Blender 2.57 RIBMosaic 0.1.1 Aqsis 1.7 Pipeline Test
One of the only reasons the pipeline has stayed in the old version of Blender was stability. As of right now that pipeline is considered production proven, actually there is a studio that uses Blender, RIBMosaic and PRMan to create animated cartoons. Project Widow has also been using this same system, simply for stability, however now that Blender 2.57 and RIBMosaic 0.1.1 are in a better state of development, we are beginning to upgrade the pipeline to these tools.
Project Widow is about pushing the edge, at first that was the stable pipeline, which over the course of two years has gone from an experiment to production capable. At the start of production the tools that we had to our disposal were not as built up as we wanted and over time they improved. So much so that this small studio in Costa Rica is using the legacy RIBMosaic in their pipeline, has their team trained on RIBMosaic and is producing cartoons with them. In light of this, plus our team is getting more adapted to Blender 2.57, the upgrade was needed to test out the next generation so we can raise the bar again. While we are now not the first to produce such imagery with these tools, we are however the primary test subject and as such need to kick it up and get cracking on the new tools.
Aside from some issues, it went off without a hitch, as this was planned for some time now, it was just a matter of time when the tools were stable enough for the transition. Of course this means that the shading and lighting data needs to be redone but since the shader source already exists for the current shading pipeline, that too will be a seamless transition. All that we need to do now is provide feedback for Jeff Doyle on RIBMosaic as well as the Aqsis team for the developmental 1.7 builds. While on one hand the shading pipeline is still sketchy, the transition from shader fragments to shader pipelines in RIBMosaic are proving to be a bit complicated, so it will be some time before the shading pipeline is considered “stable”. This happens. Plus we are still trying to just animate the damn short first, then worry about the rendering end afterwards. So work already has been done to convert the easier animated scenes to the new format, this involves removing the legacy RIBMosaic from each scene, not an easy task both time wise and emotionally as the amount of work that went into developing the pipeline took a toll, now only to be dismantled and rebuilt. Such is life in the world of code right??
One of the cooler features of the upgrade is the automatic feed through the compositing pipeline, so all Aqsis renders can be fed through that portion of the pipeline on the fly. Not an amazing feature but Blender is using Aqsis as an internal renderer now. That is something that many in the community have been wishing for many years, now it’s actually here. How the final composite workflow happens in still in the planning stage, so that alone will take up a good chunk of R+D.
Over the week quite a bit of testing went on in the pipeline, mainly to see what RIBMosaic is capable of at it’s current state, as well as learning the “new” way of doing Blender to Renderman. It has come a long way, in fact this is what we have been imagining for years now, just the ability to use Blender as it is and not to use a different window just to set up Renderman settings, options and variables. While there is some issues to deal with, such as the XML pipeline workflow, not to mention shaders are not being found, the actual export and rendering at a basic level functions fairly well.
On the Shotgun side, there are pages that are linked directly to the bug trackers of RIBMosaic and Aqsis, we usually don’t make bug reports on Blender because for the most part there are thousands of other users that provide enough reports, plus the majority of the bugs we encounter are not Blender side. If we do encounter such a bug, we make the special report but they have been very few and far between. Much of that falls on Jeff Doyle as he has a much better grasp at the code and what to look for, he is the primary developer after all and would encounter them long before we do. While we do have an internal ticket system on Shotgun, much of that is devoted to the actual pipeline for Widow, any really deep technical issues then get raised on the official software bug trackers.
In all the pipeline upgrade is going smoothly, though time consuming and we are really happy to see that we are able to undertake it despite the headaches and irritations. We never actually thought we would be working on Project Widow still so the thought of being able to actually render the problematic issues of the previous pipeline (such as animated curves) might come to a reality now.
