SenseNova 6.8 Code2Video: Open-Sourced Model That Writes and Revises Code for Videos

SenseNova has released SenseNova 6.8 Code2Video, an open-source model designed to generate, render, and revise code for programmatic motion graphics and videos. The release includes full model weights and a specialized creation Skill to guide development, previews, and revisions. This open-source release automates the video creation workflow using code generation and iterative natural-language refinement, representing a notable development in multimodal AI and programmatic animation. It shifts the paradigm from traditional black-box video generation to controllable, code-driven visual assets. The model supports reconstructing a reference video's visual treatment as editable code and allows natural-language requests to modify styling, timing, and scene elements. While current releases feature full model weights, INT4 and INT8 quantized weights are planned for future updates.