search

XPandora / PhysGaussian

[CVPR 2024 Highlight] PhysGaussian: Physics-Integrated 3D Gaussians for Generative Dynamics
1.07k stars 44 forks source link

Need some guidances #3

Closed Kedaya66 closed 11 months ago

Kedaya66 commented 11 months ago

Hello！ I am interested in nerf researches, but I am totally new to this area. So, I wanna ask about the hardware requirements of training nerf models. And could you provide some advice for a beginner to this research area. Thanks!

Kedaya66 commented 11 months ago

And how long will it cost to train nerf models.

XPandora commented 11 months ago

Hi,

You may start by reading the original NeRF paper and some of its popular follow-up work, like Mip-NeRF 360, Instant-NGP, etc. In this work, we use Gaussian Splatting as our 3D representation and typically it takes less than 1 hour to train a scene on a RTX 3090 GPU.

Kedaya66 commented 11 months ago

Ok！Sincerely thanks for your advice！

发自我的iPhone

------------------ Original ------------------ From: Tianyi Xie @.> Date: Thu,Jan 4,2024 5:58 AM To: XPandora/PhysGaussian @.> Cc: Kedaya66 @.>, Author @.> Subject: Re: [XPandora/PhysGaussian] Need some guidances (Issue #3)

Hi,

You may start by reading the original NeRF paper and some of its popular follow-up work, like Mip-NeRF 360, Instant-NGP, etc. In this work, we use Gaussian Splatting as our 3D representation and typically it takes less than 1 hour to train a scene on a RTX 3090 GPU.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>