Lenses for D800 - where's the weak link?

[ErikKaffehr]

Hi,

I got some good test files to measure MTF from Michael Ezra. My understanding is that Michael found the ZD to be superior while shooting models in studio, but found on the test shots that the Nikon had better detail.

I measured MTF on the Nikon 50/1.4 and the Mamiya 80/2.8 both at f/5.6 and plotted in the same diagram. The diagram shows the MTF (contrast transferred) for all frequencies up to twice Nyquist limit. This essentially means that the curves are comparable. Info past 4000 (ZD) or 5000 (Nikon) LW/PH is artifacts of SFR measurement.

This indicates that the Nikon should be able to achieve superior detail. The Mamiya curves can be pushed up with sharpening but so can the Nikon curves.

I don't know hoe this correlates with Michael's experience, but I guess that we all will learn something new checking these data, and that is always a good thing.

[ErikKaffehr]

Hi,

More observations.

Signal noise is better on the Nikon, but this depends much on exposure.

The Nikon D800E/Nikon 50 lens combo has much better MTF on a lp/mm basis. It is hard to say how much is coming from sensor and how much from lens.

Best regards
Erik

[MichaelEzra]

Hi Erik, thanks for looking into these files.
For the noise comparison, it might be more accurate to compare files from Sigma @5.6 and Mamiya @5.6 as their raw exposures were very close.
D800e was at 100 ISO and ZD at 50 ISO.

The black square of the slanted edge was not a very deep black, but a dark grey (as came out of my laser printer).
I carried raw development using low contrast & linear base tonal curves, do you think results would be different if higher contrast files were used?

Comparing these findings vs the charts from 25/2 Zeiss Distagon on the Nikon D800E is Zeiss that much better?
I see that the contrast of the Zeiss chart is much higher (but this could be due to raw conversion setting), do you think it could be a factor?

[ErikKaffehr]

Hi Ezra,

The contrast on the chart matters little, Imatest compensates for that.

I'm going to recheck the noise ratios on Sigma and ZD. The question is not about ISO, f-stop and shutter speed, but how far to the right the images are exposed.

My understanding is that you are feel that you got better detail on the ZD than on the D800. Checking with Imatest indicates the opposite, I have not really found any explanation. What is your take on the issue now?

I also checked the Nikon 50/1.4 and the ZD at 80/2.8, reason being mainly that I expect Mamiya with it's larger pixels being less affected by diffraction.

Regarding the Zeiss 24/2 sample, it was coming from Tim Parkin, I don't recall how it was "developed" but raw processing matters a lot. I'll try to check out about the sample. Right now I feel that it would be good to find out about the issue you have and not looking for the ultimate lens. The new Zeiss Distagon 55/1.4 seems to be a very good lens, but I think the price will be exceptional.

One of the reasons I am much interested in this that there are quite a few poster who claim that even old 25 MP backs are superior to todays CMOS based FF sensors. So I am very much interested finding out why and how.

Best regards
Erik

Hi Erik, thanks for looking into these files.
For the noise comparison, it might be more accurate to compare files from Sigma @5.6 and Mamiya @5.6 as their raw exposures were very close.
D800e was at 100 ISO and ZD at 50 ISO.

The black square of the slanted edge was not a very deep black, but a dark grey (as came out of my laser printer).
I carried raw development using low contrast & linear base tonal curves, do you think results would be different if higher contrast files were used?

Comparing these findings vs the charts from 25/2 Zeiss Distagon on the Nikon D800E is Zeiss that much better?
I see that the contrast of the Zeiss chart is much higher (but this could be due to raw conversion setting), do you think it could be a factor?

[JohnBrew]

There is a gentleman on one of the Nikon forums who adapted several Mamiya MF lenses for use on a D800. While he liked the image quality the weight differential put him off.
Best lenses on my D800 are in order:
1. 85 1.4G, my example is bitingly sharp, bokeh is a little nervous wide open, terrific when stopped down.
2. Zeiss 50 Makro, fabulous all-rounder
3. 105 2.5 Ai, don't discount it just because it's "old" - very sharp w/wonderful bokeh.
4. Zeiss 21, THE landscape lens and a big thank you to ACR for their lens correction tool

The only lens I own which was absolute garbage on the D800 is a Nikon 24 2.8 Ais. I would like a good 28 but the new 1.8G has too much field curvature for me.
My 2 cents.

[MichaelEzra]

My understanding is that you are feel that you got better detail on the ZD than on the D800. Checking with Imatest indicates the opposite, I have not really found any explanation. What is your take on the issue now?

