Unverified50% confidenceSolutionExact time
ATAS transfers the retrieval-augmented generation (RAG) paradigm to the video domain: retrieval before generation is equivalent to preparing a set description and event archive for the current shot, with the difference being that the units of video retrieval are geometric structures and dynamic observations.
1
Sources
50%
Confidence
Long-term
Relevance
10/15/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe scene graph serves as a generative constraint that guides the sampling of diverse reset states, thereby covering object poses and grasp modes that do not appear in the video.66% similarUnverifiedSemantic retrieval for video requires frame extraction first, followed by generating embedding vectors for each frame and building an index.65% similarUnverifiedATAS's experimental results show that on four different video generators, ATAS improved all reported cross-shot consistency metrics and achieved the highest overall benchmark score for every backbone model.65% similarUnverifiedThe idea of mmVB can be extended to more complex scenarios such as video understanding and multimodal RAG.65% similarUnverifiedReference video templates extract motion information (such as optical flow, pose skeletons, and depth information) as conditional constraints for the generation process; common implementation methods include ControlNet, IP-Adapter, and Motion LoRA.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1037438API
curl https://kongchang.com/api/v1/knowledge/claims/1037438MCP
get_claim(id=1037438)