24 MP X-Trans III has the same resoution of 36MP Bayer sensor.

[bitbonk]

I just read this weird claim on fujirumors.com:

 

RESOLUTION: Fuji said that the new 24 MP X-Trans III MP has the same resoution of 36MP Bayer sensor. And this might be true, since already dpreview tests show how the 16MP X-Trans sensorII has a “Bayer-equivalent” resolution of 24MP.

 

Can some link me to that dpreview article where this claim was made and maybe to the place where Fuji itself claims this? Knowing a bit about the X-Trans sensors (but maybe not enough), it just doesn't make sense to me.

 

The resulting images of the 24MP X-Trans III sensor will have 6000x4000 pixels and the resulting images of a 36MP Bayer sensor will have 7360x4912 pixels (e.g. Nikon D800E) how is the resolution the same?

[umad?]

http://www.fuji-x-forum.com/topic/2043-live-from-tokyo/?p=18542

 

 

If one takes low pass filter and different lenses into account, this of course can be true, but is no in general 

 

(It was the same with the last gen, which could achieve similar results as an 22MP Canon 5D3) 

 

 

Of course this is mostly marketing. But it also shows how confident Fuji is in their lens lineup. 

[danwells]

There was a graph in Fuji's presentation (linked all over Fujirumors) that shows this... Actual resolution in the print is affected by the lens, the low-pass filter and the rest of the filter pack over the sensor, as well as by sensor characteristics, mainly noise. Fuji tends to be pretty darned close to optimum on all of the above (no low-pass, X-Trans seems to be pretty kind to resolution, VERY good lenses). Practical, printed results from the 16 MP X-Trans were MUCH better than many other 16 MP sensors (especially older sensors or Micro 4/3, and, by the time of the X-T1, most 16 MP sensors were Micro 4/3), and were not as far off 24 MP APS-C as one would expect - they were in the expected range for about a 20 MP sensor, and could catch a 24 MP sensor under certain conditions (until the FE lenses came along, E-mount APS-C always underperformed due to lenses - even now, you have to use a full-frame lens on the APS-C body, because the APS-C lenses are mostly cheap kit zooms). If you wanted to show 16 MP with higher resolution than 24 MP, the test was easy - XTrans II with a great lens (most of the primes will do, as will the 16-55 f2.8 or the 50-140) against a NEX-7 (the first and noisiest generation of the 24 MP sensor, and has a low-pass filter) with just about anything other than the best of the FE lenses.

 

To get the 24 MP X-Trans III (which, by all preliminary accounts, is a truly wonderful sensor,  coming very close to the theoretical limits of 24 MP with a good lens - and if Fuji has anything going for them, it's good lenses!) to outperform a 36 MP sensor, I can think of two routes. First, I suspect it might well outperform the not yet introduced APS-C 36 MP sensor in a relatively fair test (Fuji clearly spent a lot of time on the sensor, and would have bought the 36 MP if they preferred it). I have no proof, but that may not be a great sensor - Sony would have introduced it by now if they were happy with it, and it would have also been logical for a high resolution, lower speed variant of the Nikon D500. Add in a typical APS-C E-mount lens and you may well have something that underperforms the Fuji.

 

Even if you look at a FULL-FRAME 36 MP sensor, especially an early model with a low-pass filter, the real resolution is not that much higher than a great 24 MP sensor - the highest numbers for 24 MP sensors are around 3200 lines/picture height, while a D800 is around 3500 (all of these numbers need to be taken with a BIG grain of salt, as two cameras with the SAME SENSOR, both with no AA filter, can come out a few hundred lines apart)! Put Nikon's standard 24-85 FX zoom on the D800, and a 90mm f2 on the Fuji (or the 16-55 f2.8, to go zoom to zoom), and there goes the difference! Yes, the lenses are no fair, but Nikons spend a lot more time with lousy zooms on them than Fujis do, even 36 MP Nikons...

[bitbonk]

Sounds convincing, does anyone have that preview article mentioning this too?

