Connect with us

Technology

Meta enters AI video wars with powerful Movie Gen model

Published

on

Meta enters AI video wars with powerful Movie Gen model

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More


Meta founder and CEO Mark Zuckerberg, who built the company atop of its hit social network Facebook, finished this week strong, posting a video of himself doing a leg press exercise on a machine at the gym on his personal Instagram (a social network Facebook acquired in 2012).

Except, in the video, the leg press machine transforms into a neon cyberpunk version, an Ancient Roman version, and a gold flaming version as well.

As it turned out, Zuck was doing more than just exercising: he was using the video to announce Movie Gen, Meta’s new family of generative multimodal AI models that can make both video and audio from text prompts, and allow users to customize their own videos, adding special effects, props, costumes and changing select elements simply through text guidance, as Zuck did in his video.

Advertisement

The models appear to be extremely powerful, allowing users to change only selected elements of a video clip rather than “re-roll” or regenerate the entire thing, similar to Pika’s spot editing on older models, yet with longer clip generation and sound built in.

Meta’s tests, outlined in a technical paper on the model family released today, show that it outperforms the leading rivals in the space including Runway Gen 3, Luma Dream Machine, OpenAI Sora and Kling 1.5 on many audience ratings of different attributes such as consistency and “naturalness” of motion.

Meta has positioned Movie Gen as a tool for both everyday users looking to enhance their digital storytelling as well as professional video creators and editors, even Hollywood filmmakers.

Movie Gen represents Meta’s latest step forward in generative AI technology, combining video and audio capabilities within a single system.

Specificially, Movie Gen consists of four models:

Advertisement

1. Movie Gen Video – a 30B parameter text-to-video generation model

2. Movie Gen Audio – a 13B parameter video-to-audio generation model

3. Personalized Movie Gen Video – a version of Movie Gen Video post-trained to generate personalized videos based on a person’s face

4. Movie Gen Edit – a model with a novel post-training procedure for precise video editing

Advertisement

These models enable the creation of realistic, personalized HD videos of up to 16 seconds at 16 FPS, along with 48kHz audio, and provide video editing capabilities.

Designed to handle tasks ranging from personalized video creation to sophisticated video editing and high-quality audio generation, Movie Gen leverages powerful AI models to enhance users’ creative options.

Key features of the Movie Gen suite include:

Video Generation: With Movie Gen, users can produce high-definition (HD) videos by simply entering text prompts. These videos can be rendered at 1080p resolution, up to 16 seconds long, and are supported by a 30 billion-parameter transformer model. The AI’s ability to manage detailed prompts allows it to handle various aspects of video creation, including camera motion, object interactions, and environmental physics.

Advertisement

Personalized Videos: Movie Gen offers an exciting personalized video feature, where users can upload an image of themselves or others to be featured within AI-generated videos. The model can adapt to various prompts while maintaining the identity of the individual, making it useful for customized content creation.

Precise Video Editing: The Movie Gen suite also includes advanced video editing capabilities that allow users to modify specific elements within a video. This model can alter localized aspects, like objects or colors, as well as global changes, such as background swaps, all based on simple text instructions.

Audio Generation: In addition to video capabilities, Movie Gen also incorporates a 13 billion-parameter audio generation model. This feature enables the generation of sound effects, ambient music, and synchronized audio that aligns seamlessly with visual content. Users can create Foley sounds (sound effects amplifying yet solidifying real life noises like fabric ruffling and footsteps echoing), instrumental music, and other audio elements up to 45 seconds long. Meta posted an example video with Foley sounds below (turn sound up to hear it):

Trained on billions of videos online

Movie Gen is the latest advancement in Meta’s ongoing AI research efforts. To train the models, Meta says it relied upon “internet scale image, video, and audio data,” specifically, 100 million videos and 1 billion images from which it “learns about the visual world by ‘watching’ videos,” according to the technical paper.

Advertisement

However, Meta did not specify if the data was licensed in the paper or public domain, or if it simply scraped it as many other AI model makers have — leading to criticism from artists and video creators such as YouTuber Marques Brownlee (MKBHD) — and, in the case of AI video model provider Runway, a class-action copyright infringement suit by creators (still moving through the courts). As such, one can expect Meta to face immediate criticism for its data sources.

The legal and ethical questions about the training aside, Meta is clearly positioning the Movie Gen creation process as novel, using a combination of typical diffusion model training (used commonly in video and audio AI) alongside large language model (LLM) training and a new technique called “Flow Matching,” the latter of which relies on modeling changes in a dataset’s distribution over time.

