 |
|
|
|||||
Microsoft Research Asia has unveiled a new experimental AI tool called VASA-1 that can take a still image of a person or the drawing of one and an existing audio file to create a lifelike talking face out of them in real time. It has the ability to generate facial expressions and head motions for an existing still image and the appropriate lip movements to match a speech or a song. The researchers uploaded a ton of examples on the project page, and the results look good enough that they could fool people into thinking that they're real. While the lip and head motions in the examples could still look a bit robotic and out of sync upon closer inspection, it's still clear that the technology could be misused to easily and quickly create deepfake videos of real people. The researchers themselves are aware of that potential and have decided not to release "an online demo, API, product, additional implementation details, or any related offerings" until they're sure that their technology "will be used responsibly and in accordance with proper regulations." They didn't, however, say whether they're planning to implement certain safeguards to prevent bad actors from using them for nefarious purposes, such as to create deepfake porn or misinformation campaigns. The researchers believe their technology has a ton of benefits despite its potential for misuse. They said it can be used to enhance educational equity, as well as to improve accessibility for those with communication challenges, perhaps by giving them access to an avatar that can communicate for them. It can also provide companionship and therapeutic support for those who need it, they said, insinuating the VASA-1 could be used in programs that offer access to AI characters people can talk to. According to the paper published with the announcement, VASA-1 was trained on the VoxCeleb2 Dataset, which contains "over 1 million utterances for 6,112 celebrities" that were extracted from YouTube videos. Even though the tool was trained on real faces, it also works on artistic photos like the Mona Lisa, which the researchers amusingly combined with an audio file of Anne Hathaway's viral rendition of Lil Wayne's Paparazzi. It's so delightful, it's worth a watch, even if you're doubting what good a technology like this can do. This article originally appeared on Engadget at https://www.engadget.com/microsofts-ai-tool-can-turn-photos-into-realistic-videos-of-people-talking-and-singing-070052240.html?src=rss
Category:
Marketing and Advertising
Who hasnt spent sleepless nights pondering what would happen if we applied Vilfredo Pareto's (the early 20th-century Italian economist) theories to Mario, the Mushroom Kingdoms Italian high-jump champion and part-time elephant cosplayer? Data scientist Antoine Mayerowitz, PhD, tackled that age-old question, and the resulting work provides an objective way to tell us the best Mario Kart 8 racer combinations. Hint: It sure as hell aint Koopa Troopa. When you break down the build options (including driver stats and various vehicle details) in Mario Kart 8 Deluxe, there are over 700,000 possible combinations. Yikes. But once you eliminate duplicates that differ only in appearance, you can narrow it down to only 25,704 possibilities. How do you narrow it down to find the best racer from there? Enter Mr. Pareto. Paretos theories, most notably the Pareto front, help us navigate the complexities of choice. They can pinpoint the solutions with the most balanced strengths and the fewest trade-offs. Paretos work is about efficiency and effectiveness. Now were talking. Nintendo When choosing a Mario Kart racer, you have to consider their stats for speed, acceleration, handling, weight, offroad and mini turbo. Thats a lot to weigh. Even if you decide that speed and acceleration are the most important, youre still left with imbalances. For example, its tempting to go all in on speed (like Bowser or Wario), but they have weak acceleration. However, if you prioritize acceleration instead (such as Baby Mario or Dry Bones), you may be left with quick surges that plateau at a lousy top speed. Meanwhile, some racers are always dominated in the most important stats meaning their balance of speed and acceleration consistently comes out behind. Koopa is one example of that, so dont pick him if you care about winning. (But you can absolutely choose him because he has cute bug eyes and a snazzy shell.) Nintendo Mayerowitzs Pareto front analysis lets you narrow your possibilities down to the 14 most efficient. And it turns out the games top players were onto something: One of the combinations with the most ideal balance of speed, acceleration and mini-turbo is Cat Peach driving the Teddy Buggy, roller tires and cloud glider one already favored among Mario Kart 8 competitors. Of course, if that combination isnt your cup of tea, there are others that allow you to stay within the Pareto fronts optimal range. As Eurogamer points out, Donkey Kong, Wario (my old standby, mostly because he makes me laugh) and Princess Peach are often highlighted as drivers, and you can use Mayerowitzs data fields to find the best matching vehicles. Keep in mind that others have identical stats, so racers like Villager (female), Inkling Girl and Diddy Kong are separated only by appearances. To find your ideal racer, you can head over to Mayerowitzs website. There, you can enter your most prized stats and view the combos that give you the best balance (those highlighted in yellow), according to Paretos theories.This article originally appeared on Engadget at https://www.engadget.com/finally-someone-used-paretos-economic-theories-to-find-the-best-mario-kart-8-racer-211046789.html?src=rss
