Subject: Re: Absolutely Perfect Predictions
From: Keith Morrison
Newsgroups: rec.arts.sf.written
Ian Montgomerie wrote:
It might be able to get more information from adjacent frames
and use that in combination with what you described below.This is already possible. Motion-compensated deinterlacing with HD
upsampling. However, the previous poster is right - you can’t apply
any reasonable image filtering techniques to make it look like it was
actually shot in HD. The fine detail is lost, and there are harsh
limits on how much of it you can interpolate back in (without
tremendous advancements in AI and Arbitrary Computing Power (TM)).
What? You mean all those shows where they have an image that looks like it was shot on the equivalent of a 0.25 megapixel camera and
ten seconds of snazzy computing later they’ve extracted details at a resolution of about 0.5mm from about 300m away aren’t realistic?
I am just shocked.
I’m waiting for a scene in any show where the Heroes are standing behind the computer tech looking at a screen.
"It’s a blobby shape in the driver’s seat."
"Can you blow it up?"
(Tech looks at speaker funny and shrugs.)
"O-kay. There you go. Now it’s a big blobby shape in the driver’s seat."