At each step, the model learns to predict the velocity at which samples should “move” toward the target distribution. Flow Matching differs from standard diffusion-based models in key ways:

Zero Terminal Signal-to-Noise Ratio (SNR): Unlike conventional diffusion models, which require specific noise schedules to maintain a zero terminal SNR, Flow Matching inherently ensures zero terminal SNR without additional adjustments. This provides robustness against the choice of noise schedules, contributing to more consistent and higher-quality video outputs  .

Advertisement

Efficiency in Training and Inference: Flow Matching is found to be more efficient both in terms of training and inference compared to diffusion models. It offers flexibility in terms of the type of noise schedules used and shows improved performance across a range of model sizes. This approach has also demonstrated better alignment with human evaluation results.

The Movie Gen system’s training process focuses on maximizing flexibility and quality for both video and audio generation. It relies on two main models, each with extensive training and fine-tuning procedures:

Movie Gen Video Model: This model has 30 billion parameters and starts with basic text-to-image generation. It then progresses to text-to-video, producing videos up to 16 seconds long in HD quality. The training process involves a large dataset of videos and images, allowing the model to understand complex visual concepts like motion, interactions, and camera dynamics. To enhance the model’s capabilities, they fine-tuned it on a curated set of high-quality videos with text captions, which improved the realism and precision of its outputs. The team further expanded the model’s flexibility by training it to handle personalized content and editing commands.

Movie Gen Audio Model: With 13 billion parameters, this model generates high-quality audio that syncs with visual elements in the video. The training set included over a million hours of audio, which allowed the model to pick up on both physical and psychological connections between sound and visuals. They enhanced this model through supervised fine-tuning, using selected high-quality audio and text pairs. This process helped it generate realistic ambient sounds, synced sound effects, and mood-aligned background music for different video scenes.

Advertisement

It follows earlier projects like Make-A-Scene and the Llama Image models, which focused on high-quality image and animation generation.

This release marks the third major milestone in Meta’s generative AI journey and underscores the company’s commitment to pushing the boundaries of media creation tools.

Launching on Insta in 2025

Set to debut on Instagram in 2025, Movie Gen is poised to make advanced video creation more accessible to the platform’s wide range of users.

While the models are currently in a research phase, Meta has expressed optimism that Movie Gen will empower users to produce compelling content with ease.

Advertisement

As the product continues to develop, Meta intends to collaborate with creators and filmmakers to refine Movie Gen’s features and ensure it meets user needs.

Meta’s long-term vision for Movie Gen reflects a broader goal of democratizing access to sophisticated video editing tools. While the suite offers considerable potential, Meta acknowledges that generative AI tools like Movie Gen are meant to enhance, not replace, the work of professional artists and animators.

As Meta prepares to bring Movie Gen to market, the company remains focused on refining the technology and addressing any existing limitations. It plans further optimizations aimed at improving inference time and scaling up the model’s capabilities. Meta has also hinted at potential future applications, such as creating customized animated greetings or short films entirely driven by user input.

The release of Movie Gen could signal a new era for content creation on Meta’s platforms, with Instagram users among the first to experience this innovative tool. As the technology evolves, Movie Gen could become a vital part of Meta’s ecosystem and that of creators — pro and indie alike.

Advertisement

Source link
Advertisement
Continue Reading
Advertisement
Click to comment

You must be logged in to post a comment Login

Leave a Reply

Technology

Lego’s website was hacked to promote a crypto scam

Published

on

Lego's website was hacked to promote a crypto scam

People who visited Lego’s website on the evening of October 4 were welcomed by a banner with illustrated golden coins bearing the company’s logo, claiming that the “Lego coin” is now officially out. It even promised “secret rewards” to those who’d buy some. But Lego wasn’t truly launching an official cryptocurrency coin, and according to The Brick Fan, the button to buy led to an external cryptocurrency website selling “LEGO Tokens” with Ethereum. The website was, seemingly, hijacked by bad actors who switched its banner and used it for some sort of crypto scam.

As users on the Lego subreddit have noted, the incident happened overnight for Lego’s headquarters. The company responded relatively quickly, though, and removed the unauthorized banner and links. As of this writing, the Lego Fortnite collaboration banner is back up, and the “buy now” link leads to the collection. Lego told Engadget that no user accounts were compromised and that it has identified the cause of the issue. It also said that it was implementing measures to prevent anything similar from happening again in the future. However, the company has declined to share details about that “cause” or the measures it’s implementing.

Here’s the company’s official statement:

“On 5 October 2024 (October 4 evening in the US), an unauthorised banner briefly appeared on LEGO.com. It was quickly removed, and the issue has been resolved. No user accounts have been compromised, and customers can continue shopping as usual. The cause has been identified and we are implementing measures to prevent this from happening again.”

