DXOMark's sharpness scores appear to be tied to the resolution of the test camera. So test on higher resolution camera and you get an apparently "sharper" lens. The RF lenses are bound to score lower because DXO is testing them on lower res cameras. DXO is inherently unreliable in this respect. According to DXO, the ultra-expensive mirrorless RF 50/1.2L is not the slightest bit sharper than the junky EF 50/1.4 from 1993. Both score 22 for sharpness. DXO even rates the buzzy-motor original EF 35/2 (from 1990!) as "sharper" than the new RF 50/1.2L Anybody who believes that should consider buying a large bridge in Brooklyn.
Of course, it is tied to the resolution of the test camera. This is true for all lenses. This indeed makes DXOMark's testing inherently more reliable because a lens sharpness not only depends on the lens itself but also on the camera used for the test and, DXOMark keeps that in mind when they test the lenses.
According to DXOMark, EF 50/1.4 USM scores 22 on the EOS 5Ds R and 20 on 5D mark iv. 5Ds R has a 50mp sensor while, 5D mark iv has a 30mp sensor which is probably the same sensor in the EOS R. This means the junky EF 50/1.4 lens is not quite as sharp as the new RF 50/1.2 but isn't quite far below either; only slightly less sharp. Is there anything wrong with this conclusion? I don't have any reason to think so.
The Sony FE 50mm/1.8 scores very high in sharpness. According to DXOMark, it scores sharper than some much more expensive lenses (though it may fall short in some other factors like CA, bokeh rendering, flare, etc.). Even worse, it costs a fraction of the price of those more expensive lenses.
That said, do we know how DXOMark recommend interpreting their scores relative to other lenses? I admit I have not checked on that. Maybe, they have their own recommendations.
No, the old 50/1.4 is far below the RF 50/1.2. The-Digital-Picture has comparisons in their Lens Image Quality tool. It's not even close. The old 50/1.4 at 1.4 is hazy throughout: center, middle and corners, topped off with purple fringing. The new RF 50/1.2 is neither hazy nor purple. It's 25 years newer and a big step up in quality. Yet on DXO they both score the same. That's not useful. If we relied on DXO, we could all save a lot of money by buying 1990's lenses, thinking they're just as sharp as the lastest big-priced lenses.
Jul 24, 2020 at 08:39 PM
Previous versions of NJPhotographer's message #15293693 « Sony-shooters thoughts on the Canon R5/R6 »
