New FREE AI Video Generator & Feature Length AI Films!

The video introduces a new free AI video generator platform called Hyper, developed by Yia Meow and Zuu Wong, former Google Deep Mind alumni, who have raised over $19 million. Hyper offers text-to-video conversion with smooth animations and styles, image-to-video generation, and video repainting. The platform provides a user-friendly interface with options for HD and SD video generation, with the HD version currently limited to around 2-second clips. The video also discusses the potential for AI-generated feature-length films through Movie LLM, which uses GPT 4 and text-to-image models to create detailed scripts and visuals. Additionally, the video touches on the drama between Mid Journey and Stability AI, following a 24-hour outage at Mid Journey attributed to Stability AI employees. Mid Journey CEO David Holtz banned Stability AI employees from using their service, while Stability AI's Imad Mustafa responded on Twitter, denying the accusations.

Takeaways

• 🎬 A new free AI video generator called Hyper has been introduced, offering text to video and image to video capabilities.
• 🚀 Hyper is backed by Yia Meow and Zuu Wong, former Google Deep Mind alumni, and has raised over $19 million in funding.
• 🖼️ The platform features impressive animated styles and the ability to generate images and animate them directly on the platform.
• 🎨 Hyper also offers a unique feature called video repainting, which transforms videos into a different artistic style, like watercolor.
• 📊 The interface of Hyper is user-friendly, with options for HD video generation, image animation, and various creative styles.
• ⏰ HD video generations on Hyper tend to be around 2 seconds long, but there are workarounds to extend the duration.
• 🧩 The platform includes a community feed showcasing a wide range of creative outputs, some nearing the quality of feature-length films.
• 🔍 The research on Movie LLM suggests the potential for generating full-length AI films from a prompt, leveraging GPT 4 and text to image models.
• 🌟 The style immobilization process in Movie LLM ensures consistent visual themes and characters throughout the generated film.
• 📚 The Movie Net dataset, consisting of data from over 1,000 movies and 60,000 trailers, is used to enhance the film generation process.
• 🤝 Mid Journey and Stability are experiencing a conflict, with allegations of scraping and a subsequent ban of Stability employees from using Mid Journey's service.
• 🔧 Both Mid Journey and Stability are working on advancing their AI models, with Mid Journey training its video model and Stability preparing to release Stable Diffusion 3.

Q & A

What is the new AI video generator mentioned in the transcript?

The new AI video generator mentioned is called Hyper, a platform for creating videos from text or images.

Who are the founders of Hyper?

Hyper was founded by Yia Meow and Zuu Wong, who are both former Google Deep Mind alums.

How much funding has Hyper raised?

Hyper has raised over $19 million.

What are the features offered by Hyper for video generation?

Hyper offers text to video, image to video, and video repainting or in-painting features.

What is the limitation of the HD video generation in Hyper?

The limitation of the HD video generation in Hyper is that the generated clips tend to be around 2 seconds long.

What is the workaround suggested for extending the duration of the HD video generation?

The workaround suggested is to bring the clip into a nonlinear editor like Premiere or DaVinci Resolve and extend it using the speed/duration feature with Optical Flow turned on.

What is the latest research on generating full-length AI films?

The latest research involves Movie LLM, which enhances long video understanding with AI-generated movies, leveraging the power of GPT-4 and text to image models to generate detailed scripts and corresponding visuals.

What is the process Movie LLM uses to generate a film?

Movie LLM uses a process that includes movie plot generation, style immobilization, and video instruction by data generation, utilizing GPT-4 for initial breakdown and generating chapters, scenes, and dialogues.

What is the controversy mentioned between Mid Journey and Stability AI?

The controversy involves Mid Journey experiencing a 24-hour outage due to bot-like behavior from Paid accounts, which were traced back to Stability AI employees scraping for images and text prompts. Mid Journey's CEO banned all Stability AI employees from using their service.

What was Stability AI's response to the accusations made by Mid Journey?

Stability AI's Imad Mustafak responded on Twitter, stating that they have not been scraping and have been using synthetic and other data, and that SD3 (their model) outperforms all other models.

What is the current status of Mid Journey's video model?

Mid Journey is still in the process of training their video model, which is reported to be quite good.

When is Stability AI expected to release Stable Diffusion 3?

Stability AI is expected to release Stable Diffusion 3 into the wild at any moment.

Outlines

00:00

🎬 Introduction to Hyper: A Free AI Video Generator

The video script introduces a new, free AI video generator called Hyper, developed by Yia Meow and Zuu Wong, who are former Google Deep Mind alumni. The platform has raised over $19 million and offers text-to-video and image-to-video services with smooth animations and styles. It also features video repainting, allowing users to transform videos into different styles like watercolor. The interface is user-friendly with options for generating videos in full HD or standard definition and extending video length. The script also mentions a workaround for extending short HD video generations and compares the output quality of Hyper to that of Pika, another AI video generator.

05:00

🚗 Consistency in AI-Generated Videos and Mid Journey Drama

The paragraph discusses the consistency and quality of AI-generated videos, particularly noting that cars in the generated scenes maintain their form and do not morph erratically. It also humorously points out traffic planning flaws in a simulated city scenario. The script praises the imaginative and visually appealing outputs, such as a ghost train in the sky and a detailed interior zoom-out. However, it acknowledges the occasional oddities in AI video generation. The paragraph also touches on a conflict between Mid Journey and Stability AI, where Mid Journey's CEO accused Stability AI employees of scraping their service, leading to a temporary ban of Stability AI employees from using Mid Journey. The situation led to a public exchange on Twitter between representatives of both companies.

