{"paper":{"title":"Pretrained Video Models as Differentiable Physics Simulators for Urban Wind Flows","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A repurposed video diffusion model acts as a fast differentiable surrogate for urban wind flow simulations and enables direct gradient-based optimization of building positions.","cross_cats":["cs.CE"],"primary_cat":"cs.LG","authors_text":"Ay\\c{c}a Duran, Bernd Bickel, Janne Perini, Michael A. Kraus, Moab Arar, Rafael Bischof, Siddhartha Mishra","submitted_at":"2026-03-22T13:08:01Z","abstract_excerpt":"Designing urban spaces that provide pedestrian wind comfort and safety requires time-resolved Computational Fluid Dynamics (CFD) simulations, but their current computational cost makes extensive design exploration impractical. We introduce WinDiNet (Wind Diffusion Network), a pretrained video diffusion model that is repurposed as a fast, differentiable surrogate for this task. Starting from LTX-Video, a 2B-parameter latent video transformer, we fine-tune on 10,000 2D incompressible CFD simulations over procedurally generated building layouts. A systematic study of training regimes, conditionin"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"The resulting model generates full 112-frame rollouts in under a second. As the surrogate is end-to-end differentiable, it doubles as a physics simulator for gradient-based inverse optimization: given an urban footprint layout, we optimize building positions directly through backpropagation to improve wind safety as well as pedestrian wind comfort.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That fine-tuning a general video diffusion model on 10,000 procedurally generated 2D incompressible CFD cases produces a surrogate whose predictions remain accurate enough for gradient-based optimization on real urban layouts without introducing systematic biases that would invalidate the discovered optima.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"WinDiNet repurposes a 2B-parameter video diffusion model as a differentiable surrogate that generates 112-frame urban wind flow rollouts in under one second and enables direct gradient optimization of building positions.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A repurposed video diffusion model acts as a fast differentiable surrogate for urban wind flow simulations and enables direct gradient-based optimization of building positions.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"2594eafaebbf4c5b8f7c538b9b33756ebb9fde42e05ab4fcc08351c1af2ead60"},"source":{"id":"2603.21210","kind":"arxiv","version":3},"verdict":{"id":"547247d2-3f1c-45c0-bf68-fa72728395c5","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T06:57:29.953804Z","strongest_claim":"The resulting model generates full 112-frame rollouts in under a second. As the surrogate is end-to-end differentiable, it doubles as a physics simulator for gradient-based inverse optimization: given an urban footprint layout, we optimize building positions directly through backpropagation to improve wind safety as well as pedestrian wind comfort.","one_line_summary":"WinDiNet repurposes a 2B-parameter video diffusion model as a differentiable surrogate that generates 112-frame urban wind flow rollouts in under one second and enables direct gradient optimization of building positions.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That fine-tuning a general video diffusion model on 10,000 procedurally generated 2D incompressible CFD cases produces a surrogate whose predictions remain accurate enough for gradient-based optimization on real urban layouts without introducing systematic biases that would invalidate the discovered optima.","pith_extraction_headline":"A repurposed video diffusion model acts as a fast differentiable surrogate for urban wind flow simulations and enables direct gradient-based optimization of building positions."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2603.21210/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"30eae4841a1f53be9c3f682b9b2d12565541711c5ee70a2b873075d134fe6443"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}