The FX vs. DX dilemma

DimensionSeven

Leading Member
Messages
893
Solutions
1
Reaction score
17
Location
Kaposvár, HU
Ever since Nikon came out with the FX sensor (full Leica frame), I've been attracted to it. The per pixel noise quality not only on the high iso range, but the extremely smooth base iso was really appealing. The smaller DOF capabilities was also really nice. With the low pixel density of the 12MP sensor diffraction became almost a non-issue again, like in the film days. I think FX has a future with a brighter horizon than DX: with today's 16-18 MP DX sensors pixel density is very high, reaching the level of compact cameras, and that only means diffraction by f5.6 or around with no additional gain in resolution, as virtually no current lens can outresolve these sensors. On the other hand, FX has a bigger playground area with lots of reserves for 20MP or above with diffraction being only a problem at apertures still acceptable for landscape work.

The thing is, I'm mostly shooting stopped down landscapes and use filters (polarizers, ND filters, ND grad. square and color grad. square filters, infrared filters) a lot. There was no FX lens in the wide angle territory fitting the bill for quite a while: the 14-24 2.8 is a monster and cannot take filters, while the 17-35 f2.8 is overpriced compared to his performance in the digital age. Then, the 16-35 f/4 came out a year ago, and I thought that was the one for me. It takes filters, does not weight a ton, outperforms the 17-35 in terms of sharpness and although costing a lot, it's not more than the original price of the 2 DX lenses I'd had been willing to part with for it (Sigma 10-20 and 10.5 fisheye). I even had my hypothetical lens set up for FX: 16-35 f/4 for the wide angle, 50 f1.8 AFD or 50 f1.4 AFS for midrange (maybe a Tokina 28-75 f2.8 for a walkaround zoom), and my current 70-300VR for the long end. The Sigma 10-20 and the 10.5 fisheye would have been sold, and the 16-85VR would have been used exclusively on my D90-IR.

Untill today, I thought that FX was my future too. But. There was Nikon day at my favourite local photoshop (not Adobe's lol!). There were most current top notch lenses (think 14-24 f2.8, 24-70 f2.8, 70-200 f2.8 VR2, 24 3.5 PCE, 85 1.4 AFS, 24 1.4, etc) and DX/FX DSLRs (except for the new D7000) displayed and everyone was allowed to play with them. I've had a brief flirt with a 16-35 f/4 on a D3x (omg...), and liked it (although the lens itself feels really front heavy).

The shock came as I tried to put a polarizer along with the Cokin P holder on top of the lens. I had to zoom in to 24mm to exclude vignetting with the normal holder and to 23mm with the wide angle holder! Without the polarizer, I barely won 1 mm with both holders. With the B+W ND110, being thicker than my slim CPL, the case would have probably been even worse I guess. OMG! This surprised me, as with the Sigma 10-20, I can get away at 12 mm from any vignetting in landscape mode, and that is equivalent to 18mm in FX land... This just rendered the 16-35 f/4 useless for my kind of work, as I prefer to use all kinds of filters for wide angles. The Cokin Z-pro series, that's one size bigger than the P series, costs twice as much, the X-pro series (that would be required probably) costs 4 times as much as the P series. Replacing my current P size filter setup with the X pro versions would cost half the price of the 16-35 f/4 itself, and that's only Cokin, which is plastic with no surface protection, no coating, etc. Lee or above filters would be in the price range waaay above my budget. Not speaking from the size of them - where on Earth would I put them? I'd probably need a new backpack - a smaller one that holds the square filters externally or an ever bigger one (I'm currently using a Tamrac Expedition 7, so it's not a small one by any means already) that holds them inside. Also not speaking from a new tripod as well, as my current 190 Pro-B Manfrotto would be outweighted with an FX body and the front-heavy 16-35 FX.

Damn, looks like I have to admit: FX is just not ment for me. It's for those who need the best high iso possible or the shallow DOF for portrait/bokeh work. For landscape photography, FX is only for the professionals, who can justify both the additional side costs and additional weight of the bigger sensor.

Only one question remains: which filter-capable DX ultrawide lens comes even close to meet the demand of the D7000's 16MP DX sensor in terms of sheer resolution? Is it the Tokina 11-16 f2.8 perhaps?

