I've also settled for the Adaptive Subdivision with the Mitchell-Netravali antialiasing filter as our final image sampler, and it creates very high quality images with no visible edge aliasing, just the way we like it.
Seeing I'm in charge for the rendering settings, being "the V-Ray guy" and everything, I've put together a flicker- and noise-free solution for our project. What I'm doing is rendering out the lighting information for every frame (prepass) and then render the full thing blending the lighting information to get a smooth and flicker free animation. This method is not only improving the quality of the render output, but also decreases overall rendering time considerably. It's one of them win-win situations really.
I've also settled for the Adaptive Subdivision with the Mitchell-Netravali antialiasing filter as our final image sampler, and it creates very high quality images with no visible edge aliasing, just the way we like it.
0 Comments
As we're getting close to the deadline I've been spending some time on creating some long needed texture maps for "Dave". He now looks a whole lot better and fits more with our current art direction. I also created a "overrun" variation with a lot of dirt and shoeprints for the third train incident. I've also been experimenting around a little with compositing in Photoshop to achieve different looks to the final output, and I really like where this is heading.
We have now pretty much just chopped up the movie into four parts and we're now ready to start animating one part each. I've included my first shot here which currently is in its second day of development. The first day pretty much went into camera motion and timing of the animation. I also started animating the feet. On the second day I kept iterating on the animation and I think I've got a pretty good rough going here. I feel the timing is a little off here during the running. He seem a bit slow for someone in such a hurry. I'll try to address that as I go on.
I'm having a hard time deciding on which image sampler to use for this movie. Adaptive DMC is definitely the faster one but it produces too much noise (to the degree of making background objects unrecognizable) and lowering the noise threshold increases the rendering time significantly. Fixed sampling creates a more predictable and clear render but the rendering time per frame is too long here. Finally we have the Adaptive Subdivision which appears to be much like the DMC but without the extreme noise. I find the rendering time on Adaptive Subdivision is still a bit high (currently at around 4 minutes per frame), so I will have to keep experimenting through the animating process to get us a satisfying render time to quality ratio.
Speaking of rendering times, I'm considering replacing all reflective surfaces with specular highlights and disable reflection calculation completely as that will shave the rendering by approximately 30 seconds. I've also finished up the main scene adding in the props as XRef objects for easy future adjustments and created some trash and scattered it around for a added touch. Also, the VRay Dirt Map suddenly started bugging out on me for reasons unknown, so I had to get rid of my procedural dirt shaders and have instead increased the occlusion density to compensate. Anyways, here's a few renders of the updated scene: Øyvind had some trouble setting up the rig for our main character and has asked me to do it. Of course I accepted the challenge. So I've been spending the weekend on creating a functional rig for Dave. This is my most advanced rig to date complete with IK/FK blend switches for the arms. It isn't perfect but it sure is enough for our short film.
I've been spending some time updating the shaders for most of my props to give them a bit more of an gritty, diffuse cartoony look. I honestly think it looks way better this way, besides less reflections == less render time. Can't complain about that.
Seeing as we need a couple of short dirt and smoke effects, and don't want to spend hours at learning and simulating "realistic" particles, I've been experimenting a little with a generic smoke effect using a simple ParticleFlow setup. It's amazing what you can do with some spheres, falloff/procedural maps and a bit of creativity these days...
For this part we decided to create a character each, all the way from concept to rigged character. We're all done with the modeling stage at this point. I'm pretty much done with my entire character right now, but the others aren't that far behind. As I've been working ahead of the others during the process, I noticed the others had copied and pasted the hands of my model into theirs. I find that amusing, lazy bastards :P. I've also been kind enough to retopologizing Ole Petter and Christoffer's crazy head topology. Everything seems ready for rigging and skinning now. This project has officially got me interested in rigging - To look deeper into the rigging process. I believe my rigging skills have multiplied several times just during the construction of my rig. I still don't like skinning that much though. _____________________________________________________________________ |
About the authorOle Kristian Busk Archives
August 2018
|