NVIDIA's latest project - basic sketch to photo realistic image

Impressive, the end of the camera dawns :(
 
Hardly... You're forgetting the neural network needs to be trained on photographs of real scenes. Also, a real photograph will only suffice if you want to see an image of an actual place, not a virtually rendered one.
 
I can see applications in context aware removal / addition of objects etc personally. Remove unwanted Aunt Edna from a shot and adjust the shadow, reflection in the window etc too. Like I said, it's not there yet - it's one thing to 'paint by numbers' AI style, and another to do what I said above, but not so far outside the realms of possibility as it once was.
 
You must log in or register to reply here.
Back
Top