Source link

Advertisement

Continue Reading

Science & Environment

Fed rate cuts should favor preferred stocks, Virtus fund manager says

Published

on

Fed rate cuts should favor preferred stocks, Virtus fund manager says


A place for "preferred" stocks

One financial firm is trying to capitalize on preferred stocks – which carry more risks than bonds, but aren’t as risky as common stocks.

Infrastructure Capital Advisors Founder and CEO Jay Hatfield manages the Virtus InfraCap U.S. Preferred Stock ETF (PFFA). He leads the company’s investing and business development.

“High yield bonds and preferred stocks… tend to do better than other fixed income categories when the stock market is strong, and when we’re coming out of a tightening cycle like we are now,” he told CNBC’s “ETF Edge” this week.

Hatfield’s ETF is up 10% in 2024 and almost 23% over the past year.

His ETF’s three top holdings are Regions Financial, SLM Corporation, and Energy Transfer LP as of Sept. 30, according to FactSet. All three stocks are up about 18% or more this year.

Advertisement

Hatfield’s team selects names that it deems are mispriced relative to their risk and yield, he said. “Most of the top holdings are in what we call asset intensive businesses,” Hatfield said.

Since its May 2018 inception, the Virtus InfraCap U.S. Preferred Stock ETF is down almost 9%.



Source link

Continue Reading

Servers computers

#shorts Review tủ rack kỹ thuật 10U

Published

on

#shorts Review tủ rack kỹ thuật 10U

source

Continue Reading

Technology

Look North World recreates Hasbro titles in Fortnite, starting with Clue

Published

on

Look North World recreates Hasbro titles in Fortnite, starting with Clue

UGC game studio and publisher Look North World announced today it is partnering with Hasbro to bring three of the latter’s board games to Fortnite. Specifically, Look North World will recreate the games as islands via UEFN, allowing players to enjoy recreations of the familiar gameplay. The first game it’s adapting is Clue, which launches today as the island Murder Mystery: Clue. Other islands based on Guess Who and Connect Four are planned to launch later in October and December, respectively.

Murder Mystery: Clue uses Fortnite’s assets to recreate the essential elements of Clue. Gameplay takes place in timed rounds, where players vote on maps and are then assigned the roles of Killer, Detective, or Guest secretly. Each player has their own agenda and win conditions — presumably the Killer’s is to do the slaying in a particular fashion while the Detective’s is to discover their dastardly deeds.

Eugene Evans, SVP of digital strategy and licensing at Hasbro and Wizards of the Coast, said in a statement, “Bringing our classic games into new mediums like Fortnite is a key strategy as we continue to grow our digital games portfolio through both licensing and internal development. We start with games that have attracted fans for decades, and when we partner with a studio like Look North World, we know they understand the essence of what makes these games resonate with players.”

Hasbro has found success licensing its board games for digital experiences — Scopely’s Monopoly Go, for example, continues to grow and expand its dedicated audience. It’s also worked with Look North World in the past, participating in its funding in July.

Advertisement

Join us for GamesBeat Next!

GamesBeat Next is almost here! GB Next is the premier event for product leaders and leadership in the gaming industry. Coming up October 28th and 29th, join fellow leaders and amazing speakers like Matthew Bromberg (CEO Unity), Amy Hennig (Co-President of New Media Skydance Games), Laura Naviaux Sturr (GM Operations Amazon Games), Amir Satvat (Business Development Director Tencent), and so many others. See the full speaker list and register here.


Alex Seropian, Look North World’s CEO, said in a statement, “Look North World moves at the speed of culture to deliver the experiences gamers want, on the platforms where they are already playing and creating. We are excited to work with Hasbro, a brand that understands the power of user-generated content as a unique opportunity to connect with passionate gamers. Hasbro is embracing community-driven trends—bringing iconic games like Clue, Guess Who and Connect Four into the spaces where players are most engaged.”


Source link
Continue Reading

Technology

SoCreate wants to transform screenwriting software with AI imagery and community sharing tools

Published

on

SoCreate wants to transform screenwriting software with AI imagery and community sharing tools

Many screenwriters have embraced modern tools over traditional PDFs to craft their film or TV show pilots. SoCreate, the latest entrant in the screenwriting software arena, is challenging established players like Final Draft and Celtx with its fresh approach to storytelling. And, notably, generative AI imagery is involved.

SoCreate offers many of the same features that most screenwriting software offers, such as templates to easily create an industry-standard screenplay with correct formatting. However, founder and CEO of SoCreate, Justin Couto, believes popular platforms are still lacking, particularly when it comes to visual and creative tools.