--
http://dimensionseven.deviantart.com
 
Have you looked at the still excellent 20-35/2.8 AF-D? It might be a bit "tight" but is a good FX landscape lens. Will try to put on the series P holder to see if there's any vignetting, either on the bare lens or on top of a slim UV filter.
Ever since Nikon came out with the FX sensor (full Leica frame), I've been attracted to it. The per pixel noise quality not only on the high iso range, but the extremely smooth base iso was really appealing. The smaller DOF capabilities was also really nice. With the low pixel density of the 12MP sensor diffraction became almost a non-issue again, like in the film days. I think FX has a future with a brighter horizon than DX: with today's 16-18 MP DX sensors pixel density is very high, reaching the level of compact cameras, and that only means diffraction by f5.6 or around with no additional gain in resolution, as virtually no current lens can outresolve these sensors. On the other hand, FX has a bigger playground area with lots of reserves for 20MP or above with diffraction being only a problem at apertures still acceptable for landscape work.

The thing is, I'm mostly shooting stopped down landscapes and use filters (polarizers, ND filters, ND grad. square and color grad. square filters, infrared filters) a lot. There was no FX lens in the wide angle territory fitting the bill for quite a while: the 14-24 2.8 is a monster and cannot take filters, while the 17-35 f2.8 is overpriced compared to his performance in the digital age. Then, the 16-35 f/4 came out a year ago, and I thought that was the one for me. It takes filters, does not weight a ton, outperforms the 17-35 in terms of sharpness and although costing a lot, it's not more than the original price of the 2 DX lenses I'd had been willing to part with for it (Sigma 10-20 and 10.5 fisheye). I even had my hypothetical lens set up for FX: 16-35 f/4 for the wide angle, 50 f1.8 AFD or 50 f1.4 AFS for midrange (maybe a Tokina 28-75 f2.8 for a walkaround zoom), and my current 70-300VR for the long end. The Sigma 10-20 and the 10.5 fisheye would have been sold, and the 16-85VR would have been used exclusively on my D90-IR.

Untill today, I thought that FX was my future too. But. There was Nikon day at my favourite local photoshop (not Adobe's lol!). There were most current top notch lenses (think 14-24 f2.8, 24-70 f2.8, 70-200 f2.8 VR2, 24 3.5 PCE, 85 1.4 AFS, 24 1.4, etc) and DX/FX DSLRs (except for the new D7000) displayed and everyone was allowed to play with them. I've had a brief flirt with a 16-35 f/4 on a D3x (omg...), and liked it (although the lens itself feels really front heavy).

The shock came as I tried to put a polarizer along with the Cokin P holder on top of the lens. I had to zoom in to 24mm to exclude vignetting with the normal holder and to 23mm with the wide angle holder! Without the polarizer, I barely won 1 mm with both holders. With the B+W ND110, being thicker than my slim CPL, the case would have probably been even worse I guess. OMG! This surprised me, as with the Sigma 10-20, I can get away at 12 mm from any vignetting in landscape mode, and that is equivalent to 18mm in FX land... This just rendered the 16-35 f/4 useless for my kind of work, as I prefer to use all kinds of filters for wide angles. The Cokin Z-pro series, that's one size bigger than the P series, costs twice as much, the X-pro series (that would be required probably) costs 4 times as much as the P series. Replacing my current P size filter setup with the X pro versions would cost half the price of the 16-35 f/4 itself, and that's only Cokin, which is plastic with no surface protection, no coating, etc. Lee or above filters would be in the price range waaay above my budget. Not speaking from the size of them - where on Earth would I put them? I'd probably need a new backpack - a smaller one that holds the square filters externally or an ever bigger one (I'm currently using a Tamrac Expedition 7, so it's not a small one by any means already) that holds them inside. Also not speaking from a new tripod as well, as my current 190 Pro-B Manfrotto would be outweighted with an FX body and the front-heavy 16-35 FX.

Damn, looks like I have to admit: FX is just not ment for me. It's for those who need the best high iso possible or the shallow DOF for portrait/bokeh work. For landscape photography, FX is only for the professionals, who can justify both the additional side costs and additional weight of the bigger sensor.

