doubling of pixels from 100 to 200 is only a 40% increase in each axis. (square root of 200) This simple arithmetic (partly) explains why going from 26mp to 40mp makes such little difference.

I tested Fuji's 56mm f/1.2 WR at f/4 on a 26mp and 40mp camera, and there wasn't too much difference. However, using pixel shift to get a 160mp image pretty much showed a 4x improvement in detail (+ much better tonality & colors), where small text from across the room that appeared as misshapen blobs at 40mp became very readable at 160mp. It might be fun to compare the X-H2's pixel shift to a GFX100S with & without pixel shift.

The camera does not make for a good photographer, it's the other way round. The X-H1 does not have IBIS, that's the X-H2.

Firstly, just wanted to say Thank you for an excellent video on resolution comparison. I have recently bought a Fuji 50S II and have considerable experience with medium format digital for many years. I appreciate that the Fuji sensors aren’t really medium format sensors as they are still considerably smaller than earlier Hasselblad and Phase One sensors. I have had a number of Hasselblad H cameras (actually a Fuji designed body) with digital backs starting from H2D years ago with a 22MP sensor. I presently have a Hasselblad H3DII-31 (31MP) and a H4D-50 (50MP). These two cameras not only have different size sensors between them, but are also slightly larger than the Fuji 50S II. The 50S II sensor is almost 1.7 times larger in area than a full frame 35mm sensor. The H3DII-31 sensor is very close in size to the Fuji 50S II sensor, albeit the Fuji is CMOS and the Hasselblad sensors I’m referring to above are CCD, which I personally prefer the output from, to me personally more film like. I believe my H4D-50 produces considerably sharper images than my excellent Fuji 50S II (excluding the fact of low light performance of CMOS over CCD) and any lenses differences (I have Hasselblad 80mm and 150mm lenses, both Fuji based designs I think?). The difference I believe is not simply pixel number, but also pixel size in um. The H4D-50 sensor is just over twice the area of a 35mm full frame and because of that produces in my opinion sharper images than my Fuji 50S II images. I’ve obviously conducted similar resolution/result tests to your ones to see how all my cameras compare. I actually prefer for B&W images my H3DII-31 camera, as it’s images are super smooth again because of the larger individual pixel dimensions, which are actually bigger individually than my H4D-50 pixels because of it only being 31MP against 50MP, although the sensor is far closer in size to the Fuji’s sensor. I also think the Hasselblad’s 16-bit sensor against the 14-bit of the Fuji’s sensor adds to slightly better and smoother colour tonality. I would however say, although I still absolutely love my Hasselblad mirrored cameras (I come from the film age and previously processed and printed my own B&W and Cibachrome) the Fuji 50S II is a great camera especially with it’s in-body stabilisation which allows me to handhold with the 35-70mm down to easily 1/8 of a second at 70mm with razor sharp images. In conclusion, I think a number of factors affect the actual image quality, but I personally believe the larger the physical sensor area the better all other things remaining equal. I also personally believe (only my view) that the Fuji 100S doesn’t improve much over the 50S II because it’s simply too large a pixel count for the sensor’s total area! They are all great cameras and I do appreciate that my Fuji for non studio use is basically 1/2 the weight of my Hasselblad H3D/H4D cameras. Thanks again for an excellent video.

Your video is a wonderfully concise presentation that confirms the importance of lens quality as the foundation stone of image quality. I expect that the GFX lenses are the key to the visual differences, assuming LR is doing a fair RAW conversion of these GFX and X-H files. I own one X-H1 and one X-Pro1, XF16-55-f2.8, XF 55-140-f2.8 and XF 100-300. My experience with image processors has me using CaptureOne since 2015. C1 was the best for the X-Trans sensors. I believe still is. Could the GFX files benefit from the C1 RAW converter as do the X-Trans files? (This video supports my intention to get a GFX50sII or its successor without going near the X-H2/X-T5 bodies.)

All images on the left are sharper than the right with all your tests. For some reason your images on the right are blurry in comparison. The xh1 with the older lenses also provides a better resolution than using the same older lenses with the new xh2. If I was looking to get a second body, id get the xh1 and put the extra money towards a couple of lenses. If I was looking at getting into medium format then the gfx 50sii would be the go and again use the extra money to buy a gfx lens

Nice video! Capture One is going to give you better results when it comes to X-H2, X-T5. Maybe you can get your hands on the XF 18mm f1.4 WR or the 23mm F1.4 WR and make another video, I would love to see a comparison with the GFX using those lenses. The Viltrox 27mm f1.2 Pro is another good option for the new 40 MP sensor.

