The project is coming along quite nicely! While lighting the scene, I have noticed one particular element that has made this part of the pipeline move along much quicker than originally anticipated. Before we began lighting, we planned that a general lighting setup for the entire scene was needed, followed by shot-specific lighting for each of the four shots. Once we finally got deep into the lighting phase, we noticed the lighting held up in each of our four shots consistently. It didn’t take long for us to understand what was actually going on here. As our previs demonstrates, the camera in all four shots stays on one side the temple, city, and environment the entire time—it’s almost like all four shots are parts of one long continuous shot moving through the scene. With that in mind, we realized we’re basically lighting everything for one shot, therefore reducing the amount of lighting work we thought was needed.
Since we have been lighting, we’ve noticed some strange behaviors occurring with our shadowing casted over the environment. Observe the problematic, crescent shaped shadows in the image found below.
These unusual visuals are unacceptable, and like anything else in visual effects, we had to problem solve and figure out why we were getting these results. So after a while of looking through everything and trying to understand what was going on, we realized the problem existed within the geometry we sculpted in Mudbox—there were a few polygons protruding too much, creating the peculiar looking shadows. At this point, we had two options available to fix the problem. First, one was to turn our key light into an Area Light, but the issue with this route was that all of our shadows became too soft. And, if you have ever observed shadows casted by mountains before, these shadows tend to have a crisp edge along them. So our other choice was to “smooth” the entire landscape by one iteration. While the shadow problem was fixed with this method, the trouble with it is that the environment loses some of its form, definition, and becomes softer looking due to the extra geometry generated by the smoothing process. The simple solution to this was to “smooth” the environment but turn the “continuity” attribute to “0.” Not only does this generate enough polygons for better, smoother shadow results, but the geometry maintained its shape, form, and it added an extra level of detail that went unnoticed by us when translating the environment from Mudbox to Maya. As you can see below, the shadow results are much more desirable.
On one final note, consistency between all the shots is a major concern of ours. Just like the one-shot light rig we have for the entire scene, we have decided on one other aspect to keep the visual quality similar, no matter the shot. Taking note from Pixar, we have worked out a pipeline that allows for a major chunk of our project to be done and completed through the 3-D renderer without the need of compositing every element. While there will be some compositing involved with our project, we are interested in keeping everything consistent, and keeping some components together within the 3-D renderer maintains a certain level of regularity that otherwise might be a little more difficult if you tried to composite everything for each shot. Yes, taking this approach adds more work for us upfront in the 3-D applications, tweaking everything to look appropriate right out of the renderer, but it saves us more time and requires less work for us in the end come time for compositing. …It’s a give and take sort of situation.
This workflow is not the only way of doing things but it works for our project, and below are the results thus far of using this procedure. Please understand that these shots are not composited, the environment is the low-res model, all with low quality render settings, more tweaks are still needed, and other elements, like particle fx and matte paintings, still need to be included.
- Joey
No comments:
Post a Comment