Welcome to GENETICS!
GENETICS (Generating Enhanced Natural Environments and Terrain for Interactive Combat Simulations) is an NGA-funded research project being undertaken by the Naval Postgraduate School's MOVES Institute to use readily available source data to procedurally create vast detailed landscapes at runtime without the need for a terrain database or proprietary tools.
I am outta here! Here's my dissertation and slides I used for my defense. For those that wonder what is going to happen to GENETICS, it is now undergoing a major rewrite by the Delta3D, so bug them often about the GENETICS release date and accompanying tutorials!
Here's my IMAGE presentation. Beware! It's nearly 40mb.
The MORS Symposium was last week and I'd like to thank those that hung in there until the last day to hear my talk. I definitely want to thank those who, after hearing my talk, still asked for a copy of the software! (Your check is in the mail.)
Finally, we have now demonstrated that GENETICS, when given the same parameters and source data on two different machines, created identical landscapes. Screenshots to follow soon.
I'm calling this image "Unsafe flying within GENETICS". So the update is that networking is... err... working. I'm using Delta3D's HLA capability is near idiot-proof. Also, I've implemented a bit of collision detection with respect to placing trees so that vegetation objects won't crowd each other so much.
On the less happy front, the experiment with using OpenGL shader programming to place the vegetation objects (ala "instancing") is not going well. At this point, we might have to cut our losses and rely on a smart GLSL person to come along and figure this out. That means that the current scenario of ~2M vegetation objects running at 15 fps within a 1024x768 window might be around for a while.
Testing a higher-resolution dataset (about 10 times the vegetation coverage than previous attempts). Only works for sub-geocell scenes. Pictures are in the Gallery showing vegetation placement on top of blurry (Landsat7) satellite imagery.
The new computer came in! We now have a Shuttle SN95G5 box with an AMD Athlon64 FX-55, 2Gb RAM, and a NVIDIA 6800GT. That resulted in an instant doubling of performance in some scenes over our "old" Shuttle SN41G2, AMD AthlonXP 3000, 1Gb RAM, NVIDIA 6800GT box.
We've been added to the Virtual Terrain Project's repository!
Made some changes to the Gallery. Latest images now appear on the main page. Clicking an image gives you the full resolution screenshot.
Good news! Performance since January has nearly quadrupled. We are now loading over 1.2M vegetation objects in our initial 1 deg x 1 deg terrain cell and displaying these objects at 15 fps (1600x1200 resolution). What does this mean? We can now create denser forests! Now on to optimizing the REALnat objects. See some examples of our forests in our Gallery (larger images forthcoming).
We have made some initial attempts at implementing slope aspect into the vegetation age chooser. Thus, a favorable aspect angle will tend towards older vegetation objects and less favorable will tend towards younger vegetation objects. While the initial results aren't completely satisfactory, at least it adds variety to the landscape.
A brightness control was added to increase the brightness of the ground texture (useful for dark satellite images).
We're happy to announce that this project has already met with some early success. We've ported the SOARX continuous level of detail algorithm to our Delta3D game/simulation engine. We can import DTED elevation postings to create our terrain mesh, TIGER vector data to create roads, and NLCD images to place vegetation. Terrain surfaces are either GeoTIFF images or psuedo color-ramped.
We've demoed our software at the MOVES Open House in Aug 2004, on the exhibit floor at the I/ITSEC Conference in Dec 2004, and at the G.A.M.E.S. Synergy Summit in Jan 2005. We'll also be presenting GENETICS at IMAGE 2005 in Scottsdale, AZ on 11-15 July 2005.