3.6 KiB
Minutes from 91st OpenVDB TSC meeting, May 18th, 2021, (EDT)
Attendees: Nick A., Ken m., Jeff L., Dan B., Andre P.
Additional Attendees: Jeff Budsberg (DWA), JT Nelson (Blender), Johannes Meng (Intel), Bruce Chernaik (Intel), Sergio Rojas
Agenda:
-
Confirm Quorum
-
Secretary
-
Project showreels for SIGGRAPH
-
8.1 Progress Update
-
V9.0
-
VDB Extrapolate / Fast Sweeping Questions
-
SIGGRAPH course
-
Blind Data (PR1050)
-
DynamicNodeManager Grain Size (PR1024)
-
Next Metting
-
Confirm Quorum
Quorum is present.
- Secretary
Secretary is Nick Avramoussis.
- Project showreels for SIGGRAPH
The TAC has asked for updated media material from each ASWF project to be able to update each project's showreels. Any and all material that showcases VDB in production is valid. Individuals with available material should look to contact Rachel Romoff at the ASWF.
- 8.1 Progress Update
All known issues have been fixed and merged. The targeted release date is the week beginning 31st May (~2 weeks from this meetings date). ILM, DNEG and SideFX able to test. Jeff to merge consolidation of pendingchanges and Dan to create a 8.1 branch which represents the release milestone. Ken to check any outstanding EXR build issues.
- V9.0 & VFX Platform
The VFX platform draft for 2022 has been published with VDB still listed as 8.0. The VFX platform has not reached out this year for confirmation though they do want to start working more directly with ASWF projects. Jeff contacted VFX platform for comment - current policy is that they do not add new versions until they are available. Jeff, they should not list versions in the draft unless they have been somewhat confirmed (instead leave them blank). It would be good to have them attend the TAC meetings. 9.0.0 is targeted for November. Dan, we could do a release candidate for 9.0.0 so that it at least exists. This would really be a last resort.
- VDB Extrapolate / Fast Sweeping Questions
Dan, when should the Renormalize SDF mode on the Extrapolate SOP be used over the Renormalize SDF SOP. The SOP seems to be faster than the extrapolate. Renormalize solves the eikonal equation on every voxel in the narrow band, where as fast sweeping only performs 8 sweeps. However the sweep orientation/ direction effect performance in relation to the topology structure of a SDF. Fast sweeping should not move the interface at interval steps (8, 16. etc) where as renormalize will. The real power of the fast sweeping tools are extension and extensibility (e.g. it can do asymmetric narrow bands). Renormalize SDF SOP propagates 1 voxel at a time such that the performance will scale with the narrow band size, in comparison the fast sweeping will scales more on the topology layout relative to the sweep orientation. No clear answer, more testing to be done in this area. Majority consensus that having two tools to do the same thing in Houdini isn't a big problem.
- SIGGRAPH course
Deadline is May 28th to provided JT Nelson each contributors video clip. Title slides only need to exist at the start of the presentation. JT has already been provided a test by Jeff Budsberg which seems to play and compress well. Each contributor has 36 minutes to present. The current order and topics are as follows:
- Ken. Intro and NanoVDB
- Jeff L. NanoVDB adoption
- Jeff B. OpenVDB in production
- Rich. OpenVDB AX
- Dan. OpenVDB Merging/Writing fast parallel algorithms Ken to handle intro slide, Dan to handle end slide.
- Blind Data (PR1050)
- DynamicNodeManager Grain Size (PR1024)
Time.
- Next meeting
Next meeting is May 25th, 2021. 12pm-1pm EST (GMT-5). Nick to present some AX topics on the 1st June (or the next meeting after 25th)