I printed and sold a 4foot print for 1000euros shot on XT2 with the kit lens to a medium format photographer. He thought it was shot on medium format. The print was great. With todays softare, upscaling, denoise, sharpener etc. I recon 16mp is enough for anything up to 100cm. I just got a 12mp image printed 70x50cm and it was perfect from even 5cm away. Megapixels are overrated. The most expensive images of all time were all shot on 35mm film. They are all inn galleries around the world in big prints.

Hi Tim, have you looked at up-scaling. I use a Fuji X-H1 and print at 16” x 20”. I found that cropped images lost their edge as they dropped to a sub 300 dpi/ppi. As such, I started to use GigaPixel AI to upscale, so the print resolution was always 400 dpi/ppi, this improved the print image significantly. It would be worth doing your side by side comparison of the X-H1 image upscale versus the X-H2

Hi Tim, enjoyed your review, I think the differences you have highlighted are not really about the lenses, taking both comparisons 24 v 40 and then 50 v 100, the fact that there appear to be no improvement in overall image quality is due to the fact that the only way to increase the number of pixels on a sensor of a given size, whether aps-c or full frame, is to make them smaller, the smaller a pixel becomes the less efficient it becomes at doing its job which is to capture the quality of light, this is why there was little or no difference between the two aps-c sensors and the two full frame sensors, but a mind blowing difference when you compare a aps-s against a full frame sensor, where each pixel is almost four times the size of the pixels found on the smaller format.

Thanks Tim for a very interesting comparison. Other photographers, like Ian Worth, came to similar conclusions. This will save me a lot of money in delaying an upgrade. Next step might be going from the X series to GFX but I've already earlier been contemplating the GFX 50 SII rather than the GFX 100S.

Your first conclusion ( even if lenses do not resolve sensors…) about the limit of some of your lenses is right. About the GFX vs APS-C, keep in mind that resolution comparisons are only relevant for the same sensor size: From the same distance longer focal length gives you more details ( here , going from 40mm to 70 mm increase the size of a detail by 1.75) Then the 50 Mpx of the GFX are producing a pixel density which is 3.3 inferior to the 40Mpx on APS-C . So you increase the size of a detail by 1.75 and you record it with 3.3 less pixels. The challenge for the APS-C combo, for the lens particularly, is too high. Pixel density equivalence: 16 Mpx micro 4/3 26 Mpx APS-C 61 Mpx FF 100 Mpx GFX Notice that the 50 Mpx of your GFX ( equivalent 30 Mpx FF) is not a big challenge for lenses. You should try FF lenses which have a good circle of image ( standard or tele) instead of MF manual lenses. Those lenses were made for 6x6 and 4.5x6 film cameras.

Sorry if someone has already said this, but I wonder if the difference is greater between the GFX images if you opened them in Capture One? (I stopped using Lr 10 years ago after a slated roof suddenly lost all detail. I opened the same raw file in C1 and the slates were clearly visible. Lr may have improved since then.

Very helpful video. It answered my ,"sell XT-4 for an XH2?" question. I had an XH1 and loved the grip and how it felt shooting so I thought XH-2 instead of XT-5. Based on this video, 26MP to 40 will be a very negligible difference and not worth the upgrade. The fuji 50 to 100mp medium format comparison was excellent and suprising to me! Between this being a hobby for me and not a money maker, I'll keep what I have. Again, thank you for clarifying the mp dilemma for me. Happy New Year Tim!

pretty sure the 35-70 doesn't do the 100mp sensor justice. I'd love to see that comparison on one of the best lenses, such as the 45, 55, 110 or 250 for clarification, I'm just talking about detail. for the x mount i assume the newer gen primes could probably resolve the 40mp, such as the 18, 23, 33 and maybe 56

Thanks for this. I've been considering 3 of these (minus the H1) so these comparisons are very useful to me. After watching, I now have my choices narrowed down to the H2 and the GFX-50S II, I'm leaning towards the latter.

