Skip to content

ctk2156/DeepAo

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

38 Commits
 
 
 
 
 
 
 
 

Repository files navigation

DeepAo

DeepAO: Efficient Screen Space Ambient Occlusion Generation via Deep Network

This repo contain generating Ambient Occlusion via a CNN and disploy it in Unity. If any question, please put on issues pannel directly or contact 15692002156@163.com.

Dataset Sample Visualization

Image text

Download

Dataset Generation Pipline

We use 3dsMax with Vray engine to generate our dataset. Our AO dataset contain Ao map and corresponding Depth map and Normal map. You can make your own dataset by following these steps:
1.Make your own 3d scene, you can get 3D material from Free3d.
2.Configure the Vray engine.

  • AO map: Configure the Vray engine according to the instructions here for bake Ambient Occlusion Map(Method 4 or 5, It depends on your needs).
  • Position map: VRaySamplerInfo is recommended instead of VrayZDepth, because the depth value obtained by VrayZDepth is nonlinear. You can get the xyz coordinates of each point from VRaySamplerInfo. Our experiment only use Z coordinates(Z depth).
  • Normal map: You can use VRaySamplerInfo or VRayNormals, VRaySamplerInfo can generate normal in camera space and world space. VRayNormals can only generate normal in camera space. Our experiment use normal in camera space.

3.Planned shooting path and bind to camera.
4.Use Batch Render to render the scene.

About

DeepAO: Efficient Screen Space Ambient Occlusion Generation via Deep Network

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors