Seedance 2.0 rolls out
Seedance 2.0, a new image‑to‑video AI model that supports multi‑character scenes and dynamic cameras, went globally live on HeyGen and is now integrated into Runway’s paid plans. Runway announced an introductory promotion (50% off the first three months with code SEEDANCE) as it adds the model to its creative toolkit. The update is being framed as a step toward more cinematic, multi‑character AI video workflows. (x.com) (x.com)
A lot of artificial intelligence video tools can make one pretty shot. The hard part is making two people stay consistent across cuts while the camera moves like a real operator instead of a security camera. (heygen.com) That is the pitch behind Seedance 2.0, which HeyGen says went fully live across its platform this week and which Runway says is now available inside its paid plans. Runway’s launch page says the offer is 50% off the first three months with the code SEEDANCE, ending April 13 at 9 a.m. Pacific Time. (heygen.com) (runwayml.com) Image-to-video means you hand the system a still frame, like a product photo or character image, and ask it to turn that frozen moment into moving footage. The usual failure is that faces drift, hands mutate, and the camera motion feels pasted on after the fact. (runwayml.com) (heygen.com) Seedance 2.0 is being sold as a fix for that specific problem. HeyGen says it can handle multi-character scenes, dynamic camera moves, and physics-accurate motion from one prompt bar, while Runway says it can generate multi-shot video with sound, dialogue, and consistent characters from text, images, video, or audio inputs. (heygen.com) (runwayml.com) HeyGen is pushing a different angle than Runway. Its claim is that Seedance inside HeyGen can be used with verified human faces and the company’s “digital twin” system, because HeyGen built its own consent and identity checks around those avatars. (heygen.com 1) (heygen.com 2) Runway is positioning the same model more like a general creative engine. Its help center says Seedance 2.0 is a third-party model available on the Standard plan or higher, and its product page frames it as one option inside a larger menu of image, video, and audio models. (runwayml.com 1) (runwayml.com 2) (runwayml.com 3) The technical shift here is not just prettier clips. Runway says users can feed the model text, reference images, audio, and video in one workflow, and HeyGen says users can choose presets like Multi-Scene Cut, Dynamic Camera Move, and Multi-Character Scene instead of stitching separate tools together by hand. (runwayml.com) (heygen.com) That matters for the kind of video people actually need to make. A product ad, a talking avatar, and a cinematic background shot used to live in different software bins; HeyGen is bundling those into Avatar Shots, its artificial intelligence video generator, and Video Agent, while Runway is bundling Seedance into the same workspace as its other generation tools. (heygen.com 1) (heygen.com 2) (runwayml.com) The timing also says something about where the market is going in April 2026. Runway is still promoting its own Gen-4.5 model on its homepage, but it is also giving paid users access to Seedance as a third-party option, which means the battle is shifting from “whose model wins” to “which platform lets creators mix the best models fastest.” (runwayml.com) (runwayml.com) So this rollout is less about one model landing on two websites and more about a new workflow becoming normal. The new baseline is one prompt, multiple characters, moving cameras, synced audio, and a finished clip that comes out of the generator closer to an edited scene than a raw experiment. (runwayml.com) (heygen.com)
