2.8 KiB
Minutes from 47th OpenVDB TSC meeting, April 30th, 2020, (EDT)
Attendees: Nick A., Jeff L., Ken M., Dan B.
Additional Attendees: Johannes Meng (Intel), JT Nelson (Blender), Andre Pradhana (DWA), Bruce Chernia (Intel)
Regrets: Peter C.
Agenda:
-
Confirm quorum
-
Secretary
-
Github protected branches
-
VDB Activate PR #692
-
Multithreaded File Compression
-
AX Update
-
Siggraph
-
Forum Posts
-
USD
-
Next Meeting
-
Quorum was confirmed.
-
Secretary was Jeff Lait
-
Github protected branches
All branches are protected. So as soon as you create a branch it can't be deleted. Only master should not be deleted.
Possible v3.3 and 6.2 branches should also be protected.
This showed up because we've started making branches on the master branch to hold changes. Unanimous consent to propose new rules for protected branches.
- VDB Activate PR #692
Awaiting Ken's feedback. If he has not replied we can push it through tomorrow.
- Multithreaded File Compression
Require big enough data sets to break into blocks. The grid's are working at too fine of a layer to benefit. We'd need to add a paging support to be able to coalesce the operations
- AX Update
Most feed back has been addressed, other than the matrix issues.
A feature branch will be updated showing how it can be integrated on Nick's fork. Matrix is still an outstanding issue, specifically scalar matrix promotion.
CMake is still blocking this. Next week ideally will have demonstration.
Easiset way to build AX is to add LLVM to VDB as an option. This would allow VDB core to experiment with LLVM. However, the concern is that LLVM is hard to support so we should avoid locking down one version.
- Siggraph
We have a BOF that will have to be virtual.
- Forum Posts
Managed to defeat the Google UI and approve 15 people for the forum.
An issue with points docs was pointed out.
Our Vec3D code somewhere hard codes down to floats that needs to be fixed.
Question about associating grids together. 100 grids with same topology need duplicate accessors. Blind data could work. Or a way to get a linear node number into an offset table. We should reply and suggest solutions. The origin info in the node can be re-used as a blind data. This ties back to the float[N] grid type discussed previously. Dan will attempt to form a response.
- USD
Some schemas exist already. They don't seem very VDB specific. Is there a problem we need to solve? Does VDB have to be more than an opaque volume format. As a render format, it doesn't need any over-operations. But what about USD as an interchange format? USD is a metadata format, so VDB can live within that without needing to be exposed at the USD level. What are images in USD? They seem to be references to file on disk.
- Next Meeting
May 7th 2020. 2pm-3pm EDT (GMT-4).