Thanks for producing an awesome video and comparison, Adobe is not the best raw converter for the X trans XH1 and XH2, for GFX I believe its fine due to the sensor type. You will notice a massive difference if you were to use Capture One with sharper results on the XH cameras and would close the gap a bit when compared with the GFX, Obviously the GFX will always give u better Image on still subjects i just wish the AF was better to suite my type of work.

You didn't mention if you had IBS off or not when doing your test images. Assuming you used a tripod, if you didn't turn off IBS, I think that might explain your results.

It is always the hype that gets people into buying the next camera with the most megapixels, but apparently it is the sensor size that makes the difference coupled with a good lens that lets in more light and the lens design itself makes for the pixels to be able to absorb it. Also there is a limit to how many megapixels can be stashed in a sensor as each pixels increase the size decreases and that is the drawback of the quality of an image. Many cameras are sold well because of the IBIS and the speed of the processor and how a movie is produced sometimes, also the monitor is limited to resolve the actual resolution of the camera and that is coupled with the video card in the computer the ram that is able to display all those pixels that are projected on the monitor. I can see now where printing the image will make a difference in trying to see a detailed image than just being displayed on the monitor. Next time you could print some of the images from each camera and see how it looks to the naked eye than the reflected image on the monitor. I personally find that if i shoot a panorama with few frames that are well aligned I get an excellent detail that I could not see through the lens many times. The 50 megapixel GFX will produce a very good panorama with even two images and possibly give you more detail than the one hundred megapixel GFX2 quiet possible. Good testing though from real life shots. Cheers.

You have a very interesting topic. However, i wonder why You did not mention the name brand and f number of the 40mm 3rd party lens that you were so fond of. I also wonder if all the test shots were done by tripod or hand held. If hand held you could have some variable results in sharpness…because stabilization may not be as fool proof as tripod steadyness.

Great in depth comparisons, however a couple of points that I would like to mention. 1) LR is notoriously poor in handling the X-Trans sensors of the Fujifilm X bodies, whereas the GFX Bayer sensor is easier. Consequently it would be interesting to repeat some of these tests using something like DXO or Silkypix. 2) Comparing 50 Mpx to 100 Mpx is not twice the resolution, but 1.4x (square root of 2) as the pixel count is per area, and resolution is along 1 axis only. Would you agree?

It is easier to discern the differences using a lens test chart, but even here you have to make sure that the camera and lens combinations are not both out-resolving the chart because you will still see no difference. I ended up re-testing my X-H2 and X-H2s at twice the distance from the chart in order to see the difference (which was then easy to see). Natural scenes are very difficult because you may not have details sufficiently small to out-resolve the sensor.

Which brand is the 40mm macro lens.

How come the X-H2 looks like it has more muted color science than the X-H1? Was it just a weather difference?

This is *exactly* what I've been looking for, thank you. Having had the X-H1 for a while now and considering getting the X-H2 mostly for wildlife purposes (with the Sigma 100-400) this really makes it easier to hold back. Although there will obviously be more detail and better AF, the difference is indeed not quite as huge as the figures make it seem. No doubt the X-H2 is better, but with the X-H1 at 500 bucks and the X-H2 at 2000, in my opinion and for my purposes it's not 1500 bucks better.

Like you, I have all four cameras. Strangely enough, I too see better images on the lower resolution versions of each. I suspect that there may be a pixel size factor in play. What you gain in resolution, you offset in worse signal to noise at pixel level. Notice how both higher resolution camera looks very slightly softer than their lower megapixel counterparts.

What is the 40mm that you use on xh1 amd xh2

The out of the box results of Fuji’s X-Trans sensors is horrendous in Lightroom. You should run the XH2 file through the “Enhance Details” option then apply a bit of sharpening before comparing to the GFX image. Please let me know if you do it, I’m curious as to how they compare. Thanks.

