The DIRSIG was formally created in the late 80's at RIT as a project by a group of students. Since then, the model has been improved and maintained by a combination of full-time research staff and graduate student projects.
The faculty, staff and students attempt to publish in-depth papers on various aspects of the model on regular basis. We believe that this is an important difference between this model and others. We attempt to keep copies of recent papers on the DIRSIG website on the "Documentation" page (http://www.dirsig.org/docs). On this webpage you will also find a list of published journal and conference papers.
The model is generally limited to the same wavelength regions as the MODTRAN and FASCODE atmospheric models, which is approximately 0.4 - 20 microns, or 450 - 15,000 wavenumbers.
Again, the model is generally limited by wavelength restrictions of the MODTRAN and FASCODE atmospheric models. If you are using MODTRAN exclusively, then you are restricted to the 1 wavenumber resolution limit of MODTRAN4 or the 0.1 wavenumber resolution limit of MODTRAN5. By also using FASCODE, you can increase your resolution in some special cases.
The DIRSIG software is distributed as binary executables only for the following supported architectures:
Sun Microsystems UltraSparc Servers and Workstations running Solaris 2.9+ (machine must be up-to-date with all patches)
Silicon Graphics (SGI) Servers and Workstations running IRIX 6.X
Intel Pentium and AMD Athlon Workstations running the Linux 2.4+ Kernel and Glibc 2.0+ (for example, RedHat 8+, Suse 8+, etc.).
For more information, read the Installation chapter.
At this time, we encourage that you become a funding source for an academic research project. This will entitle you to some degree of support. At this time, the university does not have the labor power nor stability in funding (for this type of work) to maintain a dedicated staff for support.
In some cases, we have worked directly for organizations to help them incorporate the use of the model into their efforts.
The run time for a simulation varies based on the type of simulation being performed. A small-format, low-spatial resolution sensor looking at a simple scene may take a few minutes. A large-format, high-spatial resolution sensor looking at a bigger scene may take as long as a few days.
The parameters that dictate run-time are the total number of pixels to be modeled, the amount of sub-pixel sampling, the spectral resolution, the scene complexity (number of objects, number of facets in each object, etc.), the number of materials, the size of the images used for material, texture, etc. maps and if the thermal model is turned on.
DIRSIG3 was written entirely in C with the exception of the radiometry library which was written in primative C++. DIRSIG4 is entirely written in C++ using a very object-oriented, modular design consisting of 250+ individual classes.