4.1 KiB
Minutes from 45th OpenVDB TSC meeting, April 9th, 2020, (GMT)
Attendees: Nick A., Jeff L., Ken M.
Additional Attendees: Johannes Meng (Intel), JT Nelson (Blender), Peter Cheng (DW), Andre Pradhana (DW)
Regrets: Peter C., Dan B.
Agenda:
-
Confirm quorum
-
Secretary
-
OpenVDB AX recap
-
Forum/Github Issues update
-
Plan for Improved Morphology
-
Google Summer of Code
-
CI Updated
-
VDB Activate/Activate SDF
-
OVDB-136
-
OpenVDB & USD
-
Next meeting
-
Quorum was confirmed.
-
Secretary was Nick Avramoussis
-
OpenVDB AX recap
Discussion ongoing between Nick and Jeff as various AX features are nailed down. A number of proposed changes have been approved - bitwise operations on integer vectors, restricting comparison operation on container types and correct implementation of fmod. Two outstanding requests are the beahviour of matrix and scalar operations and container (vector) extension. Jeff's preferred behaviour for the prior is to treat scalar promotion as an identity matrix with its identity components set to the corresponding scalar. For the latter, VEX currently extends vectors with a value of one to better match the behaviour of various applications against matrices of otherwise incompatible sizes and to better correspond to a default value of colour (alpha) or homogeneous coordiante (w). The alternative would be to explicitly error on this behaviour. AX currently already errors on assignment of containers of mismatching sizes.
- Nick proposed that every TSC meeting additionally begin with a quick recap of newly opened forum posts or github issues. Unanimous consensus. Two topics for discussion.
4a) OpenVDB Documentation
New forum post requesting update on OpenVDB documentation. Agreement that OpenVDB technical documentation needs a revamp and could benefit from extension. This is a significant task with no obvious starting point. Andre mentioned that the unit tests have served as a good place to observe various intended usage of VDB methods. Using the unit tests would provide a base set of examples to start writing documentation. Nick, concern that as the unit tests have not necessarily been designed for this that some of the logic in there could be "inaccurate" in terms of an optimal implementation solution. Additional care to be taken that the unit tests are first and foremost that, with clear documentation of the intended example. Nick to create a ticket with a task to identity some of the larger tools missing documentation and corresponding unit tests which could be used as a first doc attempt.
4b) Hessian computation of VDB grids
Github issue asking if OpenVDB supports the computation of the Hessian matrix from a scalar Grid. Currently unsupported. Ken, should be fairly trivial to implement. However, as the result per value is a matrix, there is no native way to store this information as VDB does not register matrix grids by default. Registering these grid types would be one option - however compile and runtime variants of "strided" grids (where each voxel/tile value contains an array of scalar values of a fixed size) were discussed as alternative solutions which could add more flexibility. Some potential confusion when inferring values on a grid with an "intrinsic" type i.e. Vec3f vs float[3] collocated vs staggered interpretation. General agreement that runtime variants would be interesting to pursue. Nick to create a ticket and reply to issue.
- Plan for Improved Morphology
Ken and Nick to catch up at a later date.
- Google Summer of Code
Two mentors are needed. Dan has already volunteered to be one. Nick has volunteered to be the second, with an agreement with Ken to share responsibility if required.
- CI Updated
General notice that all VDB builds should now be passing on new PRs and that PRs should not be merged if any builds fail. Dan has a fix in the works to re-enable the Houdini 18 Debug builds.
- VDB Activate/Activate SDF
Two open PRs (PR692, PR695) could do with additional review from Ken and Nick.
- OVDB-136 (Point Merging/Moving)
Nick needs to review PRs PR652 and PR631.
- OpenVDB & USD
Additional discussion to be had next meeting.
- Next meeting
April 16th 2020. 2pm-3pm EDT (GMT-4).