I sold my 50R and bought an X-T5. I love the portability, ergonomics and speed of the X-T5 (didn't like the ergonomics of the 50R). Since then I've been trying to convince myself that the image quality is on a par with the 50R, and failed! I'm really missing the look of the GFX files, so I've actually just pushed the button on a (relatively) cheap used 50S, which I'm going to use for those situations where fine detail is really important. I know you were focussing on resolution and detail, but I find there are other areas where the medium format sensor shines, notably tonality and the fall-off of highlights, compared to the crop sensor camera. Really looking forward to receiving the 50S. Most of my work (and certainly all of it where the medium format will be important) is slow and tripod based, so the lack of IBIS doesn't bother me too much. I'll eventually pick up another 35-70. (Sold the first one with the 50R), but in the meantime I'm going to be using the 50S with a Samyang 24mm tilt-shift and a couple of adapted Mamiya 645 lenses. I know the Samyang gets mixed reviews, but I must have a particularly good copy as I was getting good results with it on the 50R. (Makes nice panoramas on the X-T5 as well). Anyway, thanks for a well-structured comparison. (Edit to add; regarding "only 10 megapixels difference", if you "crop" the GFX sensor down to the same aspect ratio of the X-H2, the difference in megapixels is even smaller!)

Thanks for the great work !

The 10-24 and most zooms are not really ‘up to’ the resolution of the 40Mp sensor, I think this comparison is more about going to the limits of the lenses. Macro lenses are, of course optimised for close focusing distances, using it in an ‘infinity’ situation won’t be at its peak. The newer/latest Fuji lenses, 33mm, 18mm are much more capable with the 40Mp sensor. The 50R would be expected to prove a lot better, the larger format/pixels accommodate the lens resolution much more efficiently than smaller formats. Obviously APS-C to 44x33mm will be starkly revealing, FF ( 36x24)mm, 50/60 Mp, the difference will be more subtle.

Great comparison 👍🏻

I think your are right with your conclusion of the reduced lens resolution for the APS-C format to 16 or 26 megapixels. In my opinion based on the bayer or the x-trans color pattern which reduces the resolution in color down to about 70 percent it makes sense that the resolution of the sensor should be higher as the resolution of the lens. This significantly reduces the moire effect caused by the color patterns. For the highest possible 4k video resolution you need 4x (bayer pattern) or 6x (x-trans pattern) sensor resolution to archive the 1:1 color pixel ratio for red and blue to a 4K-TV set. Using Pixeshift can be a solution for still objects.

Thanks for these comparisons. It confirm what I always thought. The 50R (50S & 50SII) sensor is by far the one showing the greatest colors to me. They say the 100S has a better dynamic range but the dynamic range of the 50 sensor is superior and more pleasant in my opinion. I believe it is due to its specific Bayer Matrix sensor.

24MP is far too much for a APSC! Even with youtube compression you can clearly see that the XH1 image is sharper than the XH2. This is due to lens resolving power. Not even speaking about ISO noise due to pixel pitch. I whish a manufactuer could do a 16M ASPC or 24M FF with today's sensor technology.

The 10-24 is definitely not good enough to make most of the 40 mp sensor. Once I sell my xt3, I'll sell the 10-24mm and get and a prime for ultra wide.

Resolution is important only when you want to blow up the picture for bigger prints…. Otherwise even 3 MP is enough on APC or FF. I think pixel density is more important than pixel count!!!!

Interestingly the Fuji 16-55mm is one of Fujis recommended lenses to use on a 40mp sensor. I think this is about pixel density and you've proved this with your upscaling test. There is a limit to detail that you can get from sensor size

X-H2 has other benefits to consider. The fastest shutter speed of 1/160,000 sec for example allows shooting any lens at optimal aperture in bright sunshine, without the need for ND filters. The digital 2x magnification is also handy, as well as the shooting speed, etc..

To actually SEE doubling of the linear resolution we need a 4x area of the sensor. All things being equal it is pure algebra. IBIS is off course a sword with twin edges.

I think Adobe again isn't giving the best result with the 40 Mp sensor. It took a long time for them to get even close to other Raw rendening software. I use LR/PS as well as Capture One. Capture One gives a clearly better result than Adobe. Some have tested rendering and demosaicing in DXo and to do the rest of editing in LR. The result if you compare Fuji X-trans sensors around 25 MP with the 40 MP gets a clear boost with DXO doing the demosaicing and rendering. The software you are using seems to play a significant role in getting the best result.

Well I own all 4. And my 50r and 100s have produced 72x60” prints. And the look aaaaamazing

I have a feeling the new GFX 50 and if they push up to 60 And same body as the new GFX100ii Will be a weapon Tho am tempted with the sales a XH2 printing same size as XT3 and getting better image quality so what would that be on the 40mp 70% THO after seeing this I may save money and wait for new GFX 50 - it will have great video

Oh my god. I was following your synth channel before and now I see u in photography. ❤❤❤ really surprised and happy to find out this channel.

