h a l f b a k e r yBunned. James Bunned.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Please log in.
Before you can vote, you need to register.
Please log in or create an account.
|
| |
This would be a difficult piece of software to write. And if you tried handing it over to AI, you *might* find that there aren't enough Escher drawings of the right kind to use as training data - not least because it goes a step beyond what Escher did. |
|
| |
OTOH, that might make it an interesting challenge for someone in the AI field. |
|
| |
Actually I've done a couple of these in my youth, by hand obviously. There's a pretty straightforward explanation of how to do it in this Escher book I got as a kid. It really is just straight up geometry. |
|
| |
For starters, everything's a square. Now put a bump on one and a corresponding dent someplace else. Each figure would be analized as to what it has in the way of bumps and dents and put into a list. Keep in mind, pictures can be mirror imaged tweaked as necessary. |
|
| |
I couldn't do it but I assume somebody could. |
|
| |
Bear in mind that software still struggles a bit with "Which of these pictures has a traffic light in it?" |
|
| |
I hear ya, but this would just be taking that new thing on the iPhone where you hold it down and it knows what picture is the person and eliminates the background. That would be a large part of this. |
|
| |
Then the next step is analyzing those "holes" and "bumps" and figuring how much you'd have to nudge them or leave in a little of that background to get them fit together. |
|
| |
You know, an easy thing to do might be to have the camera guide you into how to pose so the pieces could be fit together. |
|
| |
This has been done already. At Ford we did a bunch of employee photos to do a huge picture of the Fusion when it hit the market. I had a scaled down pic of it on my fridge. |
|
| |
I also have a JiGaZo puzzle set which is a different sort of idea; you feed the software a picture to duplicate and it tells you what pieces to put where from its set of square pieces. |
|
| |
And the pictures are tessellated? That is, the piece is shaped like the person's outline and its arm fits seamlessly into a hole in another person's outline? (get your mind out of the gutter, you know what I mean) |
|
| |
Now to be clear, this isn't just taking a picture and turning it into a puzzle, it's automatically tessellating images so they fit together. |
|
| |
Is there an AR puzzle solver app yet? It would let you look through your phones camera and highlight which piece goes where |
|
| |