Sam Gibbs writing for Gizmodo UK (sourcing data from a lengthy Eurogamer article):
It turns out that despite having 50 per cent more power in the GPU department, the in-game graphical performance of the surrogate PS4 only managed around 25 per cent faster frame rates, like-for-like in the gaming benchmarks. The interesting thing here is that the homebrew Xbox One test rig kept up with the PS4-like kit if the resolution was turned down from 1080p a smidgen, which makes me think that on the whole the two are going to be very evenly matched.
Pretty much all tests, signs and evaluations we’ve heard is that at least out of the gate you won’t notice much difference between the two systems in terms of raw graphical performance. But years down the line as graphical needs get pushed further upward I’m curious if the extra PS4 power will make a noticeable difference, at least on first party games.