Only one question remains: which filter-capable DX ultrawide lens comes even close to meet the demand of the D7000's 16MP DX sensor in terms of sheer resolution? Is it the Tokina 11-16 f2.8 perhaps?

--
http://dimensionseven.deviantart.com
 
It's not as sharp at the edges wide open, but it distorts less at the same time, so what "worse" means quite depends on the application - although overall I'd pick the 16-35/4 given what I shoot.

To me the biggest shortcoming of the 20-35 is that it doesn't focus very close, for those shots that needs to exaggerate something that's close. For those I usually use a fisheye, though.

As to your question, I've done my own research before, and after sampling pictures taken with many DX ultrawides, the Tokina 11-16 is the only one that has a fighting chance to match the D7000 in corner sharpness wide open. All others are quite soft at the corners. Another one you might want to take a look at is the Nikkor 10-24.
It's optically worse than the 17-35, therefore it's not an option.
--
http://dimensionseven.deviantart.com
 
You have to use Z pro to avoid vignetting @dx11 Tokina. And then you will want to reverse the holder’s frame backwards and use one slot only (no multiple filters if you like this) so looks some adjustment needed anyway 
Resolution wise.. that’s the right pick.

Best,

Hynek

--
http://www.sunwaysite.com
 
Damn, looks like I have to admit: FX is just not ment for me. It's for those who need the best high iso possible or the shallow DOF for portrait/bokeh work. For landscape photography, FX is only for the professionals, who can justify both the additional side costs and additional weight of the bigger sensor.
I think the whole frenzy over FX is way overblown. Our DX sensors were just fine before FX came along. Now everyone wants FX?

Don't forget that when comparing cameras from the same generation, the iso jump is due mainly to the physics. If you have the same number of pixels on a sensor that is 1.5 times bigger in each dimension, you automatically gain a full stop of iso - and that's without any jump in technology. So why stop at FX - make a bigger sensor with even better iso performance.

For me, DX works fine as I have a kid that plays sports and the resolution on the subject is better for a given lens. I really think the question should depend on what your intended uses are.
 
The per pixel noise quality not only on the high iso range, but the extremely smooth base iso was really appealing. The smaller DOF capabilities was also really nice. With the low pixel density of the 12MP sensor diffraction became almost a non-issue again, like in the film days. I think FX has a future with a brighter horizon than DX: with today's 16-18 MP DX sensors pixel density is very high, reaching the level of compact cameras, and that only means diffraction by f5.6 or around with no additional gain in resolution, as virtually no current lens can out resolve these sensors.
Most of what you "understand" is based on internet fiction :(

In prints up to 20 inches wide there is no easily detectable difference between D300 and D700 to about 1800 ISO - with perhaps 5% of the time DX being ahead by a nose, 5% FX being ahead by a nose, and 90% of the time - no difference.

The narrower dof has advantages (and disadvantages) but is of course only available wide open - when most lenses are not at their best - especially in the corners of FX.

Diffraction never was and never will be affected by either film grain size or MP - though it is affected by format size. Suggesting diffraction sets in by f5.6 on DX is completely wrong.

Also remember there are usually more wide aperture aberration losses at f4 - 5.6 than their are diffraction losses at f16 - f22.

There is a complete myth about lenses not being able to out resolve sensors. The laws of physics have not changed :)

The resolution you get is a blend of lens resolution and sensor resolution. When both are equal the resolution you get is half lens or half sensor resolution in isolation. Increase the resolution of either and you get more resolution in a finished print.

As the eye cannot detect all the detail that 12 MP can put in a 20 inch wide print viewed at a comfortable viewing distance we have reached a point where extra MP is more about cropping than anything else.

Nobody is allowed to publish what the D700 can do yet - but it obviously has more resolution than 12 MP.

Nikon got usefully more resolution and significantly better noise performance (maybe 2 stops) when they went from 10 MP D200 to 20% more MP in the 12 MP D300.

12 to 16 MP is a 33% increase so why not more resolution AND perhaps 1 stop better noise performance?

Lets wait and see - including who Nikon might source for an FX sensor between 12 and 24 MP.

--
Leonard Shepherd

Practicing and thinking can do more for good photography than buying or consuming.
 
