A Camera That Prints Descriptions Instead of Pictures

Status
Not open for further replies.
so if a picture is worth a thousand words, then these guys type a pretty impressive 166 words a minute in order to get the results out in 6 minutes.
 
fail. Camera is not a printer, the correct way of describing this, a polaroid* camera that captures description rather than photo and can print as well.
 
This could be used for writers or Dungeon Masters. :) Instead of imagining places and trying to describe them, they could use this camera to take a "photo" of their room/basement and use this description.
 
This was a let down, I thought it was going to be some recognition software, instead it is taking a picture and having someone write a description...I can do that for free.
 
who does the description and how specific is it? Is it some geeky guy in a basement somewhere? He's like, she looks kinda hot, her boobs could be bigger and her nose is like weird and she is obviously NOT a blonde....(end description)
 
I am all in favor of research even if does not have any immediate use.
But no matter how hard I try, I can not imagine a practical use for this except as an engineering exercise.

 
I dont see the point of this. It sounded cool untill they mentioned Human Intelligence Task people. This is no magic/technology if there is a dude on the other end getting paid a dollar to describe a photo. I thought the camera itself was intelligent enough to interpret the picture it just took..
 
How about a photo printer that can print on the reverse side the meta data, iso, gps, date, time ect.
this is my idea copywrite. patent pending, my IP 😀 just to make sure...
 
Assuming it gives an accurate descrption of what the scene really contains.....i could see it being of use to blind people.

But i have my doubts on accurately describing a scene...
 
this has already been done using current and older gen cameras, using a modded firmware you can view the RGB values of each pixel, thus a description that completely describes the image

then using some scripting, you can take the text output of the rgb values, then create an image from it which can be printed and you will have a very accurate visual of what is in the pixel, with perfect pixel mapping (and best of all, no crowd sourcing needed since if the RGB values of each pixel is not enough, then an interpreter can map those values to actual pixels.
 
Status
Not open for further replies.