This is utterly insane. A research team from several universities has put together a system whereby you can draw a terrible stick picture sketch, label your blobby objects what they’re supposed to be, and it will essentially photoshop something together for you that meets your criteria. I kid you not. It’s early, experimental, and questionably useful, but it’s just too cool to not share with you guys.
Photosketch uses a sophisticated image filtering system (as well as humans) to pick objects that are what you’ve written and more or less in the position you want them to be. It then takes just the object and pastes it in over the background — whatever you’ve decided that is. It’ll come up with a few composite images, based on different parts, and then you get to choose the one you like. Are you kidding me?
I don’t want to be premature here, but I’d say tentatively that this does appear to be the greatest thing of all time.
They showed it at SIGGRAPH. If this works the way they say it does (and I doubt it does at this point, really), it’d be a party game all to itself. And can you imagine playing with this thing on a tablet? God damn! The future, people!
[via Metafilter]
That is amazing!!
This is amazing, I am surprised this article about it has yet to go viral. I am excited to try it, eventually.
so much win, so little time
Reminds me of a site circa 2006. http://labs.systemone.at/retrievr
It’s simply amazing…I wonder if this could be something Adobe might be interested in and possibly include in Photoshop somehow :)
Give the people a break please. They tried hard to release this product. You could have been more enthusiastic.
Totally ’shopped. I can tell by the pixels. And by seeing quite a few ’shops in my time.
What? You mean they didn’t run out and take a picture of a dinosaur chasing a monkey?
YOU’RE A FREAKING GENIOUS.
I believe Bill was joking, GENIOUS
It’s spelt GENIUS, you GENIUSES!! ;)
/learn-to-spell-if-you-wanna-play-on-the-internet blog.
Well fooey.
I know how it’s spelled. I was merely making fun of his spelling. I suppose I should have used more punctuation.
Your link to the research group (put together a system) is dead … and I mean dead .. dead
Fixed, thanks for alerting me.
I’m impressed. I think it would be even cooler if they added an option to automate somewhat less of it. The automations seem to work great, but advanced users should be able to pick and choose from the best results on their own if they don’t like the given options (“that’s great, but it would look more like I wanted it if I could just use THIS man with THAT dog”) and also drag the items around, maybe even resize and rotate them, themselves.
All they would need to do (for advanced mode only – what they’ve got is great for everyone else) is provide a photoshop (or GIMP) file with layers showing the top 5 or 10 results, and let you finish the photoshop yourself. In most cases you would probably go with one of their choices, but just providing the ability to fine-tune things would make this 10 times better.
“I think it would be even cooler if they added an option to automate somewhat less of it.”
That’s called Photoshop :)
yes, but photoshop isn’t free.
It would also be terribly inconvenient for a person to have to edit the images by cutting out objects and masking the holes.
where can I try this? it a web site or do I need to download a program or something?
nice, but the composites aren’t too realistic. No real shadows, lighting doesn’t match, etc.
The image extraction is impressive, tho.
wouldn’t mind that image extraction built into photoshop.
My only question, which I can’t find answered anywhere, is when this program/plugin is actually coming out for consumer use.
if they can do this with photographs now, in a few years, they will be able to do it with video.. Reminds me of that scene in Running Man when they fabricate Ben Richards’ death. :)
sweet! hope it makes it to market soon.
Not trying to be a jerk but I call fake.
Except it isn’t..
A better question is weather or not this program is randomly grabbing images available on line, and who’s copyrights they are infringing.
I’ve seen this done before – nothing new. Images look fake. Imagine cutting out a bunch of magazine pictures and pasting them on another magazine picture, all kinds of stuff like shadows etc. missing. You would have to do a LOT of manual editing to make it look anywhere close to realistic.
What they need to do first is put up a site that actually works. The link is forever dead.
Finally we can get xkcd in full-color!