[mbhh68]

Yes! In the in-depth review of the X-Pro1 dpreview took a closer look at the x-trans-sensor technology and compares the theoretical maximum resolution with the one of the traditional bayer-style sensor: http://www.dpreview.com/reviews/fujifilm-x-pro1/18

[scoobysnapstories]

its really good to have a bigger pixel for aps-c know why? now you may use third party lenses like nikons/canons etc. and use a focal reducer or booster. and the quality will amaze it. thats where the whole same sensor size will come in play against a nikon d800 etc. try it for yourself and youll see what i mean, well when the xpro-2 comes. =) cheers

[r o b e r t]

I'm not buying I just wish they'd just stop with the "big" claims i.e "worlds fastest" etc. 

[danwells]

If they're close to the theoretical maximum of 4000 lines/picture height, they won't quite reach what 36 MP cameras can do, but they won't be far off. The best number I've seen out of a 36 MP camera on any test is about 4200 lines. By comparison, a good 24 MP camera (no AA filter) is just over 3500 lines tested in the same way. These results vary quite a bit depending on how you interpret the chart (different sites have the same camera hundreds of lines apart), but the gap of about 600-700 lines between good 24 MP APS-C (Nikon D7200) and 36 MP full-frame of the same make (Nikon D810) is rather constant if you look at both tests from the same source (whether you interpret the chart tightly or loosely, as long as you do it consistently, the gap's not much different)  X-Trans (also consistently) outresolves Bayer by about 300 lines at 16 MP, so the expected difference at 24 MP (assuming relative sensor quality is similar, and, if anything, the 24 MP X-Trans may be relatively better) will be somewhere between 300-450 lines in favor of X-Trans. That eats up about half the gap between the D7200 and the D810, leaving the 24 MP X-Trans III equivalent to something like a 30 MP APS-H sensor.

 

If there IS a 36 MP APS-C sensor in the wings, and it's not a great performer, I suspect it'll outresolve the 24 MP X-Trans III at low ISOs, maybe not by much, but it might well fall behind at higher ISOs. Micro 43 sensors have a tendency to fall behind larger sensors of similar resolution, and the 36 MP APS-C has a Micro 43 like pixel pitch.

[danwells]

The a6300 coming out with the X-Pro 2 sensor in it suggests that either a 36 MP APS-C sensor doesn't yet exist, or it was developed, but Sony wasn't able to get satisfactory noise or dynamic range performance out of it, so they shelved it (at least for a generation). There has been enough noise about it that I suspect it actually DOES exist in fairly late prototype form (as in there are (probably Sony, maybe Fuji as well)) 36 MP APS-C bodies around in small numbers, but they take pictures with a lot of detail, but poor dynamic range. Fuji engineers made comments that sounded like they'd looked at such a sensor, not specifying resolution, but something >24 MP, and ended up choosing the X-Pro 2 sensor instead. That could have just been the known Samsung 28 MP sensor, but there has been enough of a kerfluffle about a REALLY high resolution a6000 replacement that I suspect it was a Sony prototype over 30 MP.

[haroldp]

When x-trans was introduced, most beyer sensors had low pass filters, now many do not.

[gdanmitchell]

"Even if you look at a FULL-FRAME 36 MP sensor, especially an early model with a low-pass filter, the real resolution is not that much higher than a great 24 MP sensor - the highest numbers for 24 MP sensors are around 3200 lines/picture height, while a D800 is around 3500 (all of these numbers need to be taken with a BIG grain of salt, as two cameras with the SAME SENSOR, both with no AA filter, can come out a few hundred lines apart)! Put Nikon's standard 24-85 FX zoom on the D800, and a 90mm f2 on the Fuji (or the 16-55 f2.8, to go zoom to zoom), and there goes the difference! Yes, the lenses are no fair, but Nikons spend a lot more time with lousy zooms on them than Fujis do, even 36 MP Nikons..."

 

As a person who shoots Fujifilm X-trans cameras and also shoots high MP full frame DSLRs, I regard the claim that the likely-excellent 24MP X-Pro2 sensor will equal full frame 36MP sensor performances to be marketing hyperbole. Before I get into the details, I intend to get the X-Pro2 and I look forward to using the 24MP sensor camera after using a 16MP Fujifilm X-series camera for three years with outstanding results.

 

Increasing photo site density has the potential to improve the pixel resolution of photographs, though often by less that some might imagine. For example, the 50% increase in photo sites obtained from going from 16MP to 24MP will produce a print with about 22% larger horizontal or vertical dimensions with the same original sensor resolution. I regularly print at the equivalent of 18" x 27" from 16MP Fujifilm sensor originals, so the 24MP sensor will let me print with the same source resolution at about 22" x 33". That is larger, but not hugely so. 

