Differentiable Block Compression for Neural Texture

EGSR 2025

36th Eurographics Symposium on Rendering

PDF: Eurographics Digital Library

Authors

• Tao Zhuang: Tencent Technology (Shenzhen) Co., LTD
• Wentao Liu: University of Science and Technology
• Ligang Liu: University of Science and Technology

Citation

@inproceedings{10.2312:sr.20251199,
booktitle = {Eurographics Symposium on Rendering},
editor = {Wang, Beibei and Wilkie, Alexander},
title = {{Differentiable Block Compression for Neural Texture}},
author = {Zhuang, Tao and Liu, Wentao and Liu, Ligang},
year = {2025},
publisher = {The Eurographics Association},
ISSN = {1727-3463},
ISBN = {978-3-03868-292-9},
DOI = {10.2312/sr.20251199}
}

Environment

• CUDA 13.0

• LibTorch 2.9.1

• VS 2022 latest

• CMake latest

Compilation

Use cmake to generate the sulotion, build the project DBC_examples_NeuralMterial

Run

run the built exe (bin/DBC_examples_NeuralMaterial.exe) with command arguments

<object_name> <codec> <optimize_mode> <MoP> <Ns> <Nr> <feature_size>

for example

lubricant_spray BC7 1 1 2 2 512

It will compress bin/image/lubricant_spray texture set on the Neural Material model with DBC framework.

For more detailed information, please refer to the source code.

Result

(decompressed) result images is located in bin/image/pred