Thanks to recent advancements in scalable deep architectures and large-scale pretraining, text-to-video generation has achieved unprecedented capabilities in producing high-fidelity, instruction-following content across a wide range of styles, enabling applications in advertising, entertainment, and education. However, these models' ability to render precise on-screen text, such as captions or mathematical formulas, remains largely untested, posing significant challenges for applications requiring exact textual accuracy. In this work, we introduce T2VTextBench, the first human-evaluation benchmark dedicated to evaluating on-screen text fidelity and temporal consistency in text-to-video models. Our suite of prompts integrates complex text strings with dynamic scene changes, testing each model's ability to maintain detailed instructions across frames. We evaluate ten state-of-the-art systems, ranging from open-source solutions to commercial offerings, and find that most struggle to generate legible, consistent text. These results highlight a critical gap in current video generators and provide a clear direction for future research aimed at enhancing textual manipulation in video synthesis.

A guy with no background in film or artificial intelligence is working on making an entire movie -- in a provocative attempt to demonstrate that generative AI art models can open ambitious levels of filmmaking to the masses. German tech entrepreneur Fabian Stelzer told PC Magazine that his 70s-style sci-fi film "Salt" will feature all artificial voices except his own, and that generative models will also create the film's footage and sound effects, too. From what Stelzer's been posting to Twitter to tease his progress, the film apears to be about space travelers who encounter a planet with an overgrowth of bizarre salt -- but beyond that basic premise, the neuroscientist is allowing his Twitter followers to determine which specific directions the film will go in a sort of "Choose Your Own Adventure" style for the AI age. "I definitely want to have a'Director's Cut' at some point, or a'Community's Cut,' but the real goal is to transcend the medium of film into something new," he said. "Like enable everyone in the community to eventually use a model that lets them write their own scenes."