Slashdot Mirror
Scientists Create Easier Way To Embed Objects Into Video
Ashutosh Saxena writes "Stanford artificial intelligence researchers have developed software that makes it easy to reach inside an existing video and place a photo on the wall so realistically that it looks like it was there from the beginning. The photo is not pasted on top of the existing video, but embedded in it. It works for videos as well — you can play a video on a wall inside your video. The technology can cheaply do some of the tricks normally performed by expensive commercial editing systems. The researchers suggest that anyone with a video camera might earn some spending money by agreeing to have unobtrusive corporate logos placed inside their videos before they are posted online."
I thought there was more than enough advertisement on YouTube as it was already.
The researchers suggest that anyone with a video camera might earn some spending money by agreeing to have unobtrusive corporate logos placed inside their videos before they are posted online.
Just like web surfers no longer even glance at banner ads anymore, people will learn to ignore any corporate logos in videos (even if they really ARE there in real life!).
Does anyone notice that the more pervasive advertising is, the less effective it is?
In other words, people build filters for it. I know within younger generations, advertising is almost invisible.
I recall older people at work asking me "did you notice that new ad on the webpage?"
To which I responded "uhm... our webpage has ads?"
Because I spend enough time on the web to have almost totally filtered them out (yeah, adblock does a bunch of that for me, but even without it....)
I don't think I could tell you after a TV show, who the sponsors were. Commercial time is just blank in my mind because I tune it out.
I don't think I've EVER clicked on an ad in a webpage. I don't know for sure, but television and radio advertising rarely affect my purchasing decisions, at least not in a way I can discern.
So, legitimately, how powerful is a wall-hanging logo for Pepsi in some random goofy youtube video ACTUALLY going to be?
Am I a total oddity in not even noticing most advertising?
Modern advertising/branding isn't about actively convincing you anymore. It's about creating a pervasive environment of exposure in which you become familiar with a brand/product/logo whatever. In the store people are then more likely to subconsciously reach for Tide or Tylenol (despite the fact that generics are composed of essentially the same active ingredients) because they are familiar.
Nobody pays much attention to TV commercials anymore, and haven't for some time. Have advertisers markedly decreased their buying of TV commercial time? No, because you don't have to pay attention for it to work.
Couldn't I use this to remove the objects/logos/animations just as effectively? I would likely pay for that!
1. Initially from a computer forensics pov, it would be trivial to detect if a video has been altered, however i think with further improvements in the embedding technology where the actual advert piece is better rendered to take into account surrounding lighting conditions it might become more difficult, however not impossible to detect intentional modifications
2. Just as with current browser ad-blockers, the these ads can also be blocked out, in-fact the technology proves that complex camera conditions such as rotational pan(the heros examples) and occlusion (fat chick on couch) can be easily determined, so creating a blank out mask of a texture that is close to the surrounding surface would also be quiet doable, perhaps not real-time at the moment, but doable nonetheless, and most definitely live sometime in the future perhaps.
Arash Partow's Philosophy: Be a person who knows what they don't know, and not a person who doesn't know.
Actually you're wrong, conceptually its a similar process but the method is very different. The video processing done in football games requires 4+ cameras to be position around the arena, the camera locations are known exactly, furthermore the cameras are all high frame rate, hi resolution.
From the article it seems to suggest one can take some arbitrary video that has non-degenerative camera motion and embed images/video into it, doing this requires knowing the optical flow of all the moving objects in the scene, the camera motion and positions (to scale) from frame to frame - these are very difficult/different things.
In conclusion, going from deliberate views to arbitrary views is somewhat complex, so don't pooh-pooh the technology straight-away just because you don't understand it or have seen something similar before.
Arash Partow's Philosophy: Be a person who knows what they don't know, and not a person who doesn't know.
There are cases for generic and cases for non-generic.
There's nothing to say that brands are evil, just that brands with heavy advertising aren't necessarily more worthy of a purchase.
When it comes to drugs, generics actually ARE composed of essentially the same thing, but when it comes to card or liquor or computer parts, "generics" are definitely not.
However, purchasing a car bumper because the TV showed a hot woman rubbing on it is different than purchasing it because it's a superior product.
That's the contention this whole thread. Some people seem to think that advertising legitimately will make me pay more for an identical or inferior product, without my conscious knowledge, which I have argued as a bit of bunk, at least the enormous majority of the time, for me personally.