零儲存危機隱私保護：您上傳的證據影像僅在記憶體中進行瞬時神經分析，完成後立即徹底銷毀。無日誌、無雲端備份。
Alibaba Wan 2.1 Video Detector: Spotting WanX Flow-Matching Diffusion
Forensic inspection of Alibaba Wan 2.1 14B flow-matching video foundation model and 3D causal VAE architectures.
使用 Sealed Rose 驗證媒體真實性
即時神經網路檢測 · 免費 · 無需註冊 · 零儲存 · 記憶體瞬時處理
Alibaba Wan 2.1 is an open-source 14B parameter foundation video model widely adopted by independent creators, open-source communities, and bad actors. Because the model weights are public, it can be run locally with zero content filters or watermarks. Sealed Rose exposes the unique 3D causal VAE latent artifacts and flow-matching trajectory anomalies inherent to Wan 2.1.
關鍵警訊與破綻特徵
13D Causal VAE Latent Smearing
Moving high-contrast edges leave subtle horizontal latent ghosting trails across successive frames.
2Non-Newtonian Fluid & Particulate Physics
Raindrops, dust, and water surfaces occasionally fuse into solid background geometry rather than following Navier-Stokes fluid mechanics.
3Specular Light Detachment on Moving Vehicles
Reflections on metallic car bodies and window glass slide across geometry during camera rotation rather than tracing true 3D Fresnel ray angles.
Wan 2.1 Flow-Matching Architecture Forensics
- Flow-Matching Vector Fields: Straight-path ODE solvers leave distinguishable trajectory velocity markers in pixel space.
- 3D Causal VAE Latent Smear: Time-causal VAE compression creates periodic temporal smoothing artifacts every 4th frame.
- Unconstrained Open Weight Synthesis: Lack of safety gating allows bad actors to synthesize realistic fake news and romance scam media.
防範措施與應對步驟
Upload to Sealed Rose AI Video Scanner
Run spatial-temporal forensic tests to check for Wan 2.1 3D VAE decoders and flow-matching mathematical signatures.
常見問題
Why is Wan 2.1 considered high-risk for disinformation?▼
Because Wan 2.1 open weights can be downloaded and run locally on high-end consumer hardware without safety filters, scammers use it to generate unrestricted deepfakes.
支援 23 種語言：