My untrained eye can't really notice a difference except that the pix version looks brighter (even washed out), the live photo version looks really uncanny valley (where the faces don't move but the world does around them is an unsettling effect).
My first thought when playing with the slider was that the Apple version was much better. But, if you focus on the faces (which was probably the intent of the person taking the photo) they do come out better in the Pix version. I think all the examples had the faces underexposed, so the "fix" for each of them was to brighten the photo.
Gotcha. It's a shame it seems to overexpose the background as a result though; that's something I could quickly reproduce in the photos app. Though if it's doing it automatically for each shot based on the faces in the photo I suppose it's novel and more convenient.
That's what I was thinking as well: it'd be really cool if they did some sort of automatic exposure bracketing and combined the best areas of each version since every one of their sample photos generally looks overexposed, especially compared to what Apple's built-in HDR mode would do.
They're also using models with multiple skin tones (dark vs light), which is a classic scenario for underexposed faces.
It's not even that the Pix version looks brighter -- it looks washed-out, like the evergreen trees are more of a blue-green color.
Agreed on the uncanny valley comment. It's interesting to me that very uniform "smile and look at the camera" is the desired outcome of pictures for so many. This page seems to work on the unspoken assumption that the perfect smile in front of some other thing is the ideal picture. I definitely prefer pictures of people doing something other than staring at a lens with a rehearsed smile. It seems that this and so much other camera algorithmic work is aimed at idealizing that posed picture style.
prefer + perfect = prefect?
D:
Most people only care about that smile :/
I felt the same way; I was amused because I consistently preferred the originals.
Downloaded it myself and tried it, and the difference is noticeable. I have to say, the result was better than what the default iOS camera app took. The Live Image is hit or miss (it really depends on what you are capturing), but it works.