The narrower dof has advantages (and disadvantages) but is of course only available wide open - when most lenses are not at their best - especially in the corners of FX.
Only available wide open? An 50mm lens for example at f5.6 will have less DOF on FX than on DX every time (provided you use the same field of view of course).
Diffraction never was and never will be affected by either film grain size or MP - though it is affected by format size. Suggesting diffraction sets in by f5.6 on DX is completely wrong.
Diffraction comes from the pattern of the Bayer layer. The smaller the pixels are, the smaller the holes in this layer are, therefore diffraction kicks in earlier. Format size has nothing to do with diffraction - only pixel density.
There is a complete myth about lenses not being able to out resolve sensors. The laws of physics have not changed :)
Is this a myth? Really? Well, try a 18-55 EFS lens on a Canon 7D... (just an example of a very poor lens on a very high density sensor)
Nobody is allowed to publish what the D700 can do yet - but it obviously has more resolution than 12 MP.
Feeling okay? :o
Leonard Shepherd

Practicing and thinking can do more for good photography than buying or consuming.
--
http://dimensionseven.deviantart.com
 
Hi DimensionSeven

Greetings from a fellow Hungarian:-)

I feel your pain, share your doubt.

I was also contemplating FX. But I decided I will stick with DX.

The new D3100 and especially the D7000 sensor shows that they can still improve high ISO a lot on DX sensors. These new sensors has at least 1 or 1,5 stops better ISO sensitivity than my D300 sensor. At ISO 100 low ISO noise also seems to be improved. (D300 is a bit noisy at ISO 200)

And with resolution at 14-16 Mpixel (it might go beyond that with D400 if it will be called like that) they are really amazing.

I think you only need FX if you really badly need very high ISO (this could happen but not with landscape) or if you need 24 Mpixel or more resolution.

I am full of DX lenses, Sigma 10-20, 18-50, 50-150, 30 1,4 etc.

To find and afford FX equivalent would be a big hurdle. I especially love the 50-150. One of the best for portres.

FX biggest problem is lenses, the sensor cost will go down sooner or later. But you will always need big expensive lenses.

If you need filters that much just stick with DX and buy a D7000. Seems it will be a wonderful camera.

Very nice landscape work otherwise congratulation!!!

János
 
...environnement with D300 and I feel it is OK until f8.

F11 is clearly worse but still OK for me (if I need more DOF)

From f16 and on it really becomes a problem.

Would a tilt-shift lens help you more? Landscape photographers use that a lot as I heard.

I lust for one for product photography. The professional one is surely, sadly out of my budget.
 
If your not a pro who might require DX for certain clients, I don't see any reason why the serious amateur would not be satisfied with DX, especially with a camera like the D7000.
--
Canon A2E, Sony R1, Panny TZ5, & Nikon D5000.
 
On FX, have you thought about the modest 18-35 AF f3.5-4.5? I have no fist hand experience (yet), but I have just bought a used copy on the net for less than 300USD. Bjørn Rørslett rates it at least as good on F5 as the Sigma 10-20 on DX. Reading his review, the 35 end seems good. I just have to see how the 18 end goes myself once I receive the lens.
 
Agree with Jerry R, DX is amply sufficient for the advanced amateur and reduces the cost and weight of the equipment.

Of course the D700 is a marvellous box, a real pleasure to use. Its weight with, lets say, a 70-200 f2.8, quite a load, can be made acceptable for walk around if you use a shoulder Black Rapid strap, but there is no way you can get around the cost.

If you are a BIF shooter, which I am, the loss of range is prohibitive and I find myself using the D300 much more than the D700, even with the 200-500 or the 500mm.

The other issue is the size of the files, but with fast and big computers that is a minor problem.

Finally the bokeh effect, easier with a larger processor, is not much of an issue if you buy f2.8 or lower glass with DX.

All and all, I think that DX is the reasonnable way to go unless pro use is at stake.

