yes, just the reviews led me to expect something more. interesting what you say about spider-man (haven't seen those films) but there was a distinct point during the superman reboot where- having previously done stuff like orbiting the world for fun- the action suddenly came down to a much more mundane 'realistic' level- catching cars and throwing punches. i think more a formalistic thing, like the director realised that having an all-powerful cgi hero would mean an essentially boring film. has anyone seen the older superman movies, i seem to remember one of those used more interesitng ways to get around this problem: the villain created synthetic kryptonite but his formula was slightly off, so rather than kill him it just made him really surly and disaffected (christopher reeve grew stubble and messy hair to convey this).
Very late to the Bladerunner party, but here is architecture office Barozzi Veigas original proposal which the movie uses (with approval of course) for Wallace's HQ:
EDIT: This is unfortunately not a built project, just a competition entry.
They don't use color in their visualizations to be able to focus on material and light. That's probably the best reason for doing B/W I have heard.
EDIT 2: More pictures:
Left: Barozzi Veiga, Right: Concept Art
I prefer the asymmetrical spaces of the competition proposal and concept art over what ended up in the movie. Changing the water (floor in the concept art) to an opening is nice.
Some guy trained a neural network on Balenciaga lookbook photos, some of the output results in this twitter thread:
This kind of thing was inevitable, but I suspect it's quite tricky to get the models right so it doesn't overfit the original outfits. Without an objective rating system for training the models you have to just trial and error tweak them for "better" results. Quite literally more art than science, which I think keeps it a bit more interesting than a lot of the style transfer neural network stuff that seems to be getting quite popular at the moment, and I suspect will be a boring gimmick quite rapidly.
Nightclub on a Friday Get my Gucci on F-f-f-firestarter Yes, I'm in the zone You coming on me like a player (WOOF!) But your game's all wrong You get your words all twisted It's the same old song
nope wrote:Some guy trained a neural network on Balenciaga lookbook photos...
The guy who did that Balenciaga neural network stuff, Robbie Barrat, does loads of other interesting neural network art as well. He makes much of his code available open source, and a French "AI Collective" used large chunks of that code to create a portrait that recently sold it for $435k without even crediting him. The license he released it under means that they are actually allowed to do that, but as some dude called Jason Bailey puts it in this pretty decent article on the situation “There’s a lot of stuff you can do that’s legal, but that makes you sort of a jerk”.
Seems kind of fucked up that they are spending all this money on a helmet that shows you Command & Conquer cutscenes instead of letting you see where you're flying the plane