On ZD I usually use F10 to get acceptable depth of field. Field curvature on Mamiya 80mmAF is not as curvy as on Nikon 50AFD1.4.
In my application field curvature is quite important - when I shoot a model laying (or flying:)) perpendicular to the lens axis, filling most of the frame length.
Comparison using models *could* be faulty, as models used with different systems were different, so there is variation in skin texture and hair thickness and some lighting variation as well.
Nevertheless, It appeared to me that if I shoot the same object using these systems not to preserve framing, but to maintain number of pixels per object dimension, ZD version would be sharper.
May be this is due to aliasing of the large pixels, hard to say.

[ErikKaffehr]

Hi Michael,

Aliasing would cause fake detail. The Nikon would also alias. You have a D800E?

I just run the 25/2 vs. 50/1.4 comparison. It seems that the 25/2 has an edge in MTF, specially at low frequencies (I will make a better plot later). I also ran the SQF figures. They are very close: if you made a 150x100 cm print and viewed at 95 cm you should not be able to see a difference between the two. Zeiss SQF = 90.5 Nikon SQF = 89. Supposedly a difference of five is discernible.

This was with no sharpening.

Best regards
Erik

[Rob C]

There is a gentleman on one of the Nikon forums who adapted several Mamiya MF lenses for use on a D800. While he liked the image quality the weight differential put him off.
Best lenses on my D800 are in order:
1. 85 1.4G, my example is bitingly sharp, bokeh is a little nervous wide open, terrific when stopped down.
2. Zeiss 50 Makro, fabulous all-rounder
3. 105 2.5 Ai, don't discount it just because it's "old" - very sharp w/wonderful bokeh.
4. Zeiss 21, THE landscape lens and a big thank you to ACR for their lens correction tool

The only lens I own which was absolute garbage on the D800 is a Nikon 24 2.8 Ais. I would like a good 28 but the new 1.8G has too much field curvature for me.
My 2 cents.

That’s interesting; I find mine quite good. Maybe I’ve had it for around fifteen years or so now, from new – I do feel that quality control has since gone down the tubes, though. This shouldn’t be allowed to happen; imagine if jet engines had a similarly low level of testing prior to use…

Rob C

[ErikKaffehr]

Hi,

The main advantage of the D800 is more pixels. I don't think that per pixel quality of the D800 is better than say a D600.

Best regards
Erik

Nevertheless, It appeared to me that if I shoot the same object using these systems not to preserve framing, but to maintain number of pixels per object dimension, ZD version would be sharper.
May be this is due to aliasing of the large pixels, hard to say.

[TMARK]

My old 60 AFd micro is very sharp. My old 28 2.8 ais is super. Very well corrected.

[ErikKaffehr]

Hi,

What is your problem with Canon? DR at low ISO? Resolution? Color rendition?

If you happen to have a decent MFDB, wouldn't a Hartblei H-cam or Alpa FPS be a good solution?

What is wrong with Nikon 14-24/2.8, Zeiss 18/3.5 and Zeiss 21/2.8? All those lenses seem to be decent and around 20 mm.

I don't have any personal experience of any of that stuff, but from what I have seen they are pretty attractive. All right, I have heard about field curvature on the Zeiss lenses, but not much else.

Sorry for asking. I can feel your frustration, just want to know what aspects it is about?

Best regards
Erik

I've been reading this thread with interest.

The following are a few comments I made elsewhere today. They seem relevant to this thread and sum up my frustration.

I look at files from the Canon 17 & 24mm TS-E lenses and weep. If only Canon made a body that took full advantage of them.

I'd literally go out and buy a Nikon D800E tomorrow if there was a lens of around 20mm that took full advantage of the camera.

I'm even considering buying a Canon 5D111 just to use the TS-E lenses. I'm that desperate.

[Fine_Art]

Hi,

I got some good test files to measure MTF from Michael Ezra. My understanding is that Michael found the ZD to be superior while shooting models in studio, but found on the test shots that the Nikon had better detail.

I measured MTF on the Nikon 50/1.4 and the Mamiya 80/2.8 both at f/5.6 and plotted in the same diagram. The diagram shows the MTF (contrast transferred) for all frequencies up to twice Nyquist limit. This essentially means that the curves are comparable. Info past 4000 (ZD) or 5000 (Nikon) LW/PH is artifacts of SFR measurement.

This indicates that the Nikon should be able to achieve superior detail. The Mamiya curves can be pushed up with sharpening but so can the Nikon curves.

I don't know hoe this correlates with Michael's experience, but I guess that we all will learn something new checking these data, and that is always a good thing.

Erik,

That sag in the lower frequencies is a typical curve of diffraction effect on a constricted lens. Was the Mamiya lens on the nikon camera with an adapter or was it on the ZD? With the thread title I assume its adapted and a large amount of the image circle is off the sensor. That is the price of adapting lenses. It's not that the mamiya lenses are inferior, it is diffraction.

