DPReviews first look at the X1D - what I learned from it

[ErikKaffehr]

Hi,

DPReview has published a "first look" review of the X1D, including raw samples. So I checked them out, what did I find?

First I looked at top left corner. It contains a USAF test chart. Although USAF charts have bad reputation I feel they are quite nice and actually give a lot of good information. The cameras I choose here were:

• The X1D, the new Hasselblad
• The Fuji GFX, it's closest competitor
• The Canon 5DsR, same megapixels
• The Phase One IQ3-100MP - the real king

So, what did I see? The Fuji GFX is not very sharp in the corner. The X1D is nearly perfect. A big surprise is that the canon 5DsR with it's cheap 85/1.8 USM keeps up so well. The IQ3-100MP has twice the MP and it shows.

Now, let's look at aliasing. There is an area of the DPReview test target that is aliasing hell. Here I switch to comparison mode, that shows the IQ3-100 MP image downscaled:


Both the X1D and the GFX shows a lot of colour aliasing, indicating that the lenses outresolve the sensors. Well I guess Fuji was serious the lens was designed for 100MP. The Canon lens is probably a good match for the 50 MP sensor on the 5DsR. Some aliasing but not a lot.

Looking at the fine text:


Although the IQ3-100MP is robbed of it's resolution advantage, the benefits of the megapixels is clearly seen, even the finest text is excellently readable on the IQ3-100MP image downscaled to 50 MP.

Medium ISO noise in the darks. Here, I replaced the IQ3-100 MP with with the 42 MP Sony A7rII:



At 640 ISO and above, the Sony has a small trick in it's sleeve. Sony's late generation sensors have dual gain conversion.

The 44x33 mm sensors have 68% surface area, so they should have something like a 2/3 EV advantage. Lenses for the smaller formats are often faster and the Sony A7rII has image stabilisation.

I have also run some MTF tests at the centre, where we have wedge usable for SFR analysis. At the centre the GFX and the X1D were amazingly close. The Canon was just a bit behind, while the Sony was far behind. The IQ3-100 MP was quite a bit ahead the pack.

So, what is my take? The Hasselblad X1D and the Fuji GFX seem to be yield excellent quality and they will improve a lot when the new 100MP 44x33 sensors from Sony arrive. Personally, I am quiet a bit impressed by the Canon 5DsR using the 85/1.8 USM lens.

The IQ3-100MP is clearly ahead. That may change with 100 MP arriving in 44x33, but high end MFD is not standing still.

Anyway, the X1D and the GFX seem to be able to deliver great image quality at a competitive price. The Fuji GFX 63 mm lens looses some sharpness across the field. Off axis the X1D lens tested by DPReview may be a better choice. But, a test with a single lens doesn't tell the story about a full system

Best regards
Erik

 

[Quentin_Bargate]

Looking at these "tests", the X1D lens is slightly better at the edges, the GFX slightly better at the center. But neither camera was designed to shoot a DPreview test chart, and both would probably be a lot happier shooting useful photographs. You can no more assess a camera with these tests than you can assess a Monet on the thickness of the brush strokes.

Unsurprisingly, a camera with twice the number of pixels resolves better than two without, even though the two without are are capable of huge prints.

Does anyone care, I wonder?!

 

[siddhaarta]

Erik, thank you very much for this profound analysis.

We all know the limitations of test charts. Nevertheless they can give hints, including lens performance.

I personally am very impressed about the quality of the Hasselblad chart. Also the Fuji is excellent.

On the other hand I did not think that the IQ3100 has so little advantage nowadays (only visible highly enlarged), and even loosing ground with the Schneider lens which apparently has issues as regards contrast and edge sharpness (or is it a testing problem?).

