3D Object Generation

MeshAnything can convert 3D assets in any 3D representation into meshes. This can be used to enhance various 3D asset production methods and significantly improve storage, rendering, and simulation efficiencies.

MagicPose4D can generate 3D objects from text or images and transfer precise motions and trajectories from objects and characters in a video or mesh sequence.

RemoCap can reconstruct 3D human bodies from motion sequences. It’s able to capture occluded body parts with greater fidelity, resulting in less model penetration and distorted motion.

NOVA-3D can generate 3D anime characters from non-overlapped front and back views.

DreamScene4D can generate dynamic 4D scenes from single videos. It tracks object motion and handles complex movements, allowing for accurate 2D point tracking by converting 3D paths to 2D.

X-Oscar can generate high-quality 3D avatars from text prompts. It uses a step-by-step process for geometry, texture, and animation, while addressing issues like low quality and oversaturation through advanced techniques.

Invisible Stitch can inpaint missing depth information in a 3D scene, resulting in improved geometric coherence and smoother transitions between frames.

And on the pose reconstruction front we have had TokenHMR, which can extract human poses and shapes from a single image.

PhysDreamer is a physics-based approach that enables you to poke, push, pull and throw objects in a virtual 3D environment and they will react in a physically plausible manner.

InFusion can inpaint 3D Gaussian point clouds to restore missing 3D points for better visuals. It lets users change textures and add new objects, achieving high quality and efficiency.

in2IN is a motion generation model that factors in both the overall interaction’s textual description and individual action descriptions of each person involved. This enhances motion diversity and enables better control over each person’s actions while preserving interaction coherence.

Video2Game can turn real-world videos into interactive game environments. It uses a neural radiance fields (NeRF) module for capturing scenes, a mesh module for faster rendering, and a physics module for realistic object interactions.

LoopGaussian can convert multi-view images of a stationary scene into authentic 3D cinemagraphs. The 3D cinemagraphs can be rendered from a novel viewpoint to obtain a natural seamless loopable video.

[MCC-Hand-Object (MCC-HO)] can reconstruct 3D shapes of hand-held objects from a single RGB image and a 3D hand model. It uses Retrieval-Augmented Reconstruction (RAR) with GPT-4(V) to match 3D models to the object’s shape, achieving top performance on various datasets.

ProbTalk is a method for generating lifelike holistic co-speech motions for 3D avatars. The method is able to generate a wide range of motions and ensures a harmonious alignment among facial expressions, hand gestures, and body poses.

GaussianCube is a image-to-3D model that is able to generate high-quality 3D objects from multi-view images. This one also uses 3D Gaussian Splatting, converts the unstructured representation into a structured voxel grid, and then trains a 3D diffusion model to generate new objects.

Garment3DGen can stylize the geometry and textures from 2D image and 3D mesh garments! These can be fitted on top of parametric bodies and simulated. Could be used for hand-garment interaction in VR or to turn sketches into 3D garments.

ThemeStation can generate a variety of 3D assets that match a specific theme from just a few examples. It uses a two-stage process to improve the quality and diversity of the models, allowing users to create 3D assets based on their own text prompts.

TexDreamer can generate high-quality 3D human textures from text and images. It uses a smart fine-tuning method and a unique translator module to create realistic textures quickly while keeping important details intact.

Controllable Text-to-3D Generation via Surface-Aligned Gaussian Splatting can create high-quality 3D content from text prompts. It uses edge, depth, normal, and scribble maps in a multi-view diffusion model, enhancing 3D shapes with a unique hybrid guidance method.