 

In any case, 22" x 33" is a rather big print. I occasionally print larger, but that is rare. So perhaps the Fujifilm claim could accurately reflect the reality that, all else being equal, the X-tran 24mp sensor can produce print quality that is indistinguishable from 36MP full frame originals at such sizes. That is an impressive and useful thing, but it is not the same as being equal to 36MP full frame. (With my current 51MP full frame system, I am confident of print quality at sizes such as 30" x 45" and even larger.)

 

It is possible to come to terms with two seemingly contradictory understandings here. On one hand, 24mp cropped sensor systems are not equal to 36MP (or larger) full frame systems. The optics and physics simply don't allow that. On the other hand, for virtually all photographers shooting such camera, there is almost never going to be a visible difference in the quality of their final work based on this MP difference, and for them the two are equivalent in practical terms.

 

I look forward to getting and using my 24mp Fujifilm system. :-)

Dan

[danwells]

I think we actually agree - the increase in linear resolution (whether it is from 16 to 24 MP, 24 to 36 MP or 36 to 51 MP) is there, but it isn't huge - now, the jump from 16 to 51 MP is a different story. What Fuji's marketing department is saying seems to be "we challenge you to tell in a print", and 16 MP X-Trans II on a 5 year old sensor was already darned good. Give them the small jump from 16 to 24 MP, coupled with a brand-new sensor with state of the art per-pixel performance, and I think it'll be close enough to a sensor with one more small jump that unfair test conditions can make it really hard to tell.

 

The three most obvious unfair conditions that exist in the real world are lenses, AA filters and raw compression artifacts. The fairest test (and one that 36 MP would surely win, if by a relatively small margin, because a third more pixels isn't enormous) would be a D810 (which has no AA filter and can produce uncompressed raw images) against an X-Pro 2, both locked down on a studio stand, and both with top quality primes, preferably short teles (Nikon has no modern 135mm lens, so the ideal comparison against the 90mm Fujinon is hard). The best we can do might be 85mm f1.4 Nikkor against 56mm f1.2 Fujinon at around f5.6? Under those conditions, I have no doubt the D810 would win (although it would take a good-sized print and a careful inspection to see).

 

Start deviating from those conditions and you can introduce enough variability that it's no longer so clear... A D800 (with an AA filter and probably a different sensor generation) loses just a little bit of the acuity of its younger brother - not enough to lose if the other conditions are still fair, but it's closer now. Go from the studio stand to a regular tripod and forget to lock the Nikon's mirror up, and mirror slap is perhaps another little bit. Now switch lenses to what most photographers are likely to have on the body, and Fuji may pick up a bit on better glass, depending on decisions.

 

Basically, a single jump in resolution like 16-24 MP or 24-36 MP (especially with X-Trans' slight resolution advantage over Bayer of the same pixel count going the other way) is small enough that other deviations can easily counteract it, whether that is lens quality, the presence or absence of an AA filter, or whatever else. The story between 16 MP X-Trans II and 24 MP X-Trans III is not only the resolution jump, but that coupled with a much newer sensor. A triple jump in resolution (16-51 MP) is going to be so noticeable that it would take a lot of other things being different for the resolution not to dominate.

 

I haven't tried this (I've seen neither an X-Pro 2 nor an a6300, although I have a Pro-2 on order), but I think that the difference between an X-Pro 2 and an a6300 ,each with its default lens, is perhaps going to be greater than the difference between the X-Pro 2 and a 36 MP camera with relatively similar lenses, despite the X-Pro 2 and the a6300 sharing a sensor. This is because the a6300 is kitted with the 16-50 powerzoom, about the WORST lens in present production in any mount, exclusive of Lensbabies and such (Sony could have done themselves a huge favor by bringing out a better APS-C standard zoom, but shot themselves in the foot again by choosing the worst of their three options as the kit lens). Fuji doesn't have a kit lens, but the lens they are showing the camera with is the 35mm f2, a very, very good lens. The other option is the Fujinon 18-55 XF, also a very good lens. The difference between a good lens and the terrible Sony powerzoom (which was built for extreme compactness on the viewfinderless low-end a5xxx series) is probably greater in many ways than a single jump in resolution. I'd rather have an X-T1 with a good lens than a brand-new a6300 with that powerzoom.

