I know many people don't like DXO but this thread is not whether you agree with them and in particular with their sharpness score overall. (Also I know sharpness is not everything in a lens so no need to mention that too) Whether you like them or not or agree with their sharpness score or its methodology, their tests demonstrate a consistent pattern (and therefore constitute a fact for the pattern but perhaps not for the score itself) for their sharpness score, which they measure in terms of their so-called P-Mpix in the following respects:
1) As the Mpix of a sensor increases, depending on the quality of the lens, the P-Mpix of the lens increases with the combination of the lens with that sensor. E.g. the P-Mpix of 85mm 1.8G scores 20 with the 24 Mpix D600 but 22 with D800.
2) If the lens is not that good, for instance with most zooms, the P-Mpix score does not change at all. E.g. both the non-VR 24-70G and the new 24-70E VR scores the same with D600 and D800 (13 and 15, respectively)
3) In a given lens, regardless of the quality, the score also increases between the same sensors that has OLPF and not or cancelled out, as in D800 vs D800E/D810. However, if the lens is of better quality (such as the ones that falls under "1" above, the bump is more). E.g. 85mm 1.4.G's score of 20 with D800 increases to 30 with D810, whereas 24-70 and 24-70E's scores of 13 and 15 with D800 increase to 19 and 21. As you see in the case of 85mm 1.8G the bump is 10, whereas with 24-70 zooms the bump is 6 in terms of P-Mpix. (As a side note, it must be noted the difference is also dependent on how strong the AA filter is implemented. In D800, as far as I know, a rather weak AA filter has been implemented. If it were stronger, than the gap would be stronger in all combos. E.g. 85mm 1.4G might have scored 19 with D800 (instead of 20) but would still score 30 with D810, which would mean a bump of 11 instead of 10.)
So here is how Nikon's both 24-70s compare to the most recent Canon's offering:
Canon's latest 24-70 lens in combination with the 20.2 Mpix 6D and 22.3 Mpix 5D Mark III that both have an OLPF, and scores a P-Mpix of 17 and 18, respectively, which are higher that with the combo of either Nikon 24-70s with not only D600 but also D800 which shows no improvement. Whereas, as you see, even with 2.1 Mpix bump the P-mpix increased a point from 6D to 5D Mark III. Note that these Nikon and Canon sensors are from the same generation in 2012 which therefore obviates any argument that this is not a fair comparison. Also, Canon's lens score increases with 5DSR to 32 due to the higher Mpix of the sensor and cancelling out of OLPF thereof. All these indicate that Canon's latest 24-70, in my view, is a significantly better lens than even the latest 24-70E VR. Therefore, while its score with 5DS with OLPF is not currently available on DXO's website, if it were available, or if this lens could be tried on a D800, the score would have increased too, whereas both Nikon 24-70 stays the same despite the increase in Mpix of the sensor. To put in another way, both Nikon's 24-70s benefit from only the cancelling out or lack of OLPF, which is something very natural and applicable for even worst lenses (for the sake of the argument even the old generation and notoriously bad Nikon AF-S VR Zoom-Nikkor 24-120mm f/3.5-5.6G IF-ED's score increases a point with D810 compared to D600 and D800), whereas Canon's most recent 24-70 is up to par to some degree with prime lenses and therefore its sharpness increases not only with the cancelling out or lack of OLPF, but also with the increase of Mpix of the sensors, as in quality primes.
My conclusion is that Nikon's latest 24-70 VR offering, considering it was released after 3 years than that of Canon's, is a shame. While it seems Nikon tried to response Tamron's VR offering, it should have offered a lens like Canon's considering the higher resolution bodies of today and the questionable need for VR in the 24-70mm range. I think Sigma has a good opportunity here.
I would love to hear your comments on my analysis.
1) As the Mpix of a sensor increases, depending on the quality of the lens, the P-Mpix of the lens increases with the combination of the lens with that sensor. E.g. the P-Mpix of 85mm 1.8G scores 20 with the 24 Mpix D600 but 22 with D800.
2) If the lens is not that good, for instance with most zooms, the P-Mpix score does not change at all. E.g. both the non-VR 24-70G and the new 24-70E VR scores the same with D600 and D800 (13 and 15, respectively)
3) In a given lens, regardless of the quality, the score also increases between the same sensors that has OLPF and not or cancelled out, as in D800 vs D800E/D810. However, if the lens is of better quality (such as the ones that falls under "1" above, the bump is more). E.g. 85mm 1.4.G's score of 20 with D800 increases to 30 with D810, whereas 24-70 and 24-70E's scores of 13 and 15 with D800 increase to 19 and 21. As you see in the case of 85mm 1.8G the bump is 10, whereas with 24-70 zooms the bump is 6 in terms of P-Mpix. (As a side note, it must be noted the difference is also dependent on how strong the AA filter is implemented. In D800, as far as I know, a rather weak AA filter has been implemented. If it were stronger, than the gap would be stronger in all combos. E.g. 85mm 1.4G might have scored 19 with D800 (instead of 20) but would still score 30 with D810, which would mean a bump of 11 instead of 10.)
So here is how Nikon's both 24-70s compare to the most recent Canon's offering:
Canon's latest 24-70 lens in combination with the 20.2 Mpix 6D and 22.3 Mpix 5D Mark III that both have an OLPF, and scores a P-Mpix of 17 and 18, respectively, which are higher that with the combo of either Nikon 24-70s with not only D600 but also D800 which shows no improvement. Whereas, as you see, even with 2.1 Mpix bump the P-mpix increased a point from 6D to 5D Mark III. Note that these Nikon and Canon sensors are from the same generation in 2012 which therefore obviates any argument that this is not a fair comparison. Also, Canon's lens score increases with 5DSR to 32 due to the higher Mpix of the sensor and cancelling out of OLPF thereof. All these indicate that Canon's latest 24-70, in my view, is a significantly better lens than even the latest 24-70E VR. Therefore, while its score with 5DS with OLPF is not currently available on DXO's website, if it were available, or if this lens could be tried on a D800, the score would have increased too, whereas both Nikon 24-70 stays the same despite the increase in Mpix of the sensor. To put in another way, both Nikon's 24-70s benefit from only the cancelling out or lack of OLPF, which is something very natural and applicable for even worst lenses (for the sake of the argument even the old generation and notoriously bad Nikon AF-S VR Zoom-Nikkor 24-120mm f/3.5-5.6G IF-ED's score increases a point with D810 compared to D600 and D800), whereas Canon's most recent 24-70 is up to par to some degree with prime lenses and therefore its sharpness increases not only with the cancelling out or lack of OLPF, but also with the increase of Mpix of the sensors, as in quality primes.
My conclusion is that Nikon's latest 24-70 VR offering, considering it was released after 3 years than that of Canon's, is a shame. While it seems Nikon tried to response Tamron's VR offering, it should have offered a lens like Canon's considering the higher resolution bodies of today and the questionable need for VR in the 24-70mm range. I think Sigma has a good opportunity here.
I would love to hear your comments on my analysis.
