Haiper.AI: New Free AI Video Generator

In this video, Theoretically Media introduces a new free AI video generator called Haiper.AI, which offers impressive quality and results, comparable to Sora. The video highlights Haiper’s features, including text-to-video, image-to-video, and video repainting. The host demonstrates the platform’s user interface and various generation options, noting that HD video generations are currently limited to around 2 seconds. To extend video lengths, the host suggests using video editing tools like Premiere or DaVinci Resolve.

The video also explores the latest research on feature-length AI movies, focusing on a research paper about MovieLLM. MovieLLM leverages GPT-4 and text-to-image models to generate detailed scripts and corresponding visuals. The process involves generating a movie plot, creating chapters and scenes, and using stable diffusion to maintain consistent visual styles. Although the paper does not provide video examples, it showcases keyframe comparisons that highlight the model’s consistency.

Additionally, the video covers drama between MidJourney and Stability.AI, with MidJourney accusing Stability.AI employees of bot-like behavior and scraping images and text prompts. This led to MidJourney banning Stability.AI employees from using their service. Stability.AI CEO Emad responded on Twitter, denying the accusations and expressing confusion over the situation.

The video concludes with updates on MidJourney’s video model, which is reported to be quite good, and the upcoming release of Stable Diffusion 3 by Stability.AI. The host encourages viewers to check out related links and subscribe for more AI tools and tips.