Marginal gains or material advantage? My reflections

Hi,

I started a thread comparing GFX 50S, Sony A7rV, GFX 100 and Phase One IQ 4150, based on DPReview's test images. An image from Canon 5DIII was added in the aftermath.

Comparing detail:

The first part of the evaluation looked at detail in a 70x100 cm image at 200 PPI. This corresponds to my largest print size, at least regarding prints from a single image.

This image size corresponds to 43 MP, all systems compared with exception of the Canon 5DIII would deliver that detail.

It was not really obvious to me what difference viewers may have observed. My impression was that viewers did not find that any image 'stood out' on the initial demo.

If we look at MTF of the scaled images, we would find that the best performer was the GFX 100 and the worst performer was the Sony A7rV (exluding the Canon 5D3.

Images scaled for 70x100 cm at 200 PPI. Note that the MTF curves are very close up to say 1500 cy/PH. Above that there is significant advantage to the GFX 100. But human vision is more sensitive to low frequency detail than high frequency detail, so the difference in high frequency detal may not be very observable.

The 70x100 cm image at 200 PPI would hold 43 MP of data. An 8K image on screen weights in a 33 MP, so this image size is also a good representation of 8K screens of adequate size.

The GFX 100 was 'better' here than the Phase One IQ 4150 MP. There are two reasonable explanations for that:

• The Phase One XF may not have the focusing accuracy of the GFX 100. In general, EVF systems has more accurate AF as they use the sensor as focusing device. DSLR systems like the XF have AF-sensors in a substitue position with the image projected on the AF sensor by two moving mirros. External sensors are also more affected by focus shift.
• The other explanation would be that the Phase One lenses are older designs and not as well corrected as the GF lenses.

I have checked MTF data for another Phase One IQ 3100 MP image from DPReview using a different lens. That was slightly better, but still behind GFX 100.

Aliasing:

The DPReview studio scene has a couple of sections that demonstrate aliasing, most noticaeable as color moiré. Here the GFX 50S was worst, followed by the Sony A7rV. The GFX 100 and the Phase One IQ 4150 hade very little or no color moiré. That was partly expected.

I think that both the GFX 100 and the IQ 4150 'outresolve' the subject. but I may think that the IQ 4150 may be limited by the lens.

SNR:

I looked at SNR (signal noise ratio) on the images scaled for 70x100 cm.

Here the IQ 4150 'wins' and the Sony A7rV comes in last.

Tonal separation of two exposures 1% apart on the Sony A7rV (worst case)

Tonal separation of two tones 1% aprt on the IQ4150

I don't know what to make of it.

About DPReview test images:

Having a large set of reasonably well made test images of a large set of cameras is a godbless in my humble opinion.

There is often critique against their test shots. I don't know if it is deserved or not.

• Getting the optimal test image may need to have quite extensive focus bracketing. I don't know if DPR is doing that.
• They probably test gear sent to them by vendors, that gear may have issues.

Regarding the GFX 100, they put much extra effort into that test. The IQ 4150MP is their 'reference' shoot, so I guess they put some extra effort into that, too.