VasTexture: Vast repository of textures and PBR Materials extracted from images using unsupervised approach

VasTexture: Vast repository of textures and SVBRDF/PBR Materials extracted from images using an unsupervised approach.

This dataset contains hundreds of thousands of textures and PBR/SV-BRDF materials extracted from real-world natural images.

The repository is composed of RGB images of textures given as RGB images (each image is one uniform texture) and folders of PBR/SVBRDF materials given as a set of property maps (base color, roughness, metallic, etc).

Note that this contain subset of repository more could be found in the main project page.

Visualisation of sampled PBRs and Textures can be seen in: PBR_examples.jpg and Textures_Examples.jpg

Link to the main project page

Link to paper

File structure

Texture images are given in the Extracted_textures_*.zip files.

Each image in this zip file is a single texture, the textures were extracted and cropped from the open images dataset. 

PBR Materials are available in PBR_*.zip files these PBRs were generated from the texture images in an unsupervised way (with no human intervention). Each subfolder in this file contains the properties map of the PBRs (roughness, metallic, etc, suitable for blender/unreal engine). Visualization of the rendered material appears in the file Material_View.jpg in each PBR folder.

PBR materials and textures which are larger then 512x512 pixels contain 'large' or 'larger' in their file  name there about 40k in this repository but 100k more can be found in the main project page.

PBR materials and textures who are seamless marked as contain 'seamless' in their file name. 

PBR materials that were generated by mixing other PBR materials are available in files  with the names PBR_mix*.zip 

Samples for each case can be found in files named:  Sample_*.zip

File with the word Seamless contain tileable seamless textures this files also contain textures and PBR>512 size that were extracted from the Segment Anything Dataset.

Since the textures were extracted fom natural images they were not natively seamless but were turned to seamless using the code at this url

Documented code used to extract the textures and generate the PBRs is available at:

Texture_And_Material_ExtractionCode_And_Documentation.zip

Details:

The materials and textures were extracted from real-world images using an unsupervised extraction method (code supplied). As such they are far more diverse and wide in scope compared to existing repositories, at the same time they are much more noisy and contain more outliers compared to existing repositories.  This repository is probably more useful for things that demand large-scale and very diverse data, yet can use noisy and lower quality compared to professional repositories with manually made assets like ambientCG.  It can be very useful for creating machine learning datasets, or large-scale procedural generation. It is less suitable for areas that demand precise clean and categorized PBR like CGI art and graphic design.  For preview It is recommended to look at PBR_examples.jpg and Textures_Examples.jpg or download the Sample files and look at the Material_View.jpg  files to visualize the quality of the materials.

Scale:

Currently, there are a few hundred of thousands PBR materials and textures but the goal is to make this into over a million in the near future.

Data generation code:

The Python scripts used to extract these assets are supplied at: 

Texture_And_Material_ExtractionCode_And_Documentation.zip

The code could be run in any folder of random images extract regions with uniform textures and turn these into PBR materials. 

Alternative download sources:

Alternative download sources:

https://sites.google.com/view/infinitexture/home

https://e.pcloud.link/publink/show?code=kZON5TZtxLfdvKrVCzn12NADBFRNuCKHm70

https://icedrive.net/s/jfY1xSDNkVwtYDYD4FN5wha2A8Pz

Paper

This work was done as part of the paper "Learning Zero-Shot Material States Segmentation,

by Implanting Natural Image Patterns in Synthetic Data".

@article{eppel2024learning,

  title={Learning Zero-Shot Material States Segmentation, by Implanting Natural Image Patterns in Synthetic Data},

  author={Eppel, Sagi and Li, Jolina and Drehwald, Manuel and Aspuru-Guzik, Alan},

  journal={arXiv preprint arXiv:2403.03309},

  year={2024}

}

License:

All the code and repositories are available on CC0 (free to use) licenses.

Textures were extracted from the open images dataset which is an Apache license.