SIBGRAPI 2013: Polygonal Mesh Extraction From Voxel Art.

Banshee

Introduction:

This article was sent and accepted as one of full papers from SIBGRAPI 2013.

Polygonal Mesh Extraction From Digital Voxel Art explains the 3D Modelizer -> Model Quality -> Topologically Correct Triangle Model (Low) feature that is being developped for the beta versions of Voxel Section Editor III.

The Conference on Graphics, Patterns and Images (SIBGRAPI), which used to be called Brazilian Symposium on Computer Graphics and Image Processing until 2009) is one of the most important latin american symposiums in computer graphics research, covering 3D, animation, modeling, medical, mathematical morphology, among other topics on computer graphics. This edition of SIBGRAPI was the first one that happened outside Brazil, at the city of Arequipa in Peru, being promoted by the Brazilian Computer Society (SBC), through its special interest group on Graphics and Image Processing (CEGRAPI) and organized in coordination with the Peruvian Computer Society (SPC), its special interest group on Computer Graphics and Images (CGI) and the San Pablo Catholic University (UCSP).

All articles must be written in english and presented in english.

This work continues our research on extracting polygonal meshes from voxel models posted at Project Perfect Mod's Voxels forum.

Our previous research generated a non-manifold quad or triangle mesh, smoothed it and extracted a texture atlas from it. A non-manifold triangle mesh is can be informally described as a mesh where components of its surface (vertices, edges or faces) are degenerated, therefore, you cannot say if they are in the interior of the volume of the mesh or at the exterior. Non-manifold meshes may not be processed by most mesh processing techniques known in the literature, since the results you'd obtain by using them would cause undesired results.

This work takes any voxel model as input and generates manifold triangle meshes by resampling the data that describes its volume. It generates models with dense, smooth and almost regular meshes that can be post-processed by other techniques to export them into games, movies or other media.

It's important to note that while our motivation is related to games, the technique itself works with any kind of volumetric data that is describen by a 3D grid of volumetric pixels with a binary density (or a way to detect if a voxel is in the interior or exterior of the graphical object). You can use it in seismological data, data extracted from 3D scanners, sensors, Kinect and other devices. In some cases, you may need to convert this data into a 3D grid of binary density voxels.

Abstract:

This work presents a method to extract polygonal surfaces from volumetric models created by artists, proposing a way of using voxel modeling tools to build B-Rep models.

The volumetric data created by voxel editors usually contain topological features that do not describe solid structures. Hence, the main objective of this work is to solve the problem of extracting triangle meshes from volumes that contain these topological features.

In order to extract surfaces successfully, a methodology was conceived to resample any volumetric model, in a way that it is possible to reconstruct a tridimensional manifold that can be polygonized without generating a surface with gaps or topological problems.

The meshes generated by this technique have good properties, satisfying some of the main criteria used to measure the quality of meshes, such as aspect ratio, smoothness and skewness.

Downloads:

-> Polygonal Mesh Extraction From Digital Voxel Art (Article)

Additional Credits:

- The Nod Bike is a volumetric model created by Reaperr for the modification Return of the Dawn for the game Command & Conquer: Tiberian Sun.
- Mad Dog Mech is a volumetric model created by WeeRaby2k.
- Kirov Airship Mark II is a model created by Arikado
- WarMiner is a model created by Gangster.

ViPr · Gauss Rifle Trooper **Joined**: 30 Dec 2004

What happened to all the various smoothing effects? Or are you just focusing on redoing the earlier stage before the smoothing happens?

Banshee · Posted: Mon Aug 12, 2013 10:14 pm Post subject:

I was just focused on doing the earlier stage before smoothing, because the smoothing part wasn't really a contribution from my work anymore.

ViPr · Gauss Rifle Trooper **Joined**: 30 Dec 2004

Can the same smoothing effects from before apply here also? What was being smoothed anyway? It was the colors, the normals, and also the 3D mesh, correct? Also, can it still have the surface properties be transferred to a 2D texture instead of being per triangle or per vertex? Also you realise the triangle reduction needs to be about 100 times doesn't it? Would MultiRes modifier of 3ds Max work well on these models? By the way, why does there seem to be some scatter effect going on with the colors?

Banshee · Posted: Tue Aug 13, 2013 6:42 pm Post subject:

Wow... so many questions. Hang on:

Quote:

Can the same smoothing effects from before apply here also?

The answer is yes for the transformation of face normals to vertex normals, face colours to vertex colours. The mesh smoothing applied in the previous article doesn't work with the data from this one, because the model that it generates isn't as symmetrical as the one from the other article. However, it is possible to use other mesh smoothing techniques in this data.

Quote:

What was being smoothed anyway?

- Colors, normals and the mesh.

Quote:

Also, can it still have the surface properties be transferred to a 2D texture instead of being per triangle or per vertex?

Yea, and the method from the previous paper works on the data this one generates. But there might be better techniques to do it.

Quote:

Also you realise the triangle reduction needs to be about 100 times doesn't it?

I don't have numbers to measure it, since it changes for each model. But it certainly needs to be heavly optimized.

Quote:

Would MultiRes modifier of 3ds Max work well on these models?

I think it could work, since the model is a manifold. But I've never tested it.

Quote:

By the way, why does there seem to be some scatter effect going on with the colors?

I don't know exactly. Anyway, the calculation of colors is something that will be investigated and improved in the next months.