Not sure how significant it might be, I don't think the 10-24 lens is on Fuji's 'list' of recommended lenses for the X-H2 sensor. I've seen images with the relatively new 33mm 1.4 Fuji lens on an X-H2 and the detail looks extremely high. The 3rd party Viltrox 75mm f 1.2 is reported to realise the extra potential in the X-H2. I agree the "50" series looks to be a sweet spot of resolution/detail/cost.

Loved the video, and very informative, as you've no doubt already heard the X-tans sensor responds best with Capture one, and with the GFX, there's only a 25% to 30% jump in actual resolution between the 50 and 100; I personally love the old pensioner 50 sensor because even at 14 bit it looks fantastic.

I used to be a amateur photo enthousiast, but due to lack of time and other interests, I haven't been shooting in a long while. You make some valid conclusions. Megapixel alone is not cure-all. If the sensor size stays the same, the sensorsites gets smaller, collect less light and needs compensated. That is usually done in software in-camera and as you pointed out, Photoshop may be even better in that, with more up-to-date algorithms. Your second observation about the glass is also correct. Not all lenses are capable of capturing the high resolution. What I know of my Canons, is that they have an option to micro adjust the lens to the sensor. It might be that the distance between the sensor and the lens on your X-H2 is not optimal, as the images to me look a little blurred at 100%. But that could also be related to that the lens itself is not capable of capturing the amount of pixels. (But for me, that is guessing.) To me, the GFX50R gives the best image quality results, but as you said, it lacks image stabilization and can't film at 4k. The answer I gave myself for the first problem was buying a monopod (which can be used for smooth panning thanks to the fluid base).

If I am using the HX1 for portraits would having a 40mp sensor really matter ?

To me comparisons like this only really make sense if you look at the same image cutout on both, instead of looking at the same relativ magnification on each one.

Tim, thank you for this analysis! This is exactly how no one else compares cameras and lenses. I’ve been at this crossroads. December moved from SONY A7R4 and A7S3. Menu’s on Sony horrible and love FUJI colors. Started a Y/T Channel about my Retired Life in Japan. Enjoy video and H2S nails my needs. Started printing again and want more details in my photos. It’s not that I Don’t like the H2 but I would like to increase my wow factor. Photography is my real joy. So much here to capture. I appreciate logical thinking… thanks for posting. Only thing that scares me is the cost of GFX lenses. Today I saw a Y/T on using Pentax 645 lenses. Curious if you ever looked at Pentax lenses for use on GFX? I hear people complain about slowness but that doesn’t concern me as I understand the difference in moving up to Medium Format. Actually annoying to hear it as often. Thanking you again!

A tremendously helpful review, thank you very much for taking the time to share your thoughts and experiences on the matter. I’ve been a Fuji X-T3 user for quite some time and am now deliberating over whether or not to upgrade to the new sensor or to jump ship entirely in favour of a full frame system (either the S5II or the A7IV). Thank you again, you’ve got a new subscriber here! Warmest regards

should try the 110mm at f/5.6, it's CRAZY sharp.

Loved this, as I do with your synth reviews

If u do this in capture one u get better photos

Your voice is very soothing

Thanks for this comparison. For the sake of giving the X-series its best shake, particularly in a landscape-type setting, I absolutely recommend using LR's "denoise" / "enhance details" feature before peeping the files (and not doing much noise reduction, since it isn't needed for base ISO files in good light). I recommend this not to try to give the sensors an unfair advantage, but because LR by default does weird s*** to Fuji files when it comes to landscape/foliage/rocks/natural textures. You end up with a bloated file size, but at least they come out looking like photos instead of weird "painterly" smears. I'd been confused about this for awhile because results are inconsistent - sometimes you really need the denoise feature, other times it does not matter. But I find the issues almost always show up when trying to shoot any sort of landscape-type shot. With bayer sensors on board, I would not expect these issues with the GFX line. If C1 can handle RAFs perfectly by default, I don't see why the mighty Adobe LR cannot. Annoying.

24megapixels not enough? Really mate? You’re sure you’re a photographer? I Can print a A2 size from a 35mm film 1972 camera pin sharp. Megapixels should never be a factor to pick a camera.

Wait you went from Sony to Samsung? Nikon is the best for hybrid. Full frame and the S lenses are second to none. Crop sensor is for amateurs