Present
- Michael Min
- Nick Porcino
- Joseph Goldstone
- Michael B Johnson
- David Stump
- Alexander Schwank
- Sam Richards
- Kevin
- JT Nelson
- Patrick Palmer
- OT
- Sergio Rojas
- Quentin Birrer
Discussion
Joseph Goldstone
Should there be a fallback for a pinhole model for simple use cases?
usd
- modularity
- well engineered
- established across interested users
questionnaire
- assess sophistication at facilities
- assess rigor vs seat-of-the-pants
- identify precision needs datum by datum
- can camera and lens information be teased apart?
- are we stuck with grids?
- working practice should guide metadata collection
- this is an opportunity to be precise in a shared manner
- too much for OpenEXR
- usd should be the place a common model is canonically expressed
- this effort is user-pull, not vendor-push therefore serves the community
- the original camera parameterization equations we use came from JPL and the Mars Rovers
- there needs to be a procedures manual on characterization and recording the data
- think spectrally, act colormetrically
David Stump
- it's not critical to get it correct first time
- it should evolve as the metadata is used and adopted
- an adoption bell-curve phenomenon, time is ripe to do this, finally
- producers won’t care about the data, they will care about cost-savings
- industry resists efforts with self-interest. Absence of profit motive makes this effort compelling
- coefficients for third order equations characterize lenses, easy to evaluate
- has json data for lenses from Cooke, can supply these for usd camera wg to use
Sam Richards
- usd could be used for overlays with take to take variations
- Cover email to be sent to heads of match move by TC members:
https://docs.google.com/document/d/1Hnmr8kIiT6VzR0HEVlsO_cK-I15K2Xu4xWVqjLUL0c8/edit?usp=sharing
- Video for commandante:
https://drive.google.com/file/d/1ERDD5w5vxn3XGFkP00GgOriio9291pld/view
- grid tests should be used to wedge the numeric characterization
- Joseph will send Sam the Cooke Part Three document to share with this group
Michael Min
- accepted workflow is to matchmove approximate data
- trying to break this mold
- a cost savings would accrue if a shared camera model started to supplant match moving
- can we identify a base model in cooperation with camera vendors
Michael B Johnson
- compositional architecture of USD makes it appropriate for scalable data
- layering of data allows variation, levels of detail, and external referencing
- USD is descriptive, not evaluative
The camera data is contained, but there’s no functionally-derived data elsewhere
- can UsdShade be used to make an evaluable graph of camera parameters?
