Latest rendered image:

    Monday March 14, 2005

      The tentative 1.0 release date has been set for Sunday, March 20. I hope all the remaining issues can be cleared up by then. The site will also be updated during the course of the week.

    Monday March 7, 2005

      I am excited to announce the imminent release of version 1.0, and the advancement of the project status to beta from alpha. This new release will contain numerous bug fixes and several new features such as: physically-based subsurface scattering, quasi monte carlo sampling and a completely flexible and customizable shader system. This new release should be available within the next two weeks.

    Tuesday July 13, 2004

      A new release is available, 0.9.6 build 72. Major enhancements include the addition of the new BRDFs mentioned below and also the DirectVolumeRenderingShader which enables R.I.S.E. to visualize CT or MRI volume data. This will probably be the last release for a couple of months since I have little time for R.I.S.E. due to the conferences I must prepare for.

    Wednesday June 16, 2004

      There is a new version of MFCGonio (v. available. MFCGonio is a virtual goniophotometer application used to help visualize scattering probability functions and BRDFs. This latest version contains the newly implemented R.I.S.E. materials to be available in a future release.

    Friday June 11, 2004

      The photon mapping code has been changed so that BRDF of the surface is evaluated for every photon during a radiance estimate. This provides for caustics on non-lambertian surfaces. There is a patch available from build 70 along with a new source package.

    Thursday June 3, 2004

      Another new release is available, 0.9.5 build 70. The binaries package has been removed. Included is an implementation of the keyframe animation system. Several new features and old bugs have been fixed. Look at the Change list in the Downloads section for a full list. Also included are several new test scenes.

    Sunday April 25, 2004

      A new release is available, 0.9.4 build 69. All packages except for the extlib package has been updated. Included are more sample scenes. There are also numerous bug fixes and several new features added. See the Downloads section for a list of changes.

    Friday April 16, 2004

      A new release, 0.9.3 build 68 is available for download. All packages have been updated except for the extlib package, since nothing in the external libraries have changed. The installer now contains the source code as well. A quick summary of changes is in the Downloads section.

    Thursday February 19, 2004

      A new release, 0.9.3 build 67 was made available yesterday. This release is broken up into several download types. In addition to downloading the source, compiled binaries for Windows and Linux can be downloaded along with sample scenes, meshes and textures. An installer for Windows is also available which installs Windows binaries and sample scenes.

    Tuesday February 3, 2004

      CVS access has been disabled because the CVS repository has been moved to an internal server. R.I.S.E. source will only be avaiable as releases, if for some reason you want the latest source, contact me directly.

    Friday December 19, 2003

      There is a new source release available for download. This is the final release for 2003. There are numerous changes including better photon mapping, irradiance caching, an iridescent painter and more.

    Friday November 21, 2003

      Source release 65 is now available for download. It includes refactored color code along with a new painter the Voronoi painter as well as a dielectric material which uses the Henyey-Greenstein phase function for scattering. It also includes support for the Radiance .HDR file format. RISE can now read and write HDR files. You can also specify a radiance map for a scene which can be used for doing image based lighting. I have decided with the recent changes to the color code that internally RISE should only be concerned with floating point colors and not even deal with 8 or 16 bit types. Eventually, the rendering engine component of RISE will only know about floating point colors with conversions to other formats being done externally by utility apps.

    Thursday November 6, 2003

      There is a new source release available for download. It includes recently refactored math code as well as better compilation options for GCC and better compilation support on SGI MipsPro compilers. The MPI version is also working again. The build target is still risempi

    Friday June 27, 2003

      There is a new source release available for download. With this new release, the primary focus for the library is in being a useful scientific visualization tool. The new release includes stuff like black body radiators, a new shading interface, 16 bit PNG images and global photon maps (very naive unoptimized implementation). The RISE application also has a -pr flag which attemps to quantify performance in a platform independed manner. I have measured the performance of RISE with different compilers on the same machine. The results are in the "Platform" section.

    Tuesday June 3, 2003

      There is a new build with source available for download. This build contains some more refactored code and should also be solid enough to compile and run on most platforms. The code looks a lot better from all the refactoring, but naturally performs slightly slower. I need to do a round of optimizing. This latest build also includes several sample scenes to give an idea of how to use the really simple parser. I have also started working on a .NET UI which will be the main R.I.S.E. application for Windows. To this end, the IJob interface is now exposed as a managed .NET class. I have only started writing the .NET application, so it will probably take several months before its even close to ready.