[haroldp]

I think danwells has stated this very well.

 

One minor point, lossless compression is exactly identical to uncompressed. As pixel counts go up, Fuji should support a lossless compression option.

 

... H

[gdanmitchell]

I think danwells has stated this very well.

 

...As pixel counts go up, Fuji should support a lossless compression option.

 

I believe the specs say that the X-Pro2 does support lossless compressed raw.

[haroldp]

I believe the specs say that the X-Pro2 does support lossless compressed raw.

Thank You, you are correct.

 

Hopefully Fuji will provide this feature with firmware updates for their more recent camera's as well.

[flysurfer]

 

 

Hopefully Fuji will provide this feature with firmware updates for their more recent camera's as well.

 

Certainly not.

[mdm]

Hopefully Fuji will provide this feature with firmware updates for their more recent camera's as well.

 

An old processor doesn't have enough power for that.

[haroldp]

An old processor doesn't have enough power for that.

Far older processors in Nikon and Canon (2007) are doing just fine wth it.

[mdm]

I've seen somewhere around the forum that Fuji recently said that only X-Pro2 will be capable of Acros film simulation, so who knows what's on their mind...

[donrleonard]

The technical details behind this claim are beyond my pay grade, as they say.

 

But recently I also printed a 16mp xt1 file shot at 20x30". I held my breath as I told photoshop to up-sample it to 240ppi and sent it to the printer. It's now hanging in my living room and looking fantastic.

[phrehdd]

If they're close to the theoretical maximum of 4000 lines/picture height, they won't quite reach what 36 MP cameras can do, but they won't be far off. The best number I've seen out of a 36 MP camera on any test is about 4200 lines. By comparison, a good 24 MP camera (no AA filter) is just over 3500 lines tested in the same way. These results vary quite a bit depending on how you interpret the chart (different sites have the same camera hundreds of lines apart), but the gap of about 600-700 lines between good 24 MP APS-C (Nikon D7200) and 36 MP full-frame of the same make (Nikon D810) is rather constant if you look at both tests from the same source (whether you interpret the chart tightly or loosely, as long as you do it consistently, the gap's not much different)  X-Trans (also consistently) outresolves Bayer by about 300 lines at 16 MP, so the expected difference at 24 MP (assuming relative sensor quality is similar, and, if anything, the 24 MP X-Trans may be relatively better) will be somewhere between 300-450 lines in favor of X-Trans. That eats up about half the gap between the D7200 and the D810, leaving the 24 MP X-Trans III equivalent to something like a 30 MP APS-H sensor.

 

If there IS a 36 MP APS-C sensor in the wings, and it's not a great performer, I suspect it'll outresolve the 24 MP X-Trans III at low ISOs, maybe not by much, but it might well fall behind at higher ISOs. Micro 43 sensors have a tendency to fall behind larger sensors of similar resolution, and the 36 MP APS-C has a Micro 43 like pixel pitch.

 

Hi Dan - I'll just say that I might be a touch handicapped as I can't help but consider the testing done on both film and lenses from days of yore.

 

As mentioned, there are theoretical limits associated with various sensors and then there are lens resolving power. It would seem that if we are going from paper to real world, a test would have to have a lens that resolves at a higher rate than any camera (sensors) compared. The ideal is to use exactly the same lens on both cameras so that the lens is a known quantity or rather a constant in the testing.

 

As for resolution of sensors. I admit I find some of the discussion a bit confusing. I say this as resolving power of a sensor (to me) suggest greatest ability to differentiate (such as the common term of resolving lines) and nothing more. Given two cameras set to their ideal "ISO" sensitivity and correct light, constant lens, it would seem logical that a 36 vs 24 outcome from same size sensors would favour the 36. Next of course is by how much and if in the real world it has significance.

 

As for usefulness of more pixels, I gather most agree that the complexities of more pixels per same size sensor reaches a point of diminished returns in creating limitations. Hopefully, I am on the right track here. In the days of film, I always tried to match "slow films" with higher resolving lenses and when it came to faster films or pushed film, regular lenses seemed to fit the bill. I admit I was one of those folks that enjoyed getting as much as I could out of technical pan film and then the next round going for faster film pushed and allowing the grain to bring a dimension to the image upon print.