We need to hook up an AI with the library of bable image archive
This topic contains 11 replies, has 1 voice, and was last updated by John 1 week ago.
July 20, 2016 at 11:08 pm #7887
I am somewhat spookily hooked by the idea, that somewhere in the algorithm are images, never made by humans, though representing something to be interpretable by our mind…… or maybe an AI.
We should combine an AI with the image archive, searching for recognizable patterns, fishing out the raw gems.
Something maybe not for today, but for the future of quantum computers..
Spooky!August 26, 2016 at 3:17 am #8437
Howdy. So in the library is the source code of an IA.
you know i heard they were making a Watson application for things like this.June 21, 2017 at 8:22 pm #16875
I thought the exact same thing.
When you upload an image, the site is obviously somehow searching the library for a match to your image. If it can search for an entire image in the infinite possibilities, then it should be able to search for just *part* of an image. For instance, a face.
I would think the odds of finding an exact match of an entire image in the infinite possible entries of the library would be larger than the odds of finding just a face somewhere in an image. And, if it’s not a specific face but, rather, an AI’s general recognition of “a face”, then I would think the odds would be even better.
Of course, the more general you went in your search, the more hits you’d get and that could become unwieldy. But if a photo I take tomorrow is located somewhere in the library and can be found within a matter of seconds after I upload it, why couldn’t it just as easily find another image that’s 90% the same? Or a bunch?
If I’m misunderstanding how this works, please feel free to tell me so. I’m an artist and level designer for games, not a scientist. But, at first blush, this makes sense to me.June 21, 2017 at 11:32 pm #16883
@John Can you help me with making a game?June 24, 2017 at 9:28 am #16952
you are seriously wrong. first, the possibilities are not infinite. second, no searching takes place in case you are uploading a file, the image library is way too huge for that. simply a string encoding is what takes place.
well, you can build a supercomputer for the face recognition, just wait the several million years until it finds the first match in the tremendous meaningless chaos of the library. :^)
(I recommend not to answer Mark. he is an american furfag fapping to anthropomorphic cartoon characters, who posted his own mom into the image library, and lied about his apparent random finding, despite the overwhelming facts against him. horribly embarrassing.)June 24, 2017 at 10:59 pm #16960
@messiah How dare you, I’m a normal human being and I don’t care for animals!June 24, 2017 at 11:03 pm #16961
Wait that came out wrong… I have a beautiful human girlfriend.June 25, 2017 at 12:36 am #16964
post herJune 25, 2017 at 4:01 pm #16980
No, what’s wrong with you?
You have a dirty mind!July 16, 2017 at 11:11 pm #17396
I’m using the word “infinite” as shorthand for “I don’t have a finite number to give you, but I know it’s mind-bogglingly huge.” We all know what I mean. But if you want to get pedantic about it, I’ll amend my comment to “If it can jump to the location of an encoded string that matches an entire image in the 10^961755 possibilities, then it should be able to jump to the location of an encoded string for just *part* of an image.”
As for the rest of it, can you explain how what you said negates the question? Even if the Library’s simply taking my uploaded image, converting it into a string of numbers, and then jumping directly to the location of the matching string of numbers without searching in any way, I think the point stands. In the conversion of my image to the string, why could you not just take *part* of the string — the part that encodes, say, the face — and then jump to any number of locations that have that same “face” chunk within their numbers? If it’s just encoding the image to have a string to match, then it should be able to match a partial string, right? And I should be able to jump from picture to picture to picture of my same selfie face in the middle of other images (or noise patterns, as the case may be).July 16, 2017 at 11:30 pm #17397
Oh, and @Mark: I really can’t. I’m a level designer. Just a small-ish cog in a giant machine. I spend 112% of my time in a small, dark office making maps and levels for various games. I generally only stop once in a while to grab more coffee and remember sleep fondly.
If you have a pitch and a budget, though, I could probably hook you up with a producer. He and I started together at Activision back in the day. I think he’s transitioning into the movie biz now, but he may still be doing some game consulting on the side.
Fair warning though: He’s pulled me in on a number of his consulting gigs and the smallest budget we ever worked with was around $600,000. I don’t know if he’d be interested in going under that. Just being realistic.
If it’s about furries, on the other hand… ;>P