    Monday May 12, 2003

      I have refactored a bunch of code and seperated the library from those applications that use it. I have also created the RISE_API and the Job class which are useful for accessing the library. All of these changes are availabled in Build 61 of the source distribution in the downloads section.

    Wednesday January 1, 2003

      Before the holidays I finished up a reasonable implementation of DRISE (heterogenous massively distributed cluster-based rendering) and PRISE (homogeneous distributed memory multiprocessor rendering of large scenes). There is a new release available with these changes. This will probably be the last release for some time, as I don't have the time to devote to this project. I may make some minor updates, but probably no big features.

    Sunday December 9, 2002

      Added support for precompiled headers, so now all .cpp files include "pch.h". I use the PCH in the MSVC6 environment and this sped up build times by a factor of about 5. I am going to implement the PCH in the GCC and SGI MipsPro versions as well. Apart from speeding up compilation times PCHs are also useful to keeping compile time, compiler dependent settings (such as turning off certain useless warnings).

    Wednesday December 4, 2002

      I have changed the Log so that it doesn't use reference counting anymore, this means that everything compiles under GCC 3.x again. Once the bug in GCC is fixed, I will add back reference counting. I also discovered that the version of the STL included in GCC 2.9x is severly lacking (examples of missing things are std::make_pair() and std::vector::at() ). Thus I have revised a few pieces to not use these functions so that we compile under GCC 2.9x again.

    Monday December 2, 2002

      I have finished with the first implementation of the distributed rendering system. I was able to have about 10 computers rendering a scene together with many of them coming and going at various times. There are still many quirks and bugs, but the basics are there.

    Saturday November 30, 2002

      The source code has been heavily refactored, it now adheres to a simple interface / implementation pattern almost everywhere. Many of the files have also been moved around into subfolders for better organization. Apparently ther is a bug in the GCC 3.x compiler when using classes with virtual inheritance and var args (...). The compiler doesn't compile, and spits out a rather useless error 'generic thunk code fails for method ...'. This bug doesn't exist in GCC 2.95.*. The bug is reported here. Since there is no workaround I can see, GCC 3.x will be unsupported until the bug is fixed.

    Monday November 25, 2002

      I am full swing into writing the distributed extension to RISE. I have written the utilitary socket communication code and it seems to work well enough. I have also cleanup up a lot of the code so that it is a bit more consistent. I have also implemented a memory tracker as part of log to help detect memory leaks.

    Wednesday October 23, 2002

      I have fixed a whole slew of bugs and also reduced memory consumption of meshes by 50%. I have also added a few little speed optimizations. There is a new release available in the downloads section, build 55.

    Sunday, October 13, 2002

      I found and fixed a minor bug in the ProbabilityDensityFunction class, which was causing a crash only on the Alpha architecture. Apparently, there was a division by zero, which is 'handled' gracefully by all the other platforms, but not Alpha. Amazing the kinds of bugs you can find by getting software to run on another platform.

    Wednesday, October 9 2002

      I have removed unnecessary templates in an effort to make everything compile faster and as a smaller executable. Thus the math and pel classes are no longer templatized. I have also limited which files include headers that contain templates. This has resulted in a slight decrease in compile time and the executable now compiles to a size several orders of magnitude smaller.

    Wednesday, October 9 2002

      I have cleaned up a bit of the code. I have also optimized the NRooks Sampling kernel, so that it generates its kernel faster. I also now regenerate the kernel for every pixel, and also regenerate the kernel every time a light is sampled. This will ensure that even the pathalogical cases of aliasing are addressed (ie. say the frequency is precisely the size of one pixel and its inoppurtunely located where there are no samples). I didn't think it would even matter, but this way, it is guaranteed to be more correct.

    Tuesday, October 8 2002

      I have decided to take a little rest from R.I.S.E for the next few months. I won't be working on any new features (stuff listed in 'Currently Working On' in the Features section). I will however be actively working on adding robust MPI support for RISE, including the ability to render massive scenes in parallel. I will also be making optimizations and generally improving the structure and code organization. I plan to add some intelligent caching schemes in an effort to improve performance, meshes in particular. I also plan to drastically reduce the memory usage (particularily for meshes).

    Tuesday, October 8 2002

      I have added support for simple spectral caustic photon mapping. I rearranged the photon mapping code to make it a bit more extensible, and then added the spectral support. Currently just like regular Pel photon mapping, only the standard ray tracer supports photon maps. I still need to get the path tracers to support photon mapping, as well as optimize the photon tracing.

    Monday, October 7 2002

      The simple web log is now active.

(c) Aravind Krishnaswamy 2001-2005

Generously hosted by: Logo