Runway claims its latest text-to-video model produces visuals that are even more accurate than before. In a blog post on Monday, the company announced that the Gen-4.5 model can create “cinematic and highly realistic outputs,” making it tougher to tell the difference between reality and AI-generated content. According to Runway, “Gen-4.5 achieves unprecedented physical accuracy and visual precision.” The new AI is designed to follow prompts better, which means it can generate detailed scenes without sacrificing video quality. Runway notes that AI-generated objects now “move with realistic weight, momentum, and force,” while liquids “flow with proper dynamics.”
The Gen-4.5 model will be gradually rolled out to all users and is expected to maintain the same speed and efficiency as its predecessor. However, there are still some limitations, including challenges with object permanence and causal reasoning. This means that sometimes effects may occur before their causes, like a door opening before someone uses the handle.
In addition to Runway, OpenAI is also making strides to enhance the realism of its AI-generated videos. OpenAI recently emphasized upgrades to physics with the release of its Sora 2 text-to-video model in September. Bill Peebles, head of Sora, noted that users can “accurately do backflips on top of a paddleboard on a body of water, and all of the fluid dynamics and buoyancy are accurately modeled.”
Runway asserts that its Gen-4.5 model is also better at managing various visual styles, enabling it to produce consistent photorealistic, stylized, and cinematic images. The startup claims that photorealistic visuals generated with Gen-4.5 can be “indistinguishable from real-world footage, with lifelike detail and accuracy.”
