SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2407.12777 · 2024

Generalizable Human Gaussians for Sparse View Synthesis

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 9 functions out of this paper's own repositories and ran 7 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
humansensinglab/Generalizable-Human-Gaussians canonical 7 of 9
FunctionStatusWhere it lives
dilate Ran humansensinglab/Generalizable-Human-Gaussians/lib/ghg/network_eval.py
pointer only (licence: NONE) · get_code("0b6a67d62274e07e")
focal2fov Ran humansensinglab/Generalizable-Human-Gaussians/lib/graphics_utils.py
pointer only (licence: NONE) · get_code("b0c6d75bb1487c71")
gaussian Ran humansensinglab/Generalizable-Human-Gaussians/lib/loss.py
pointer only (licence: NONE) · get_code("c56b7ef16f309a45")
getProjectionMatrix Ran humansensinglab/Generalizable-Human-Gaussians/lib/graphics_utils.py
pointer only (licence: NONE) · get_code("616ba1e1fd6951b2")
l1_loss Ran humansensinglab/Generalizable-Human-Gaussians/lib/loss.py
pointer only (licence: NONE) · get_code("ac0e42d6fbcfbbe6")
read_img Ran humansensinglab/Generalizable-Human-Gaussians/lib/ghg/human_loader.py
pointer only (licence: NONE) · get_code("d73cfad5ab886ac4")
repeat_interleave Ran humansensinglab/Generalizable-Human-Gaussians/lib/utils.py
pointer only (licence: NONE) · get_code("f424875683006127")
getWorld2View2 Not yet run humansensinglab/Generalizable-Human-Gaussians/lib/graphics_utils.py
pointer only (licence: NONE) · get_code("6c394c1d0cb69da4")
sequence_loss Not yet run humansensinglab/Generalizable-Human-Gaussians/lib/loss.py
pointer only (licence: NONE) · get_code("520a398e387edf23")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Recent progress in neural rendering has brought forth pioneering methods, such as NeRF and Gaussian Splatting, which revolutionize view rendering across various domains like AR/VR, gaming, and content creation. While these methods excel at interpolating {\em within the training data}, the challenge of generalizing to new scenes and objects from very sparse views persists. Specifically, modeling 3D humans from sparse views presents formidable hurdles due to the inherent complexity of human geometry, resulting in inaccurate reconstructions of geometry and textures. To tackle this challenge, this paper leverages recent advancements in Gaussian Splatting and introduces a new method to learn generalizable human Gaussians that allows photorealistic and accurate view-rendering of a new human subject from a limited set of sparse views in a feed-forward manner. A pivotal innovation of our approach involves reformulating the learning of 3D Gaussian parameters into a regression process defined on the 2D UV space of a human template, which allows leveraging the strong geometry prior and the advantages of 2D convolutions. In addition, a multi-scaffold is proposed to effectively represent the offset details. Our method outperforms recent methods on both within-dataset generalization as well as cross-dataset generalization settings.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2407.12777")
get_code_for_paper("2407.12777")
have("2407.12777")

Connect an agent — have() is free.