Runway Act-Two dramatically upgrades motion capture capabilities

WeTransfer reverses controversial AI training terms

Absen builds world's largest LED virtual production volume

Lightricks LTXV generates long-form AI video

Act-Two requires only a driving performance video and reference character image to animate any digital character, eliminating the need for expensive motion capture hardware or specialized studios.

The model supports diverse character types and art styles, from photorealistic humans to stylized cartoon characters, while maintaining performance fidelity across different environments and creative directions.

Enhanced tracking capabilities now include fine facial expressions and finger gestures, plus automatic environmental motion that makes characters appear more naturalistic without additional editing.

Pricing is set at 5 credits per second of generated animation with clips running up to 30 seconds at 24 fps, making it accessible for rapid prototyping and iterative workflows.

The upgrade addresses key limitations from Act-One by improving motion consistency and expanding the range of trackable movements, positioning it as a more comprehensive solution for animation studios.

Premium Display Quality: Features a daylight-viewable 19-inch 1920x1080 touchscreen with stunning 1200-nit brightness and 10-bit color depth, covering the DCI-P3 color gamut for perfect visibility even in direct sunlight.

Professional Recording Capabilities: Capture content in multiple high-end formats, including ProRes RAW (up to 6Kp30), ProRes 422, and DNxHR up to 4K60p, with multi-channel recording for complete production flexibility.

Complete Connectivity: Equipped with four 12G-SDI inputs, HDMI 2.0 connectivity, signal cross-conversion, and professional audio support including dual XLR inputs with +48V phantom power.

Advanced Production Tools: Access essential monitoring features including waveform, vectorscope, false color, focus peaking, anamorphic de-squeeze, and comprehensive LUT support for both SDR and HDR workflows – tools typically found only in much more expensive systems.

Trusted Across The Industry: Used on hundreds of productions, from the latest season of The Last of Us to capturing hundreds of camera feeds for MrBeast videos.

The original clause granted WeTransfer rights to use content for "improving performance of machine learning models" as part of content moderation, sparking fears among users that their intellectual property could feed AI training datasets.

Creative professionals including filmmakers, artists, and photographers led the outcry on social media, with many threatening to cancel subscriptions over concerns their work would be used without permission to develop competing AI tools.

WeTransfer's swift response included revising the terms to state content is used solely "for the purposes of operating, developing, and improving the Service" with no AI training whatsoever, addressing user privacy and intellectual property concerns.

The controversy reflects broader industry tensions as creative professionals increasingly worry about how their work feeds into AI development, following similar backlash at companies like Adobe and Meta over data usage policies.

The Writers' Guild of Great Britain welcomed the clarification, emphasizing that "members' work should never be used to train AI systems without their permission" and highlighting the importance of explicit consent in creative industries.

The volume has a diameter of 50 meters and a height of 12 meters. The total floor area of the stage reaches 5,000 square meters, with the LED display covering approximately 1,700 square meters—equivalent to four basketball courts.

Productions can achieve up to 160° viewing angles without color shifts, allowing cameras and actors to move freely around the set without technical limitations.

The modular design allows studios to reconfigure or expand the volume for different scenes and projects, maximizing flexibility for various production needs.

Energy efficiency improvements deliver up to 40% less power consumption compared to comparable LED systems, reducing operational costs for extended shooting schedules.

The model runs 30 times faster than comparable solutions while maintaining quality, thanks to its 13-billion parameter architecture that works on standard consumer GPUs rather than requiring expensive hardware.

LTXV introduces Continuous Control LoRAs that allow creators to input signals for pose, depth, and visual style throughout the entire video generation process, not just at the beginning like previous models.

The real-time streaming capability means creators can watch their video generate frame by frame while making live adjustments and refinements, reducing the typical trial-and-error workflow of AI video creation.

Unlike proprietary competitors like OpenAI's Sora or Runway's Gen-4, LTXV offers open-source weights and codebase on platforms like Hugging Face, enabling developers to experiment and customize the model for specific use cases.

The technology supports both text-to-video and image-to-video workflows, giving creators flexible entry points for different types of projects from advertising to game development.