Unverified60% confidenceFactExact time
OpenAI's Sora and Google's VideoPoet both adopt the technical approach of compressing video frames into discrete tokens followed by autoregressive prediction
2
Sources
60%
Confidence
Long-term
Relevance
7/11/2026
First Seen
Sources
Lingbot World: A 14B Open-Weight World Model with Causal Architecture and Inference Acceleration
redditr/StableDiffusion7/9/2026
Related Claims
UnverifiedOpenAI Sora 引入了视频压缩网络与 Transformer 结合的时空补丁机制，将视频拆解为跨越空间与时间的三维补丁79% similarUnverifiedSora introduces a video compression network that compresses video into low-dimensional spatiotemporal patches, then performs diffusion generation in that space.75% similarUnverified平台对外分发视频通常经历转码压缩降带宽和叠加可见及隐式水印两个处理步骤75% similarUnverifiedAlgorithms on vertical short-video platforms are extremely sensitive to completion rate; therefore, each short video should be edited around a single strong point of view rather than simply clipped from a longer video.75% similarUnverifiedWhen releasing the Sora video generation model, OpenAI emphasized that Sora implicitly learned some physical laws (such as fluid motion and object collisions) in the process of generating coherent videos.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/477996API
curl https://kongchang.com/api/v1/knowledge/claims/477996MCP
get_claim(id=477996)