Archive for August, 2009

Today when I was using MeshLab to process a triangle mesh, a window popped up with the following funny message:


You have successfully used MeshLab to open and process more than one hundred meshes!
We hope that this means that you have found MeshLab useful.

Please consider to send a short email to the developers of MeshLab, describing how MeshLab fitted your needs, attach to the email some screenshots of your processed meshes and tell us your impression about MeshLab.

MeshLab is developed on public funding and assessment of its impact on the whole community is necessary, so, please, spend a couple of minutes writing down a mail to us.
Thanks for using MeshLab

I installed this software more than one year ago, so I didn’t use it quite frequently. Actually for my last research project, only one simple seaweed model was used with MeshLab for some quick preview. But still thanks the developers of MeshLab!

I’m also a little curious whether it will pop up similar windows if the times I use it┬áreaches some larger milestone numbers.

Read Full Post »

This paper introduced a procedural noise that offers accurate spectral control, setup-free surface mapping, anisotropic filtering and fast evaluation at the same time.

The noise is formulated as a random pulse process and the Gabor kernel is chosen as the pulse, which has compact support in spatial domain for fast procedural evaluation, as well as compact support in frequency domain for precise spectral control. An interactive tool has been developed for noise design, which allows the user to determine noise pattern by tuning intuitive parameters in frequency domain, such as magnitude, orientation and bandwidth.

Various noise patterns have been created showing that this technique is really useful, and the accompanying video has proven that the design tool is indeed user friendly.

This paper is published on SIGGRAPH 2009 and can be found here or under the project webpage.

Read Full Post »

This paper presents a system to simulate and render photorealistic fire with high resolution, which can be used for feature film.

The algorithm pipeline goes as follows:

  1. Coarse particle/grid simulation for large-scale motion

    The artist can control the behavior of fire by directing a particle system, which is used as input to a 3D coarse grid simulator. Some standard grid-based simulation steps are performed for incompressibility and vorticity confinement. The update of velocity is then projected back to particles.

  2. View-specific refinement on GPUs

    The coarse particle attributes are projected onto evenly spaced planes parallel to the camera viewing plane. The high-resolution 2D Navier-Stokes equations are solved on each image plane independently on multiple GPUs. And the camera-oriented slices can be used for volume rendering finally.

    The refinement pass takes advantage of some important observations about visual perception of fire, which lead to the fact that camera-facing sprites work well for fire and smoke.

Although the underlying techniques are fairly complex, the system is easy to use, and indeed produces appealing results of fire efficiently.

This paper is published on SIGGRAPH 2009 and can be found here.

Read Full Post »

This paper presents a method to generate 3D solid textures from 2D images, which applies to solid textures composed of discrete particles.

The 3D particles density is estimated from 2D image via a so-called “unfolding” process, based on a fundamental relationship in stereology. The particle distribution is then synthesized from random initialization, followed by simulated annealing to resolve collisions. Although the algorithms only works for limited kinds of textures, the final results appear realistic and beautiful, especially after adding fine details, which can be obtained as noisy color variation by some other solid synthesis algorithm.

This paper is published on SIGGRAPH 2004 and can be found here.

Read Full Post »


This week I spent most of my time producing new video demo, which would be available online soon. It was the first time I tried to produce widescreen video with resolution 720 x 480, while most of my previous demos were of 640 x 480.

I do feel that the widescreen version looks better. And I think the popularity of widescreen videos and monitors is due to the following reasons:

  1. Human vision system is more adaptive to widescreen viewport. But this conclusion may have to be changed if we have a third eye in the middle and above the current two, just as shown in the Japanese comic. In that case I even believe a triangular monitor and corresponding triangular images/videos are better.
  2. The monitors are specified by diagonal length. With the same diagonal length, widescreen monitors have smaller area and thus cheaper cost.
  3.  Last but not the least, it is human nature to try novel stuff.

Ironically, while almost all the LCD TVs on sale in China have wider aspect ratio, all the official channels still only provide signals with standard ratio 4:3. So we have to choose between extended fatter appearance, incomplete display, or wasted blank area on either side.

Read Full Post »

I have kept thinking about improving my English writing for quite a long time. Following my mentor’s suggestion this morning, I’ve started my first English blog here. To practise writing, and also to express myself in a language I’m not really familiar with! Furthermore, I believe it’s better to sign up with my desired user name on this popular website as early as possible, before my name becomes famous.

Read Full Post »

Hello world!

Welcome to WordPress.com. This is your first post. Edit or delete it and start blogging!

Read Full Post »