pith:6MYRC43N
3DPhysVideo: Consistency-Guided Flow SDE for Video Generation via 3D Scene Reconstruction and Physical Simulation
A training-free pipeline turns a single image into a physically realistic video by reconstructing 3D scenes and guiding generation with physics simulations.
arxiv:2605.16795 v1 · 2026-05-16 · cs.CV · cs.AI · cs.GR
Add to your LaTeX paper
\usepackage{pith}
\pithnumber{6MYRC43NGU6TBFE4K46YBFTVCQ}
Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge
Record completeness
Claims
This work introduces 3DPhysVideo, a novel training-free pipeline that generates physically realistic videos from a single image... our method successfully bridges the gap from single-images to physically plausible videos, while remaining efficient to run on a single consumer GPU. It outperforms state-of-the-art baselines on GPT-based scores, VideoPhy benchmark and human evaluation.
The off-the-shelf image-to-video flow model can be effectively repurposed for both 3D scene reconstruction via point cloud guidance and final video synthesis via physically simulated point cloud guidance without any fine-tuning or additional training.
A training-free pipeline reconstructs 3D scenes from single images, applies physics simulation, and uses Consistency-Guided Flow SDE to steer an off-the-shelf image-to-video model for physically plausible video output.
References
Formal links
Receipt and verification
| First computed | 2026-05-20T00:03:22.490163Z |
|---|---|
| Builder | pith-number-builder-2026-05-17-v1 |
| Signature | Pith Ed25519
(pith-v1-2026-05) · public key |
| Schema | pith-number/v1.0 |
Canonical hash
f33111736d353d30949c573d8096751418204b488244cbc6f6ba9b975c1b8587
Aliases
· · · · ·Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/6MYRC43NGU6TBFE4K46YBFTVCQ \
| jq -c '.canonical_record' \
| python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: f33111736d353d30949c573d8096751418204b488244cbc6f6ba9b975c1b8587
Canonical record JSON
{
"metadata": {
"abstract_canon_sha256": "e55c3db1dfff8eb1782de943e2b093ff0716c9cfe1e7c3b26880ae828d15b085",
"cross_cats_sorted": [
"cs.AI",
"cs.GR"
],
"license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
"primary_cat": "cs.CV",
"submitted_at": "2026-05-16T03:56:52Z",
"title_canon_sha256": "4bc188e5d8a37f64ae9f63551b4a774a5cef033cff90418251ef6fdbdb149be6"
},
"schema_version": "1.0",
"source": {
"id": "2605.16795",
"kind": "arxiv",
"version": 1
}
}