3.2 KiB
Minutes from 41st OpenVDB TSC meeting, February 20th, 2020, (GMT)
Attendees: Nick A., Dan B., Jeff L., Ken M.
Additional Attendees: Bruce Chernia (Intel), JT Nelson (Blender)
Regrets: Peter C.
Agenda:
-
Confirm quorum
-
Secretary
-
Quorum was confirmed.
-
Secretary was Jeff Lait
-
PR 455
Looks good. But Peter suggested he did not want to merge?
Resolved: Add and name the C++ tool guassianSharpen, but keep the UI name Sharpen.
- Default branch
Default branch is now Develop. We must change branch when we do PRs.
- VDB Merge
The initial round of feedback on the UI has been read. Another push will be done for final verification.
The confusing Activity Union toggle Will be moved into the activity collation options.
- JT Nelson
Working on Blender integration of VDB. Have a fracture modifier integrated. Working on crushing technology.
Adding solid geometry support.
Should there be blender path in the openvdb directory? We've not maintained Maya branch well. We should consider outreach to Autodesk to see if they can maintain it.
- Roadmap
What does priority mean? Personal affiliation vs OpenVDB priority?
We decided this is some combination of the two.
a) AX
A new release next week. Performance improvements. End of March will be ready for feature branch. Needs documentation and testing. Language spec still incoming. Can prioritize the spec. Spec is currently in Doxygen and private, Will be ported to Google Doc for commenting.
b) Fast Sweeping
PR exists, not yet pushed. Very close.
c) Sharpening
Discussed this meeting.
d) Counting Tools
Agreed it can live in both places. Leave for Google Summer of Code
e) Merging Tools
Part of VDB Merge project, ongoing.
f) Topology Tools
Moving tree API to tool and accelerate. Likely want to deprecate tree methods.
g) Fill Tools
Straight forward method to move off tree and into tools.
h) Densify
Similar to Fill Tools
i) BBox
Similar to Fill Tools
j) Extrema
More complications. Confusion exists now, is easy to make very slow by accident. Should be separate tool.
k) Mask/Bool Resample
Low res -> High res has to be optimized.
l) Reducing Memory Footprint
No discussion noted.
m) Adding more interpolation types
Higher order, monotonic cubic
o) Point Merging
Ongoing phases.
p) Surfacing
VDB From particles is used a lot for surfacing. Need better surfacing. Artist prefered fast base surface and then do their own adjustments. But likely time to look at again.
Add a single SOP that does the single method? Like Zhu-Bridson. But rasterization is the key. What about MPM kernels? Similar rasterizer.
Solution seems to have add the rasterization infrastructure and Zhu-Bridson solution. This could then be extended to MPM kernels.
q) Mesh Support
Contentious discussion of if this is a good thing. Some members feel it is an overreach of what VDB is supposed to be. Others feel real world production experience shows it to be an extremely useful component.
Topology information can be passed as light weight information. Separate side car of primitive data and topology information. Works well for changing topology meshes like fluid meshes.
- Next meeting
March 5th 2020. 2pm-3pm EST (GMT-5).