“When I decided to go to college, I found myself gravitating towards film, which meant I needed to dive into the art of screenwriting. I immediately found the process to be dull and uninspired. It was like, we’re writing for a visual medium for movies and TV, but I have to use this archaic black-and-white document with outdated formatting based on the typewriter? I knew there had to be a better way — a more visual, fun, creative way,” Couto told TechCrunch.

SoCreate thinks one of its big selling points is its image uploader tool for screenwriters to incorporate visual concepts into their scripts, including characters, settings, and action moments. Users have the option to upload their own images or select from SoCreate’s gallery of illustrations.

Advertisement

Soon, users will be able to use an AI-powered image generator to create imagery, which will be powered by a combination of models, including OpenAI, Stable Diffusion, and others. It’s important to note that SoCreate has no plans to offer AI-generated writing tools. The image generator is solely to inspire users while writing and make the process less monotonous. 

Image Credits:SoCreate

Another standout feature is “Storyteller,” which the platform launched earlier this week. Storyteller is a dedicated hub where users can share their stories in a public library for others to read. This new feature is reminiscent of Wattpad, allowing a community of readers to access scripts for free, written by both established and aspiring writers.

The company believes Storyteller will help aspiring screenwriters market their work more effectively, building a public, “visually stunning” portfolio without needing Hollywood connections that aren’t readily available. 

However, some screenwriters may prefer not to make their scripts public for fear of being plagiarized. Users have the option to keep their work private on SoCreate, and the platform uses encryption. Additionally, there is a strict policy against plagiarism. It’s always advisable to register your work with organizations such as the Writers Guild of America or the U.S. Copyright Office.

“My personal theory and this is not legal advice, is that publishing your work online publicly protects you from plagiarism in many ways; you have timestamped proof that you were the original writer of the work and hundreds or thousands of eyes on the work that saw it on SoCreate first. A PDF doesn’t really give you that,” Couto argues.

Advertisement

Couto envisions Storyteller to become more than just a reading experience. In the future, it’ll add the ability to include AI-generated character voices, sound effects, and background music. Final Draft’s latest update includes an option where users can assign characters’ voices to read the script. 

Image Credits:SoCreate

In addition, readers can leave comments under scripts, giving them the ability to provide instant feedback when previously screenwriters were accustomed to exporting to PDF and emailing it. Users can share a link to any part of their story, from a single piece of dialogue to the entire thing, and readers can write their notes or suggestions without needing a SoCreate account.

Another standout feature is Reading Stats, letting screenwriters see if someone actually read their story, where they stopped reading, how long they spent reading, and where they left comments. 

The platform is mainly catered to people writing movies, TV shows, and short films. However, the company is also exploring templates for articles, novels, and short stories, broadening its reach to more creatives. 

“Once we nail narrative storytelling, we’ll move into new verticals, including business, education, journalism, lifestyle, and research. As readership grows, we’ll add subscriptions to access the SoCreate library, and creators will have a new opportunity to earn from their work through revenue-sharing with SoCreate,” Couto said. 

Advertisement

SoCreate launched last May and has garnered over 1,200 subscribers. Of its users, the platform says that some are writers who produced work for Amazon, Disney, Marvel, and Netflix. It also runs pilots and other programs with select high schools in California and Illinois. 

The platform is free for all users, but if they want to access the custom image tool and reviewer stats, they will have to spend $10/month for the Professional subscription. There’s also a Personal tier for $5/month, which includes unlimited projects and access to SoCreate’s image gallery. 

The company closed a $3 million pre-seed round last year and is currently raising a $5 million seed round that will be used for development and marketing.

Source link

Advertisement

Continue Reading

Servers computers

HP Server Rack and Stack – Service

Published

on

HP Server Rack and Stack - Service



Empower your business with our professional server installation service. Our experienced team specializes in deploying reliable and efficient server solutions tailored to your specific needs.

We begin by assessing your requirements and understanding your infrastructure demands. Our experts carefully plan and execute the installation process, ensuring proper server configuration, hardware integration, and network connectivity.

With our server installation service, you can experience enhanced performance, improved data storage, and streamlined operations. We optimize server settings to maximize efficiency and minimize downtime.

Our team ensures data security by implementing robust backup and disaster recovery solutions. We also offer ongoing server maintenance and support to keep your system running smoothly.

We provide comprehensive training to ensure that you and your team are familiar with server operations and management. Our experts are available to address any concerns or issues that may arise.

Upgrade your business capabilities with our reliable and efficient server installation service, enabling you to harness the full potential of your data and applications.

#ServerInstallation #EfficientInfrastructure #EnhancedPerformance #DataSecurity #ExpertInstallation

source

Continue Reading

Trending

Copyright © 2024 WordupNews.com