Also I found the dynamic range link (included in the DPreview) quite interesting as it includes data on various P1 cameras, Hasselblad, Fuji GFX and Leica S 007:

http://photonstophotos.net/Charts/PDR.htm#FujiFilm GFX 50S,Hasselblad X1D-50c

 

[ErikKaffehr]

Hi Quentin,

Actually, I think the cameras are sort of designed to shoot test charts. The optical designers need references for developing their designs.

They start with a basic design, than they let the design program optimise the lens parameters. From the design they calculate a lot of MTF-data. The lens design program covers all the parameters.

Once a lens is built it goes on the optical bench and real lens images are analyzed and MTF charts are created and compared with predicted data.

At the final stage, test charts will be shot.

Now getting back to these test charts. They actually give a lot of information. As I said, they indicate a weakness in the corners on the Fuji lens. That is one single lens, so it doesn't say anything about the other lenses in the lens programme. The test chart would also show astigmatism, but neither lens has a lot of it. it also gives some indication about how well lateral chroma is corrected. It is well corrected on all those lenses.

As I have written, I have measured MTF at the centre and both lenses were very close. The amount of moiré indicates that the lenses resolve far beyond the 50 MP of the present cameras. Fuji has stated that the lenses are designed for 100 MP, and the test charts confirm it. So, the test charts tell a story.

Now, you can go both systems and shoot a couple of thousands of images to find out which one you prefer. If you do methodical testing you can do with a couple of hundred pictures.

Checking out a proper test gives you a lot of insight about the performance of a system.

Can a test chart give erroneous results? Yes, they can. Field curvature varies with focusing distance. Field curvature is most interesting at long distances, like infinity. Shooting test charts at infinity is not very practical. With common subjects field curvature does not matter a lot, as depth of field is often limited. With a curved field you still may have pretty optimal sharpness where you focus.

What I see from these tests is that either the Hasselblad or the Fuji would be a fine performer. But, I must say that I am impressed by the Canon 5DsR, too.

What the samples I have selected to show also indicate is that is some benefit from higher resolution. The 44x33 mm format will gain 100 MP, as Sony has released they roadmap and 100 MP 44x33mm is coming next year. The test indicates pretty clearly that the X1D and the GFX will be good hosts for that sensor. The best indicator of that may be the aliasing samples.

So yes, I would agree that a test like this doesn't tell the full story but it tells a lot. Another thing is that it gives us a lot of raw images to play with. Until now raw images have been scarce.

Best regards
Erik

Looking at these "tests", the X1D lens is slightly better at the edges, the GFX slightly better at the center. But neither camera was designed to shoot a DPreview test chart, and both would probably be a lot happier shooting useful photographs. You can no more assess a camera with these tests than you can assess a Monet on the thickness of the brush strokes.

Unsurprisingly, a camera with twice the number of pixels resolves better than two without, even though the two without are are capable of huge prints.

Does anyone care, I wonder?!

 

[hcubell]

Does anyone care, I wonder?!

Yes, the OP, who doesn't own any of the four cameras, and has no plans to buy any of them.

 

[SrMphoto]

The famous DPR scene tests are something I mostly look at after buying a camera . As a recent X1D owner, I was curious how it compares with Fuji GFX, a system I have not tried but assume behaves similarly as X1D. I do not understand the huge difference in one of the corners:

View attachment 126099

 

[jdphoto]

Yes, the OP, who doesn't own any of the four cameras, and has no plans to buy any of them.

Yes, but others might be. Those who are on the sidelines can make a more informative decision from these observations. I'd say the Nikon D810 gives MFD a run for its money in the last comparison and that's at the core of all this. Is it worth it to invest in the X1D, GFX, or Phase at these minimal differences? I think the color aliasing alone is worth the review. Thanks Erik, I think those that complain have already made their purchases.

 

[SrMphoto]

Yes, but others might be. Those who are on the sidelines can make a more informative decision from these observations. I'd say the Nikon D810 gives MFD a run for its money in the last comparison and that's at the core of all this. Is it worth it to invest in the X1D, GFX, or Phase at these minimal differences? I think the color aliasing alone is worth the review. Thanks Erik, I think those that complain have already made their purchases.