[JohnBrew]

That’s interesting; I find mine quite good. Maybe I’ve had it for around fifteen years or so now, from new – I do feel that quality control has since gone down the tubes, though. This shouldn’t be allowed to happen; imagine if jet engines had a similarly low level of testing prior to use…

Rob C

Rob, interestingly enough it is fine on the D700.

[MichaelEzra]

Was the Mamiya lens on the nikon camera with an adapter or was it on the ZD? With the thread title I assume its adapted and a large amount of the image circle is off the sensor. That is the price of adapting lenses. It's not that the mamiya lenses are inferior, it is diffraction.

FYI, in this test Mamiya lens was used directly on Mamiya ZD body only. All other lenses were mounted on Nikon D800e without any adapters also.

[ErikKaffehr]

Hi,

The Mamiya lens is on ZD body (without adapter). The shot is at f/5.6 so I don't think diffraction plays a significant role. The sensors have different pitch 9 microns on the ZD and 4.8 microns on the D800E.

Best regards
Erik

Erik,

That sag in the lower frequencies is a typical curve of diffraction effect on a constricted lens. Was the Mamiya lens on the nikon camera with an adapter or was it on the ZD? With the thread title I assume its adapted and a large amount of the image circle is off the sensor. That is the price of adapting lenses. It's not that the mamiya lenses are inferior, it is diffraction.

[ErikKaffehr]

Hi,

Thanks for explaining, now I see what you mean.

Best regards
Erik

Hi, certainly resolution, I would have preferred the D800 sensor, but above all I've never particularly liked the Canon bodies.

I'm happy enough with my current MFD setup. I'm looking for a more versatile option; something that a tech camera or MFD couldn't offer.

They're decent enough but I was really hoping for a T/S solution/s.

Hey ho.

[risedal]

I've been reading this thread with interest.

The following are a few comments I made elsewhere today. They seem relevant to this thread and sum up my frustration.

I look at files from the Canon 17 & 24mm TS-E lenses and weep. If only Canon made a body that took full advantage of them.

I'd literally go out and buy a Nikon D800E tomorrow if there was a lens of around 20mm that took full advantage of the camera.

I'm even considering buying a Canon 5D111 just to use the TS-E lenses. I'm that desperate.

today you can buy the 6D which is the best camera sensor wise from Canon and use the 17mm and 24mm TS lenses.
Nikon have nothing to compare with regarding these 2 lenses

[Rob C]

Rob, interestingly enough it is fine on the D700.

That explains it: I have both D200 and D700 bodies.

I can't quite get the idea behind why a higher res. sensor should affect a lens so; yes, of course I understand the implications for larger blow-ups and the inevitable break down of information gathered, but at non-giant work of maybe A3 or thereabouts, which I imagine people generally work around for magazine use etc. why should the same lens appear less good? Or, are these negative comments entirely based on unrealistic pixel-peeping at reproduction sizes for which almost nobody ever achieves a market?

Rob C

[risedal]

Hi,

What is your problem with Canon? DR at low ISO? Resolution? Color rendition?

If you happen to have a decent MFDB, wouldn't a Hartblei H-cam or Alpa FPS be a good solution?

What is wrong with Nikon 14-24/2.8, Zeiss 18/3.5 and Zeiss 21/2.8? All those lenses seem to be decent and around 20 mm.

I don't have any personal experience of any of that stuff, but from what I have seen they are pretty attractive. All right, I have heard about field curvature on the Zeiss lenses, but not much else.

Sorry for asking. I can feel your frustration, just want to know what aspects it is about?

Best regards
Erik

They are no TS lenses.
14-24 is nice resolution wise but have severe problems with flare
Zeiss 18/3,5 nice but not excellent as 21/2,8 with more vignetting
Canons  two TS-E lenses are unique and they are gruesomely good

[Rob C]

Hi, certainly resolution, I would have preferred the D800 sensor, but above all I've never particularly liked the Canon bodies.

I'm happy enough with my current MFD setup. I'm looking for a more versatile option; something that a tech camera or MFD couldn't offer.

They're decent enough but I was really hoping for a T/S solution/s.

Hey ho.

Hi Keith,

I can't help wondering: if you get the right 135 format body and the lenses you need, would you then be happy to abandon your Hassy outfit in favour of lightness - if there's really much difference? I think that a tripod is probably as big a carriage problem as a larger format! Years before my first heart attack I discovered that carting my Gitzo G 410 around was more than my body could handle for longer than maybe half-an-hour.

Rob C