RIP Claude Fable, open-source AI unleashed, full body avatars, new Google models, new TTS: AI NEWS
Reconstructs full-body 4D human models from one or multiple videos without skeletons or depth maps. Code release in preparation.
FL
Flex4DHuman
Multi-view video diffusion model that turns monocular or sparse videos of dynamic humans into dense multi-view videos for 4D Gaussian splats.
Updated EU status checked Last shown
Flex4DHuman at a glance
- What is Flex4DHuman?
- Flex4DHuman is a research. Multi-view video diffusion model that turns monocular or sparse videos of dynamic humans into dense multi-view videos for 4D Gaussian splats.
- Does Flex4DHuman work in the EU?
- Unverified. No explicit evidence found; only a research project page exists and no code or product is publicly available yet. Last verified 14 Nov 2026.
- How much does Flex4DHuman cost?
- Pricing unknown. Research project; no public release or pricing found. Code release reportedly in preparation.
- Where was Flex4DHuman shown?
- In 1 creator video by AI Search. Most recently by AI Search on 14 Jun 2026, at 32:51 in “RIP Claude Fable, open-source AI unleashed, full body avatars, new Google models, new TTS: AI NEWS”.
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
About
Flex4DHuman is a research project presenting a multi-view video diffusion model. It takes monocular or sparse multi-view videos of a dynamic subject and generates synchronized dense multi-view videos, conditioned only on relative camera poses.
The generated dense views can be used to directly reconstruct dynamic 4D Gaussian splats of the subject. According to a YouTube creator summary, it reconstructs full-body 4D human models from one or several videos without needing skeletons or depth maps.
As of the available material, only the project page is public. Code release is described as being in preparation, and no weights, paper link or license were found in the provided pages.
Key features
- Video diffusion from monocular or sparse multi-view input
- Outputs synchronized dense multi-view videos
- Conditioned only on relative camera poses
- Direct reconstruction of dynamic 4D Gaussian splats
- No skeleton or depth map required (per creator summary)
1 moment
As seen on YouTube
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Keep exploring
Similar AI tools
Closest neighbours by meaning, not just by category.
4D
4DAnyone
Model
Research method that reconstructs a 4D Gaussian Splatting human from a single casual monocular video.
Unverified3D & Worlds
Aug 2026
LI
Lift4D
Research
Research method that reconstructs complete 4D (3D plus time) scenes from a single monocular video using Gaussian Splatting.
Unverified3D & Worlds
Jun 2026
P3
Pantheon 360
Model
3D-aware 360° video diffusion model that generates panoramic videos along user-defined camera paths from sparse panoramas.
UnverifiedVideo Generation
2026
OV
One Video One World
Research
Training-free research system that reconstructs simulation-ready 4D mesh scenes from a single monocular video.
Unverified3D & Worlds
Release date n/aOpen source
TR
TriSplat
Model
Feed-forward 3D scene reconstruction model that outputs triangle primitives directly usable as meshes for simulation.
Unverified3D & Worlds
May 2026Open source
PU
PURE
Research
Post-training method that refines feed-forward 3D reconstruction models (Pi3, DA3, VGGT) on unlabeled video.
Unverified3D & Worlds
Release date n/aOpen source