Week 7
- mialaplanche
- Feb 18
- 2 min read
Weaving Week
We chose Shot 1, our weaving/threading shot as our sub hero shot so need to put some extra attention on it. Last week I added the thin to thick fix by request of the lighting artists who said it would be easier if the weaving started thin lines and the when it actually arrived at the pattern it would be the thicker cross section. In figuring that out, I didn't notice the hair no longer followed the weave pattern perfectly and was instead jittering with all the motion of it's new "scalp", resulting in the poor render we got last week.
Initially I had an attribute VOP that was creating the wavy noisy pattern of turbulence against the straight weave to give that wavy cloth look. That attribute VOP was feeding into my attribute wrangle that was interpreting my attribute transfer reading colors of red and blue, if Cd.r < 0.9 then it's the the first weave pattern off camera and if Cd.r > 0.9 then it is the second weave pattern on camera. To fix the thin to thick I based it on position on the Z axis instead of Cd since it kept giving me a gradient instead of clamped values for the radius of the thread (which I discussed on my Week 5 blog).
The attribute VOP feeding into the wrangle was screwing with the normals of the subsequent polywire I made for the actual thread (which reads the value of width from that attribute wrangle). When I would visualize my guide deform feeding into my hair, it was jittering and flipping directions which indicated it was a normals issue. My new solution was to do the attribute VOP after the hair has already been generated on the straight flat weave version.
We also had some issue with the renderfarm just giving us back grainy renders filled with random artifacts that are not present in local renders. Running a husk didn't really reveal any issues but bumping up the minimum amount of RAM to 256 instead of default (0) required for the shot in the farm settings helped isolate just the render nodes which stopped the artifacts.
Comments