The components of the tower: The two sides of the tower, the balcony, and curiously happy renditions of Saruman and Wormtongue. Sarumun's hair and beard were added in Carrara.
The banner, the high-poly orc, and the low-poly orc.
A single orc model was used, created with MLCad's minifigure generator. I used LeoCad to export the model to a .3ds format, and imported this into my 3D application of choice, Carrara Studio. One I had the model in Carrara, I created a high detail orc and a lower polygon count version. (Less polygons means less data and faster processing, but it also can result in rougher edges and less detail.) The five or six rows closest the camera were filled in with the high-polygon orc, and the low-polygon one was used for the vast expanse of orcs in the distance. Even with these optimizations, the thousands of orcs were still very taxing on the computer, and I had to work with the scene in bounding-box mode, which represents each object in the scene with a wireframe box of the same color. I realized that the pikes the orcs were holding were too short to reach the correct height in-camera, so I extended them. I also extruded the ridge on the helmet to make it resemble the helmets the Uruk-Hai wear in the film. The flags were spears with a textured flag object attached to the end, designed to look like fabric with a crudely painted White Hand of Saruman. I gave the armor a shiny, plastic-looking sheen. The courtyard has a number of large pieces of machinery in the distance, which were simple structures built in MLCad.
Once I had imported and textured all the models in Carrara, I set out to re-create the composition of the original scene. I carefully compared my scene to the original, and after many, many camera adjustments got it to be close to the original. I played with the perspective some; the tower is not to scale with the orc army, but it appears to be due to the angle. It took some time to figure out the focal length of the lense used in the real shot, but this was essential in order to make the angles match up how they should. Using stock photographs, I cobbled together a sky and mountain background that resembles the background seen in the original footage. I used a smoke function in Carrara to simulate the smoke on the horizon. Lighting the scene was also a complicated ordeal due to the overcast day lighting of the scene. The final scene contained 11 lights of various colors and intensities.
Finishing the renderDue to the complexity of the scene, the rendering process for the still image took about half an hour. I used the highest quality settings on everything, because in scenes like this with lots of repeating textures and details (for example, the studs on Orthanc) lower quality anti-aliasing tends to result in ugly webbing effects. Anti-aliasing is a process which smooths the edges of objects to make them blend together in the scene. After rendering the image, I wanted to give it a more realistic, gritty, film look, so I adjusted the coloring and added some motion blur in Photoshop Elements
Wireframe view used during the composition process.
Final unprocessed rendered image.
Final processed image.
And that is the process I used to create this image. I enjoyed making it. It would not have been possible without the extensive LDraw database, which is an invaluble resource for creating these scenes. There are many people who use LDraw software for the purpose of modelling alone, but it can also be a great tool for creating art. I hope that others will be inspired to create scenes using the LDraw software; I'm excited that there have been some great entries to the LDraw Scene of the Month contest recently, but there is always room for more. Keep building!