Category:
Marketing and Advertising
Rebel Moon: Part 2 - The Scargiver is an empty feast. It's a relentless onslaught of explosions, sci-fi tropes and meaningless exposition that amounts to nothing. And yet somehow it's still better than the first film in Zack Snyder's wannabe sci-fi epic franchise for Netflix, Rebel Moon: Part 1 - A Child of Fire. (What do these titles really mean? Who cares.) With all of the dull table-setting complete, Snyder is able to let his true talents soar in Rebel Moon: Part 2 by delivering endless battles filled with slow-motion action and heroic poses. It looks cool, I just wish it added up to something. Anything. Spoilers ahead for Rebel Moon: Part 2. If you somehow missed the first Rebel Moon film, the basic setup is that it's Star Wars meets The Seven Samurai. Sofia Boutella stars as Kora, a former elite soldier of an evil empire who is hiding out in an all-too idyllic farming village, just planting and harvesting her days away. When a group of military baddies kills the chief of the village and starts threatening a young girl, Kora goes on a murdering spree (in defense!), leaving the community open to a retaliatory attack. She spends the first movie recruiting potential warriors to defend the village, including a fallen gladiator (Djimoun Hounsou) and a bad-ass swordswoman (Doona Bae). (Their names are Titus and Nemesis, respectively, but those don't really matter because the characters are paper thin.) Full disclosure: I tried writing a review for the first Rebel Moon and just gave up in disgust. It was a shockingly boring epic, so much so that it took me several days to watch without falling asleep. By the end, I was only left with a feeling of dread, knowing that there was still another two hours of Rebel Moon ahead of me. It's somewhat empty praise, but at least I didn't fall asleep during The Scargiver. Mostly, that's due to the film actually having a sense of momentum and a lot more action. You can turn off your brain and enjoy the pretty pictures, much like you could for Snyder's Sucker Punch, Justice League and Watchmen adaptation. He's more a stylist than a natural storyteller, but occasionally Snyder's visuals, such as a baffling montage of our heroes harvesting wheat, can be almost poetic. Netflix It's just a shame that I didn't care much about the film's characters or any aspect of its story. James Gunn's Guardian's of the Galaxy trilogy made us fall in love with a band of misfits and screwups, with storylines that directly led to their personal and emotional growth. The crew in Rebel Moon, instead, feel like cardboard cutouts from better movies, and the overall plot feels forced (there's even setup for another film by the end). Hounsou tries to sell the pathos of Titus with his eyes, but he can only do so much. And while Bae's warrior woman exudes cool (and has a very compelling flashback), she's mostly wasted when the action really heats up. Then there's Jimmy, a robot voiced by Anthony Hopkins, who is briefly introduced in the first film and pops up for a few minutes here to kick butt. Why? It doesn't matter. Somehow that character is also important enough to serve as the narrator for both Rebel Moon films (but really it seems Snyder just wanted Hopkins' voice adding gravitas). Perhaps the only real saving grace for Rebel Moon: Part 2, much like the first film, is Ed Skrein as the villainous Atticus Noble. As a sadistic baddie, he's really nothing new, but Skrein's heightened scenery chomping makes the character interesting to watch. Where Darth Vader exudes a calm sense of dread, Skrein's Noble is entertainingly chaotic, like the Joker crossed with Christoph Waltz's Hans Landa from Inglorious Basterds. He just has a lot of fun being bad that's something! Given how popular the first film was (according to Snyder and Netflix, anyway), we'll likely see more Rebel Moon down the line. Snyder previously said he'd like to do a six-hour director's cut of both films, and he recently told Radio Times that he'd like to stretch the Rebel Moon series out to four or six films. Somehow, that just feels like a threat. This article originally appeared on Engadget at https://www.engadget.com/rebel-moon-part-2-review-a-slow-mo-sci-fi-slog-195505911.html?src=rss
Category:
Marketing and Advertising
All news |
||||||||||||||||||
|
||||||||||||||||||