Discussion in 'Article Discussion' started by Gareth Halfacree, 13 Jan 2017.
Super-scaling system working a treat.
Surprisingly this method doesn't use neural networks, unlike many other methods of super-resolution.
Which is likely why it's up to 100 times faster than other super-resolution methods - I'm guessing, though, 'cos I haven't read the paper.
haha, finally when they say 'enhance' in movies it won't be 100% BS
My first thought to =p
google...plus ? this some wonderful new technology ? not seen anyone using it before.
So you reduce your data usage on phones, which is good, but offset it with increased battery drain, which is bad.
I wonder if it's optional for those having plenty of money for high data but who of course can't influence the battery life of their phone beyond how it arrived.
Not that I care too much for the 0.1% per se.
Remember data usage also uses power (especially LTE).
Sent from my SM-N915FY using Tapatalk
These superresoution algoithms have actually been around for a while (maybe a couple of years?), but for some reason they aren't that well known.
There's so muh cool stuff that's being done with deep neural networks right now. The hype train surrounding them is almost justified.
Bit more than a couple of years: the first super-resolution work was carried out in the 1980s.
I mean in it's current format. I mean, you could argue that bilinear sampling is superresolution.
Maybe I should say recent breakthrough, although the methods used in RAISR must have been around for a while, and RAISR seems to be as good as any other method for what I can tell.
So did I - it's just only recently that it's moved out of medical and industrial into consumer, 'cos now we've got systems powerful enough to do it on-the-fly.
Just goes to show, then!
Separate names with a comma.