CLICK HERE TO GO TO DIARY POST 1-2
DIARY POST 3:
OK... so for this production diary post, I'm going to discuss how we implemented the production in various scenes that required particle and fluid type effects. Obviously for any show that is telling a story about rocks hitting the Earth, you are going to need to show various objects entering the atmposphere and causing all sorts of distruction when they impact. Our initial concern was that we would be dealing with very large scale particle effects requiring millions of particles.
It's one thing to create particle effects for a motion picture where you are dealing with a framed view lasting a second or two. But with fulldome shows, you are talking about an immersive view of a scene and 10, 20, or 30 seconds of viewing time. Beyond that, you are limited somewhat when it comes to layering... I've yet to hear of someone sucessfully using a z-depth channel technique for every need you come across in fulldome production compositing. Getting the feeling of depth... objects being obscured in a mass of particles can be quite a challenge.
For the atmospheric entries, we used a combination of particle emitters with per-particle lifetime shader expressions (to show the hot initial trail dissapating into a smoke trail) and the Maya fire effects plugin... pretty basic ways of creating the smoke trails. For impact explosions we needed something more sophisticated to avoid having far too many particles per layer even when planning on using many layers to achieve a look. Even so, we experienced frame renders of many hours, even on a rackfull of the latest eight processor render servers.
These are frames from the Tunguska sequence. The large explosion was created using particles to generate the dynamics... then connected to a fluid to generate the shading. This produces a much more volumetric feel and look... thus a much more effective sense of a massive scale to the scene. At times, some scene objects would be rendered in the same layers with smaller quantities of particles to get our "depth" look. Then more layers would provide the complex/thicker overall look needed.
These frames from the final "future" impact sequence were our most time-consuming renders... involving close-up views of the particle generated shock waves, with lots of transparency calculations to get depth. The oil rig scene also featured a tital wave caused from the impact of a small comet just off the coastal area near Houston, Texas. The show will premier at the Houston planetarium.
Next time I will discuss the final staging and lighting process, the last step before rendering and final compositing.
TC
DIARY POST 4:
Let's finish this series of posts with a discussion on the final steps in the production pipeline... that of staging and lighting. Staging is the term we use for layout of a scene to create the most dramatic composition. It involves considering what will look best on the dome from the audience's viewpoint. In deciding this, traditional film-making rules still apply, but for the fulldome producer, the decision has some extra considerations, that of what works best in an immersive image. You need to consider the sweet spot (where the audience is generally centering their view) and also the effect of cross-bounce of a bright area on the dome... not an issue for a framed view.
The older black and white films are a great inspiration for fulldome composition. Since there was no color, light and dark areas were the only thing to work with in creating points of interest. This works well in the fulldome world, since we don't want to fill too much of the dome with bright areas and wash out the entire scene with all the cross-bounce.
In the film-making world, there is a term, "the magic hour," referring to the early morning and evening timeframe when the Sun is low. Because of the warmer color temperature as a result of the deeper atmosphere the light is passing through and because the shadows are longer with more edge lighting of objects, a scene looks more magical. Film has a way of taking this timeframe and enriching the scene. For a fulldome scene created with CGI this is often overlooked unfortunately.
Let's see how these thoughts apply to a scene in our show. The Tunguska event occured early in the morning... perfect for our magic desire. Originally the animator threw a light into the scene coming from the back of the dome view to front light the scene. It looks fairly good as long as the backgroung sky is not considered.
Unfortunately, with the background sky (a composite of several photographic fisheye images) added in, the image would not have much visual contrast with everything being brightly lit. Also the light is coming from the wrong direction for the Sun... and the "Tunguska rock" came from the southeast... our setup was to have it come from behind the farmer and have him look over his shoulder startled. By changing the light direction we get a more dramatic lighting feel with lots of long shadows and edge lighting.
We then adjusted the shadows to soften them and also played with the color to make it look more like a sunrise than a sunset. Although the initial scene's lighting had some validity with the more dimensional look of the fence and building... at times you need to NOT be afraid to "cover up" all the modeling work to achieve a more story-telling feel (it's tough on the modelers at times). We also added a slight blur to the brighter areas to spread them out a little (sort of a film-look bloom) as well as added the typical contrast adjustment for the dome's advrse contrast effect on images... then a fog layer to add further depth to the overall scene.
The resulting look creates a definate morning feel... much more in keeping with the story. The farmer, fence and building is silhouetted against the sky to render well on the dome projection.
Typically, we try to work with no more than two or three lights in a scene. A directional light representing the Sun and then an ambient light to fill in shadow areas somewhat. A third "edge" light will be used at times to create an edge on objects and separate them from the background. We also usually "bloom" or glow the top 10% of an image's tonal range during compositing to create a look that appeals to the eye more. Since we can't project anywhere near the tonal range that the eye has the ability to see, this blooming approach compensates somewhat to trick the eye into thinking it is seeing more.
Well, that's it for now. I hope something we said here was insightful. I would love to hear any comments. We will do another "Inside the Production" series again soon.
TC
