Sora 2 Adds Inpainting, Competes with Seedance
OpenAI's generative video model, Sora 2, has unveiled an inpainting feature that allows users to edit specific segments of AI-generated video. The model's primary competitor is ByteDance's Seedance 2.0, which is noted for its user interface but remains a proprietary, closed-source platform.
- Sora 2, released on September 30, 2025, goes beyond text-to-video by generating synchronized dialogue, sound effects, and ambient audio within the video itself. It also introduced a "Cameo" feature that allows users to insert a verified digital likeness of themselves into generated scenes. - Seedance 2.0's primary technical advantage is its "quad-modal" input system, which allows creators to use text, images, video clips, and audio files simultaneously as references in a single prompt. This enables more precise "director-level" control over the final output, such as replicating a specific camera motion from a reference video. - While Sora 2 is noted for its simulation of real-world physics, Seedance 2.0 offers a higher maximum resolution of 2K compared to Sora 2's 1080p. Seedance 2.0 also allows for audio reference input, enabling features like voice cloning for characters in the generated video. - OpenAI initially launched Sora 2 through an invite-only iOS app with a TikTok-style social feed for sharing creations, signaling a strategy focused on building a consumer product and community around the model. An Android version is also planned. - ByteDance is integrating Seedance 2.0 into its widely-used video editing app, CapCut, which has over 1 billion users, providing a massive existing distribution channel for the technology. - The competitive landscape for generative video is rapidly expanding beyond just OpenAI and ByteDance, with other major models like Google's Veo 3.1, Kuaishou's Kling 3.0, and Runway Gen-4.5 also pushing the boundaries of the technology. - This rapid advancement is creating new roles within engineering and product teams, with some analysts predicting that "AI Video Prompter" will become a mainstream job title in 2026, specializing in crafting detailed prompts and iterating on AI-generated footage.
