Cycles: Add render tests for texture nodes #16
No reviewers
Labels
No Label
No Milestone
No Assignees
2 Participants
Notifications
Due Date
No due date set.
Dependencies
No dependencies set.
Reference: blender/blender-test-data#16
Loading…
Reference in New Issue
Block a user
No description provided.
Delete Branch "add-texture-tests"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Adds render tests for most texture nodes in Blender, testing multiple
configurations per nodes, and including some previously problematic
test causes to ensure they don't appear in the future.
Some nodes (E.g. IES) aren't tested as they're better suited for
another category.
Ref blender/blender#123012
This PR was originally !6
However to test this PR on the build bot (a step I plan to do before committing), I need the branch to be in
blender/blender-test-data
, so I had to create a new PR.@Alaska I am a bit confused. For a new branch on
blender-tests-data
you can just commit directly to a branch, and create PR againstblender.git
. This PR is againstmain
, so if we land it kind of defeats the purpose of having it in a branch first?As for adding texture tests - it is something i am all up for, so it is more about figuring out logistics of how to get them into the repos.
My understanding was this:
blender/blender-tests-data
notYOUR_NAME/blender-tests-data
with the new testsblender/blender
that references the tests (That's here: blender/blender#124574)I opened this as a PR with reviewers because it's a larger collection of new tests, so I'd like them to be reviewed.
Your understanding is fully correct!
I think I understand now: i've missed the fact that the source branch is in the
blender/blender-tests-data
, and not in the fork. Makes it possible to easily test locally and poke buildbot in blender/blender#124574.Form quick looking into the PR it seems all good! I've mainly:
I've also poked CI/CD on the PR against
blender.git
to ensure all other platforms are happy.My main concern would be that I'm testing too many configurations of some things.
For example I test a lot of configurations of the wave texture, many of which probably aren't needed.
Another thing that I talked about a while ago but should probably re-mention here:
I created most of the tests in Blender 3.6, then created a few additional tests in Blender 4.1 and 4.2 for features/nodes that weren't in 3.6. Some of the nodes in 3.6 no longer exist. For example the Musgrave texture, but the versioning code is tested with tests like
musgrave_multifractal_36
(A file created and saved in 3.6 using the musgrave texture).I verified locally that the 3.6 specific test files (like musgrave) did get correctly converted to 4.2.
This perfectly follows the blackbox testing method. Perhaps some of the permutations you test can be avoided due to them falling into the same code path, but it is not an assumption you can make from a blackbox perspective.
Checkout
From your project repository, check out a new branch and test the changes.