← Capture & Create4DAnyone

4DAnyone Turns One Video Into a Moving 3D Person

The research system generates 16 synchronized views, then builds a free-viewpoint Gaussian-splat scene without a camera rig, known lens settings, or known camera positions.

ShareXFacebookLinkedIn
A guitarist surrounded by generated video views and a free-viewpoint volumetric reconstruction
4DAnyone generates views around a person, then uses them to build a moving scene that can be viewed from new angles.Image: 4DAnyone research team

A new research system called 4DAnyone can turn one casually recorded video of a person into a moving 3D scene that a viewer can circle from new angles.

The system does not need a multi-camera studio, calibrated lenses, known camera positions, or a tripod. Its nine researchers released the paper, code, and model on August 20, and the official project page lists the work for SIGGRAPH Asia 2026.

How It Works

4DAnyone first estimates a moving 3D skeleton from the source video. It renders that skeleton from target angles, giving its video model a guide for where the person’s body should appear in each new view.

The model then generates video from 16 viewpoints around the person. Two new techniques help those views stay consistent. One compresses earlier views into a fixed reference, while the other lets groups of views exchange information as the video takes shape.

Those generated videos become the input for 4D Gaussian splatting. This represents the person with many tiny colored points that move over time, so a viewer can watch the performance from a freely chosen angle.

Diagram showing a source video becoming 3D skeleton guides, generated target-view videos, and a 4D Gaussian-splat reconstruction
The system estimates a 3D skeleton, generates target-view videos, and uses those views to train a moving Gaussian-splat scene.Image: 4DAnyone research team

Results

The researchers tested 4DAnyone on 10 scenes from the DNA-Rendering benchmark and three from DyMVHumans. It led every reported image-quality and consistency measure across both datasets when compared with MV-Performer, TrajectoryCrafter, and a fine-tuned version of ReCamMaster.

The official demonstrations also include ordinary portrait videos recorded outside a capture studio. The paper presents those examples as evidence of broader use, but they do not have the ground-truth camera views needed for the same numerical comparison.

Release Details

The public code currently expects a portrait video showing one full or upper body, at least 121 frames, and only mild camera movement. The team lists support for lower-memory inference below 32 GB and a complete open-source 4D reconstruction path as future work.

The full research pipeline is still compute-heavy. The paper reports about two minutes to prepare a clip on an RTX 4090, about seven minutes to generate each four-video group on one H20 GPU, and about 30 minutes to train the final 4D scene on an RTX 4090.

Loose clothing can change across views because a body skeleton does not describe flowing fabric. An incorrect pose estimate also carries into every generated angle. The authors separately warn that realistic human synthesis can enable deepfakes and call for consent and clear disclosure.

For creators, the released model makes the first part available now. One simple video can become a ring of consistent moving views, with a more accessible end-to-end 4D workflow as the next step.

More from Capture & Create.

Browse all stories
Blackmagic’s spherical microphone in its forest launch artwork.

Spatial Audio · 2 min read

Blackmagic Unveils What It Calls the World’s First 7th-Order Ambisonic Microphone

On November 11, Blackmagic introduced a spatial audio microphone that packs 108 tiny capsules into one sphere, encodes 64 channels of sound from every direction inside the mic, and sends them over one USB-C cable. It’s due in December 2026, starting at $1,909.

Strada’s comment window over immersive whale footage.

Immersive Video · 2 min read

Strada Now Streams Apple Immersive Video From a Mac to Vision Pro Over the Internet

A collaborator can watch footage straight from your Mac and leave notes on the exact frame. Strada can also turn Blackmagic RAW into immersive dailies on that same Mac.

Bambu Lab H3C with its enclosed nine-spool filament system.

3D Printing · 4 min read

Bambu Lab’s New $2,499 Nine-Tool 3D Printer Can Switch Materials in 6 Seconds

The H3C keeps nine filaments loaded, mixes nozzle sizes in one print, and supports optional tools for cutting, drawing, and pressing patterns into a surface.

Preparing the Spatial Insider studio