This repo contains the implementation of the SIGGRAPH 2025 paper: Position-Normal Manifold for Efficient Glint Rendering on High-Resolution Normal Maps.
-
src/bsdfs/contains the Mitsuba 0.6 implementation of specular glint BRDF and diffuse BRDF. -
convert.pyconverts input normal map to the cluster and bounding box hierarchy. -
convert_G.pyconverts input normal map into GGX roughness and tangent frame for fast shadow-masking approximation.
- Mitsuba 0.6
- CUDA 11.7
- Python 3.8
- pytorch 2.0.8
- opencv-python 4.8.0.76
BSDF plugin in Mitsuba 0.6's XML format:
<bsdf type="glintconductor">
<string name="micro_normal" value=$normalmap_folder/>
<float name="sigma" value=$footprint_scale/>
<float name="thresh" value="0.0004"/> <!--cluster error threshold-->
<boolean name="fix_sigma" value="false"/> <!--use ray differential for footprint estimation-->
<float name="texture_scale" value="16.0"/> <!--scale of the texture-->
<integer name="res" value="1024"/> <!--normal map resolution-->
<string name="filter" value="gaussian"/> <!--footpint kernel type (gaussian|disk|box)-->
</bsdf>
<bsdf type="glintdiffuse">
<rgb value="0.8 0.8 0.8" name="reflectance"/>
<string name="micro_normal" value=$normalmap_folder/>
<float name="sigma" value=$footprint_scale/>
<boolean name="fix_sigma" value="false"/>
<float name="texture_scale" value="16.0"/>
<integer name="res" value="1024"/>
</bsdf>Normal map to cluster and bounding box hierarchy:
python convert.py --input <normal-or-height-map-file> --output <output-folder>An example is provided in demo/ that reproduces Fig. 11 of the paper:
- Download Mitsuba 0.6.
- Copy
src/bsdfs/to the Mitsuba 0.6 folder and add lines below tosrc/bsdfs/SConscript
plugins += env.SharedLibrary('glintdiffuse', ['glintdiffuse.cpp'])
plugins += env.SharedLibrary('glintconductor', ['glintconductor.cpp'])- Compile according to the documentation.
- In this folder,
cd demoand runsh demo.sh. - The rendering results will be saved as
scene1_isotropic.exr,scene1_brush.exr,scene1_scratched.exr.
@inproceedings{wu2025position,
author = {Liwen Wu and Fujun Luan and Miloš Hašan and Ravi Ramamoorthi},
title = {Position-Normal Manifold for Efficient Glint Rendering on High-Resolution Normal Map},
booktitle = {SIGGRAPH},
year = {2025}
}