I do not use DPR scene tests to evaluate systems, but to see how they do on this scene test . I use both D810 and Hasselblad H5D, and IMO the cost and inconvenience of an MF system is justified by the image quality even though the scene test may not prove it. It is almost like testing music systems: there is the measurement and there is the listening test.

Note that Sony a7rII also exhibits suboptimal performance in that corner. I wonder if there is something wrong with the test.

 

[ErikKaffehr]

Hi,

Most lenses drop off in the extreme corners. Sony is sort of known for optical problems with their lenses.

The Canon was included in my comparison mostly because it is 50 MP, the Nikon test image is quite a bit behind. But including the Nikon would be a apples to oranges comparison because the images would need to be scaled.

Just to make a point, I looked at the test and wrote down what I see.

Best regards
Erik

I do not use DPR scene tests to evaluate systems, but to see how they do on this scene test . I use both D810 and Hasselblad H5D, and IMO the cost and inconvenience of an MF system is justified by the image quality even though the scene test may not prove it. It is almost like testing music systems: there is the measurement and there is the listening test.

Note that Sony a7rII also exhibits suboptimal performance in that corner. I wonder if there is something wrong with the test.

 

[D&A]

Erik wrote >>>"Can a test chart give erroneous results? Yes, they can. Field curvature varies with focusing distance. Field curvature is most interesting at long distances, like infinity. Shooting test charts at infinity is not very practical."<<<

Precisely! Two medium format lenses I work with regulary, one a wide and the other an ultra wide exhibit massive field curvature. Shooting a test chart at close range with either, reveals corners/sides that are just a small amount behind centeal sharpness when stopped down to f11 and beyond, but at infinity, sides and edges are so far behind the center, that they dont even come close no matter how much the lenses are stopped down. Of course the foreground of the sides and edges of those infinity shots are just as sharp as the center subject at infinity, thus field curvature of both is readily seen.

Its interesting to note that one of the lenses (the wide angle) had a slight redesign of its aspherical element in an update of the lens recently, and now field curvature is essentually gone and the lens is relatively flat field at any focusing distance. What a difference.

Therefore test charts do have practical applications in some circumstances but again with only the lens being tested at the time and only at the distance being utilized for that testing.

Dave (D&A)

 

[ErikKaffehr]

Hi,

It may be hard to say. I would notice two things:

• There may be different focus
• Specular reflections differ

Almost any lens has some field curvature, check this: https://www.lensrentals.com/blog/2014/09/field-curvature-and-stopping-down/

It is sort of natural that the field curves towards the camera at the edges. Vendors normally add an additional field flattening group to fast lenses. But field curvature varies with distance and aperture. I would think that Hasselblad X1D and the Nikon lens have field curvature, or lack thereof that matches the subject.

The other thing is that specular reflections are very sensitive to the angle. If the light source is fixed and the camera is moved different reflections will result. As focal lengths are not equivalent, the camera position moves, so specular reflections move.

Just to say, the GFX 63 mm lens may not be very sharp in the corner. Jim Kasson has run a comparison with Sony 55/1.8 ZA lensand he found:

"So the Fuji is markedly sharper except at the very corners, where the Sony comes close. As I said earlier, ACR/Lr force lens corrections for the 63, which probably reduce corner sharpness.

The corner performance of the Fuji system doesn’t surprise me, but the center and edge do. The GFX is quite a camera."

So the Fuji is great over the large part of the image, but corners are not so great. But they still are a bit above the well respected Sony 55/1.8.

Best regards
Erik

The famous DPR scene tests are something I mostly look at after buying a camera . As a recent X1D owner, I was curious how it compares with Fuji GFX, a system I have not tried but assume behaves similarly as X1D. I do not understand the huge difference in one of the corners:

View attachment 126099