SOTAVerified

Lifelike 3D Human Generation

Generating realistic and lifelike 3D Humans from a Single RGB image input, while preserving face identity, delivering realistic texture, accurate geometry, and maintaining a valid pose of the generated human.

Papers

Showing 16 of 6 papers

TitleStatusHype
Human-VDM: Learning Single-Image 3D Human Gaussian Splatting from Video Diffusion ModelsCode2
Ultraman: Single Image 3D Human Reconstruction with Ultra Speed and DetailCode2
SIFU: Side-view Conditioned Implicit Function for Real-world Usable Clothed Human ReconstructionCode2
SiTH: Single-view Textured Human Reconstruction with Image-Conditioned DiffusionCode2
PaMIR: Parametric Model-Conditioned Implicit Representation for Image-based Human ReconstructionCode1
PIFu: Pixel-Aligned Implicit Function for High-Resolution Clothed Human DigitizationCode0
Show:102550

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1Human-VDMCLIP Similarity0.92Unverified
2UltramanCLIP Similarity0.91Unverified
3SiTHCLIP Similarity0.9Unverified
4PaMIRCLIP Similarity0.89Unverified
5SIFUCLIP Similarity0.87Unverified
6PIFuCLIP Similarity0.85Unverified