Peter
--
http://www.pbase.com/peter55/galleries
 
The narrower dof has advantages (and disadvantages) but is of course only available wide open - when most lenses are not at their best - especially in the corners of FX.
First, the corners often don't matter when shooting wide open. Second, at the same MP count, the FX sensor has less pixel density, meaning that the stress on lens resolution is not as high as on DX. And you will be able to stop down on FX one stop more than on DX for the same DOF, reducing lens aberration, improving performance.
Diffraction never was and never will be affected by either film grain size or MP - though it is affected by format size. Suggesting diffraction sets in by f5.6 on DX is completely wrong.
There's always diffraction. Even at f/4. Or f/2.8. The question is whether the sensor will record the effects of diffraction softening, and this clearly depends on pixel density, and hence on both sensor size and pixel number. Now of course, at f/16, a 16 MP DX won't be worse than a 12 MP DX or a 6 MP DX - but it won't be possible to fully exploit gain in resolution the higher MP count would make possible.
There is a complete myth about lenses not being able to out resolve sensors. The laws of physics have not changed :)
Sensor resolutions have changed, though...
Nobody is allowed to publish what the D700 can do yet - but it obviously has more resolution than 12 MP.
?!
 
I have both FX (D3) and DX (D300). At first I thought that these two offerings were complimentary, but recently I swinging more towards DX.

I have the 24mm PC-E lens, and on DX this produces a nice 3:1 pano with shifting: on FX this is only a 2:1 - not really pano IMHO.

Whilst I love the high ISO capabilities of FX, the reality is I hardly ever need this, or if I do, it's when using a long lens - but then I almost always want the extra reach/pixel-density of DX.

--
Tim

http://blog.timtucker.co.uk
http://www.timtucker.co.uk
 
Hi DimensionSeven
Hi Janó! :)
At ISO 100 low ISO noise also seems to be improved. (D300 is a bit noisy at ISO 200)
My findings exactly too. I have a D90 (practically the same as the D300 sensor), but didn't like the iso 200 outcome compared to the smoothness of the iso100 of the D80. Looked like it was grainy right at the beginning. However, by upping the iso or the shutter speed, the noise only increased very slowly, while the D80 is barely usable already at iso 400 for print or iso 800 for web use (talking about landscapes here).
And with resolution at 14-16 Mpixel (it might go beyond that with D400 if it will be called like that) they are really amazing.
Provided you have the glass to make it worth. My 16-85 VR or 10.5 fisheye are probably capable of 16MP, but not the Sigma 10-20. That was left behind at 12MP already.
Very nice landscape work otherwise congratulation!!!
Thanks, glad you enjoyed!

--
http://dimensionseven.deviantart.com
 
Hi

My Sigma 10-20 is quite sharp in the middle but never get very sharp at the edges.

Probably not the best for landscape.

Well wide angel in the DX format is still a challenge I don't know which lens would be the best. The 16-85 should be quite good just not wide enough. The 10-20 f3,5 is not as good as the old one the new Signa 8-16 might be very good (according to Photozone)

There is the Tokina 11-16 but Tokina lenses has CA issues.

But as you experienced the FX way is also not a lot better.

14-24 is fantastic just expensive and no filters :-(

Edge to edge sharpness on wide angel lenses is still a problem.
 
Diffraction comes from the pattern of the Bayer layer. The smaller the pixels are, the smaller the holes in this layer are, therefore diffraction kicks in earlier. Format size has nothing to do with diffraction - only pixel density.
The laws of optics - and my practical experience based on different resolving sensors and films - disagree :)

Digressing the size of the micro lens above each pixel has increased significantly over recent years - the main "enabler" of improved noise performance.

Put another way pixel area is dependent on the micro lens as well as the pixel pitch.

Perhaps you could try a web search for Rayleigh criterion to find out more about diffraction :)
Is this a myth? Really? Well, try a 18-55 EFS lens on a Canon 7D... (just an example of a very poor lens on a very high density sensor)
The laws of optics and my experience over 20 years show otherwise. Also remember the old Canon 18-55 comes from an era when Canon's minimum lens resolution was based on a 5x7 inch print.

Your presumption does not apply to something like the 60mm G on 6, 12 and 24 MP Nikon bodies - and it is unlikely to apply to a similar quality lens on Canon bodies with different pixel pitch and different pixel lens size.

--
Leonard Shepherd

Practicing and thinking can do more for good photography than buying or consuming.
 

Keyboard shortcuts

Back
Top