This technology looks like it has incredible promise. In a nutshell, it will erase all moving objects (like annoying tourists) from the frame when you take a picture.
Obviously, key to this is that the object of your affections stands stock still (like a monument tends to) and you have some form of optical stabilization on your camera (a tripod would do just fine but, my guess is, that stabilization can be built into the Monument Mode algorithm).
Promising. Watch the video below.
Been doing a more manual version of this for years with film/video in mokey, shake etc (and now AE), but having this as an auto feature in a phone is pretty phenomenal. How far technology has come in such a short time….
Holy crap. Not only annoying tourists, but possibly the most annoying demo I’ve ever seen. My first thought is, in order to get this to work, I have to tell all the annoying tourists to please walk around? Urg. My second thought is, I can’t wait for Apple to copy this because the demo will be actually informative and watchable.
So in the example shots he showed, where people were simply standing still and looking at the monument, this feature would be useless. And second, since it appears to need to be attached to a computer, how exactly is this going to help out in the real world? How many years away are we from our mobile processors being powerful enough to do this?
Pretty cool tech. We’ll have see if it works in the wild as well as it demos in a controlled example with two “tourists”. Suspect it won’t. Everything demos better with Nick Offerman. If it works, they should license the IP though. How many people will use a separate camera app for this one feature?
Unrelated, the demo for Project Faces was more immediately useful. Basically a Philosopher’s Stone for typography. Designers are going to use (and overuse) the heck out of that product when it’s ready. Surprised that one hasn’t generated more waves on the web.
Two ways to do this today:
Have your camera locked down on a tripod and take enough pictures, then stack the images and paint out the tourists in Photoshop (or other image editor) afterwards to reveal the monument behind them.
If you take a long enough exposure, the moving tourists will disappear automatically. If your exposure isn’t long enough or if the tourists don’t move fast enough, of course, you’ll get ghosts.
Either is a possibility from this demo,where the camera was locked down on a tripod. I’m guessing it’s not the latter, though, because you’re fighting the laws of physics with guessing what aperture and shutter speed to use ahead of time.
Is this like Apple’s Live Photos, where it sneaks a second and a half of extra pictures before and after you click, and then uses that to compare for where the tourists are and aren’t, painting them out automatically? If they don’t move, there’s not enough material in those three seconds to “paint out” the people. That’s my guess as to what’s happening.
In that case, this is an impressive “people detection” tech demo. Would be tricky to get right depending on clothing or what they’re carrying, I’d imagine.