Bb5a31f81ClaimVeo 3.1, Runway Gen-4.5, and Kling 3.0 generate audio synchronized with generated video footageAgreeDisagree
Bb5a31f81ClaimKling 4.0 and Gemini Omni 1.1 generate AI video clips with enhanced controllability and durationAgreeDisagree
Bb5a31f81ClaimAI video generation models produce individual clips lasting seconds rather than full minutesAgreeDisagree
Bb5a31f81ClaimRobotics labs use Runway's video models to generate synthetic data for training policiesAgreeDisagree
Bb5a31f81ClaimRunway released a 'turbo' version of Gen-3, which was a step-distilled model for faster inferenceAgreeDisagree
Bb5a31f81ClaimRunway announced its Gen-2 model two months after announcing its Gen-1 modelAgreeDisagree
Bb5a31f81ClaimRunway developed its Gen-3 model a few months after OpenAI released its Sora model in February 2024AgreeDisagree
Bb5a31f81ClaimThe development of camera controls for Gen-2 inspired Runway's research into world modelsAgreeDisagree
Bb5a31f81ClaimStep distillation allows large models to generate video in fewer steps, enabling real-time performanceAgreeDisagree
Bb5a31f81ClaimGen-1 functioned as a depth-conditioned model, converting video to a depth map before generating new pixelsAgreeDisagree
Bb5a31f81ClaimRunway added features like camera trajectory and 'Motion Brush' to give users more control over video generationAgreeDisagree
Bb5a31f81ClaimRunway developed Gen-2 after learning users wanted more control than simple text-to-video promptsAgreeDisagree
Bb5a31f81ClaimRunway's world model thesis is that scaled-up video models will predictably simulate physics and human actionsAgreeDisagree
Bb5a31f81ClaimGen-2 was created by chaining a text-to-depth-map model to the Gen-1 depth-to-video modelAgreeDisagree
Bb5a31f81ClaimA key limitation of current video models is generating realistic counterfactual scenarios, such as failuresAgreeDisagree
Bb5a31f81ClaimAnthropic's Claude Opus 5.5 generated a video of the Golden Gate Bridge entirely through code, without using an image modelAgreeDisagree
Bb5a31f81ClaimAnthropic's Claude Opus 5.5 was used to generate a 30-second marketing animation in a single attemptAgreeDisagree
Bb5a31f81ClaimAnthropic's Claude Opus 5.5 generated a historically accurate visualization of San Francisco's Market Street in 1906 using only code in BlenderAgreeDisagree
Bb5a31f81ClaimAnthropic's Claude Opus 5.5 was demonstrated creating claymations in the software Blender from a single promptAgreeDisagree