Originally posted by falconeye DxO nowhere claims that the "lens" score is a lens score. They are pretty insistent to point out that their scores are for camera+lens combos.
This is their description of the lens test table.
"EXPLORE LENS SCORES & REVIEWS
DxOMark's comprehensive camera lens test result database allows you to browse and select lenses for comparison based on its characteristics, brand, price, lens type, lens size, focal range and aperture. You can also selet a camera in order to have all the lenses tested on it.
With DxOMark you can display the lenses with three different views - Grid, List or Graph in order for you to have the best solution to compare the filtered results. Choose the lenses you want to compare from the results below. Then in the selection box, you can check up to three lenses at a time to compare their measurements."
It would not hurt is they displayed camera measurements used to combine the score together with lens measurements, in the overview.
Quote: In the example you give, the Nikon wins because of two factors: sharpness breaks down later if you widen the aperture up. And the D750 has a much better DxO sensor score than the 5DS.
In other words, the Nikon lens/camera is better if you don't have infinite light while the Canon lens/camera is better if you do. That's what the two scores are good for.
The Nikon overall sensor score is 93 and the Canon is 86, so there is not a big difference in score. According to dxomark 5 point difference in score equals 1/3 stop difference. Low light ISO does not differ that much either (Nikon = 2956) (Canon = 2308)
And as dxomark only use low light to evaluate the lens, they would miss that maybe the largest advantage of D750 is DR at infinite light, so there might be more of an advantage of using D750 in this condition, than in low light where the two camera perform closer. So it seems that the disadvantage Canon has in camera score of low ISO is transferred to low light conditions in lens tests, if overall camera score is used in lens tests.
Quote: What I personally don't like is that DxO doesn't allow me to refactor the equation. I.e., I either trust their photographic assumptions or their score is useless to me. I would prefer a more detailed break-down of results allowing me to do the factorization myself. Like what they do with their three sensor scores.
I agree, and some of the lens test graphs are pretty useless.
Like FI the sharpness graph, where the scale goes from 1MP to 12+ MP, So on many cameras, lenses score will be "off the scale".