Big day for the team on our 3D&2D piece
The whole file transfer and management side of things is quite complex, intricate, fiddly, essential, empowering and frustrating. The notion of doing this on a major production with 60 people sharing assets and working in parallel is quite daunting seeing all the little issues we have to solve on our midget 15 seconds with 3 people and basically one piece of software.
So far the major file-swap times have basically worked, but there are often things that dont quite come together nicely.
Using the referencing method is very powerful, but some more experience on this side would be valuable. Locking down the model file from editing would be one idea, as soon as we have tweaks and edits in the rig, then we have to deal with oddities when we re-reference the model. We were also very cautious in this process about adding geometry as that always seemed to mean a chunk of re-rigging. I also find myself holding back on one part of the production process and waiting for a 'more final' version of something because of the generic fear that any changes will break things and then they would need re-doing.
It is definitely better to fully model up a character (with lots of attention to how it will be rigged an animated) before the actual rigging begins. Plus to get the rig fully resolved and working before any animating begins. The workflow of Kevin Lanning I looked at on Day282 is a good example of this.
I cant wait until I can try my hand at this kind of thing, high and low res modelling, texturing, rigging and animating for game engines and cinematics - awesome.
Looking in detail I noticed we have 2 time versions of the photo-fall scene (shot10). One at 1.5 secs and the other at 2 secs. Since some of it is obscured, the end result may well be in between. I will have to get a copy of After Effects or something so I can start learning the video compositing side of things. iMovie just doesnt seem to sport enough features to do what we will need (I think).
[WIP of Shot10 with the falling photo]
Since I already had the photo-fall animation built from my test earlier, I grabbed that file and imported the main shot sequence file into it. Eliminating the characters, old lights and such it was back to a nice clean scene. I created a new shot10 camera and lined it up with pretty much how the test looked and referenced the animatic again. I added a new point light with raytraced shadows for this shot and then it was ready to render out.
Though the render process is nice - each time I would render it out I would see something in the lighting or framing etc that could use fixing. I created 2 boxes to mask out the wide-screen-ness of this shot. Hopefully the compositing stage will cope with this. So after a few more versions I have shot10 rendered out - cool.
[youtube vid of the photofall sequence also known as shot10]
It was interesting having James in at COFA working on his copy of group's files while I worked on mine. We communicate on the fly via GoogleTalk and then Gmail each other updates.
Even though we had a working model of 3D, I just couldnt help but run a bunch of refinements over him. I enjoyed this actually as I got to play with a bumch of things that I hadnt really got into before. I reworked his forearm UV maps, adjusted some geometry, reworked the outliner (I like using this window to select things) and a few other bits and pieces. The hurdle that took me the longest was working out scaling him. Scaling a rigged and skinned character with all his handles and such is painful. I ended up scaling the full model then scaling back the mesh to match the rig - this eemed to work out ok. In total this took quite a while, but was good experience.
[WIP shot of the construction scene now with 3D in place properly again]