10:02

📚 Movie LLM Research and Upcoming Developments in AI Video

This section delves into the research on Movie LLM, which enhances long video understanding with AI-generated movies. The process involves leveraging GPT 4 for initial film breakdown, generating detailed scripts and visuals, and using style immobilization to maintain consistency throughout the film. The script discusses the use of the MovieNet dataset, which consists of data from over 1,000 movies and 60,000 trailers, to improve the synopses generated by Movie LLM. It also mentions an example using the film 'Splice' to illustrate the process. The paragraph concludes with updates on the AI video industry, noting that Mid Journey is training its video model and Stability AI is preparing to release Stable Diffusion 3.

Keywords

AI Video Generator

An AI video generator is a software tool that uses artificial intelligence to create videos from textual or image inputs. In the context of the video, it refers to a new platform called 'Hyper' that allows users to generate videos for free. It is significant because it is developed by former Google Deep Mind alumni and has the potential to revolutionize how videos are created, as it can produce content automatically based on prompts, which can be particularly useful for content creators and filmmakers.

Feature Length AI Films

Feature length AI films refer to full-length movies that are created or significantly enhanced by artificial intelligence. The video discusses the latest research in this area, which suggests that AI can be used to generate detailed scripts and corresponding visuals for movies. This technology could potentially disrupt the film industry by enabling the creation of films from a simple prompt, reducing the time and resources required to produce them.

Hyper

Hyper is a new video generation platform mentioned in the video that is currently available for free. It is developed by Yia Meow and Zuu Wong, two former Google Deep Mind employees. The platform is notable for its ability to generate videos from text and images, offering features like 'text to video', 'image to video', and 'video repainting'. It is positioned as a contender in the AI video space due to its advanced capabilities and the significant funding it has received.

Text to Video

Text to video is a feature of the AI video generator that allows users to input text prompts and have the AI create videos based on those prompts. In the video, it is shown that Hyper can generate smooth and animated styles of video from text, which is a key aspect of the platform's appeal and utility for creators who want to bring their textual ideas to life visually.

Image to Video

Image to video is another feature of the AI video generator where the AI takes a static image and animates it to create a video. The video script provides an example where an image is generated on the platform and then animated, showcasing the versatility of the AI in handling different types of visual content creation.

Video Repainting

Video repainting is a creative feature that allows users to take a video and transform its style, for instance, turning a regular video into a watercolor-like appearance. The video mentions a demo on Hyper's site where a smoothie mixture being poured into a bowl is repainted to look like a koi fish watercolor, demonstrating the platform's ability to offer artistic and innovative video transformations.

Stable Diffusion

Stable Diffusion is an AI model mentioned in the context of generating consistent styles and characters across a film. It is used in the research for feature-length AI films to ensure that the generated scenes, characters, and locations maintain a consistent look and feel throughout the movie. This is crucial for creating a cohesive and professional final product in the film industry.

Movie LLM

Movie LLM, or Movie Language Model, is a term used to describe the AI system discussed in the video that is capable of generating full-length movies from a prompt. It uses GPT-4 for initial breakdowns of the film, including themes, character descriptions, and scene outlines. Movie LLM is significant as it represents the potential future of movie production, where AI can play a central role in creating comprehensive cinematic content.

Mid Journey

Mid Journey is a company referenced in the video that is involved in the development of AI models, particularly for video generation. The video discusses a conflict between Mid Journey and Stability AI, where the former accused the latter's employees of scraping their service for images and text prompts. This led to Mid Journey banning all Stability AI employees from using their platform, highlighting the competitive and sometimes contentious nature of the AI industry.

Stability AI

Stability AI is a company mentioned in the video that is also working on AI models, including Stable Diffusion 3, which is expected to be released soon. The company was involved in a dispute with Mid Journey, as mentioned earlier. Stability AI represents another key player in the AI video generation space, and their developments are closely watched by the industry.

Optical Flow

Optical flow is a term used in video editing to describe the process of estimating the motion of objects in a video sequence. In the context of the video, it is suggested as a method to extend short AI-generated video clips by slowing them down without causing significant loss in quality. This technique is particularly useful for creators looking to overcome the current limitation of short video generation times in AI platforms.

Highlights

A new free AI video generator platform called Hyper has been launched.

Hyper is developed by Yia Meow and Zuu Wong, former Google Deep Mind alumni, and has raised over $19 million.

The platform offers text-to-video generation with smooth and clean animations.

Hyper also provides image-to-video functionality, allowing users to generate and animate images on the platform.

An innovative feature called video repainting is available, which transforms videos into artistic styles like watercolor.

The interface of Hyper is user-friendly with options for video generation, including full HD and standard definition.

Hyper's HD video generation is currently limited to around 2 seconds per clip.

A workaround for extending video clips is suggested, using nonlinear editors like Premiere or DaVinci Resolve.

The platform includes a community feed showcasing the diverse capabilities of Hyper.

Hyper's video quality is compared to Sorare, with some outputs reaching a similar level of quality.

Research on generating full-length AI films is discussed, with Movie LLM using AI to enhance long video understanding.

Movie LLM relies on GPT 4 for initial film breakdown, including theme, style, and character generation.

A style immobilization process ensures consistent visual themes throughout the generated film.

The Movie Net dataset, consisting of data from over 1,000 movies and 60,000 trailers, is used for training Movie LLM.

Mid Journey, a separate AI platform, experienced a 24-hour outage due to bot-like behavior from paid accounts.

Stability AI's employees were accused of scraping Mid Journey for images and text prompts, leading to a ban.

Imad Mustafa, CEO of Stability AI, responded on Twitter, denying the accusations and highlighting their own dataset and model performance.

Mid Journey is currently training its video model, and Stability AI is preparing to release Stable Diffusion 3.