fix: recompute axes grid bounds when switching files by soumyasavarn · Pull Request #2988 · f3d-app/f3d

soumyasavarn · 2026-03-30T08:30:33Z

Describe your changes

GridAxesConfigured was not being reset when the importer changed on file switch, so ConfigureGridAxesUsingCurrentActors() was never re-called with the new file's bounds. Added this->GridAxesConfigured = false; alongside the existing this->GridConfigured = false; in UpdateActors() in vtkext/private/module/vtkF3DRenderer.cxx.

Issue ticket number and link if any

Fixes #2963

Checklist for finalizing the PR

I have performed a self-review of my code
I have added tests for new features and bugfixes
I have added documentation for new features
If it is a modifying the libf3d API, I have updated bindings
If it is a modifying the .github/workflows/versions.json, I have updated docker_timestamp

AI Disclosure

I did not use AI to generate any of the content of that pull request
I used AI to generate code in that pull request, if yes please disclose which part of the code was generated and with which model.
...

Continuous integration

Please write a comment to run CI, eg: \ci fast.
See here for more info.

GridAxesConfigured was not being reset when the importer changed (i.e. on file switch), so ConfigureGridAxesUsingCurrentActors() was never re-called. Add GridAxesConfigured = false alongside the existing GridConfigured = false in UpdateActors(). Fixes f3d-app#2963

soumyasavarn · 2026-03-30T08:31:57Z

\ci fast

mwestphal

one small change

removed unnecessary comment Co-authored-by: Mathieu Westphal <mwestphal.f3d@protonmail.com>

Co-authored-by: Mathieu Westphal <mwestphal.f3d@protonmail.com>

mwestphal

the feature doesnt seem to work yet.

soumyasavarn · 2026-03-31T11:55:14Z

@mwestphal I tested with the "F3D-3.4.1-147-g30c7f045-Windows-x86_64-raytracing":

dragon.vtu alone with --axes-grid → correctly centered
cow.vtp + dragon.vtu, pressing Right to switch → axes grid disappears completely

My fix makes the grid reappear on switch, but the bounds show -58 to 100
instead of the expected -100 to 100. Could this be related to actor transforms
being different when dragon is loaded as the second file vs first file?
Or is there something else that needs resetting on file switch?

mwestphal · 2026-03-31T12:09:53Z

@mwestphal I tested with the "F3D-3.4.1-147-g30c7f045-Windows-x86_64-raytracing":
* dragon.vtu alone with --axes-grid → correctly centered

* cow.vtp + dragon.vtu, pressing Right to switch → axes grid disappears completely
My fix makes the grid reappear on switch, but the bounds show -58 to 100 instead of the expected -100 to 100. Could this be related to actor transforms being different when dragon is loaded as the second file vs first file? Or is there something else that needs resetting on file switch?

It looks like the bounds computation is not correct for some reason, so you need to dig into that.

mwestphal · 2026-03-31T12:10:21Z

@soumyasavarn Merge branch 'f3d-app:master' into fix/axes-grid-not-recomputed-on-fi…

Please do not merge with master, but use rebase instead, and only when needed.

mwestphal · 2026-05-03T15:55:06Z

Hi @soumyasavarn , do you need help moving forward ?

soumyasavarn · 2026-05-03T17:08:44Z

Hi @soumyasavarn , do you need help moving forward ?

Yes, a little guidance is needed. As of now only that one line fix is what I figured out is necessary "this->GridAxesConfigured = false;" in the updateActors(). Like, what amount of fix would be there as I am new to this codebase. Meanwhile, I will be unavailable for a week. If someone else wants to work on this you can let him.

Resetting GridAxesConfigured alone left coloring/visibility/mapper state from the previous file, so ComputeVisiblePropOrientedBounds reported a clipped bbox for the new file (dragon.vtu after cow.vtp showed Z bounds of -58..100 instead of -100..100). Reset all derived *Configured flags when the importer changes, request whole-extent before alg->Update so streaming readers like vtkXMLUnstructuredGridReader report full bounds rather than piece 0, and reset the missing flags in Initialize() for symmetry. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

deleting earlier baseline

mwestphal · 2026-05-04T05:50:45Z

Like, what amount of fix would be there as I am new to this codebase

I'm afraid this would require some investigation to figure out. Do your one line fix work as expected ?

Meanwhile, I will be unavailable for a week. If someone else wants to work on this you can let him.

No worries, it can wait.

The previous attempt called vtkStreamingDemandDrivenPipeline::SetUpdateExtentToWholeExtent on an instance with an int argument; that overload does not exist (it is a static method taking vtkInformation*) and broke the build. vtkAlgorithm::UpdateWholeExtent does the equivalent in one call and works for both extent- and piece-based pipelines. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

ComputeVisiblePropOrientedBounds iterates renderer Props whose visibility and mapper inputs depend on per-file-switch state, so it returned a clipped bbox after the cow.vtp -> dragon.vtu switch. Use vtkF3DMetaImporter::GetGeometryBoundingBox instead - the importer maintains the union of imported surface bounds and is authoritative regardless of prop visibility ordering. Transform the 8 corners through the env-up matrix to support non-axis-aligned up directions. Also revert the speculative UpdateWholeExtent call in ComputeVisiblePropOrientedBounds (that path is now only used by the floor grid, which was unaffected). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

soumyasavarn · 2026-05-04T10:16:38Z

Do your one line fix work as expected ?

Yes, without that the axes didn't appear.

github-actions · 2026-05-04T10:17:54Z

Style Checks CI failed:

diff --git a/vtkext/private/module/vtkF3DRenderer.cxx b/vtkext/private/module/vtkF3DRenderer.cxx
index a98fe9f..b334ca8 100644
--- a/vtkext/private/module/vtkF3DRenderer.cxx
+++ b/vtkext/private/module/vtkF3DRenderer.cxx
@@ -978,12 +978,10 @@ void vtkF3DRenderer::ConfigureGridAxesUsingCurrentActors()
     {
       double bnds[6];
       importerBbox.GetBounds(bnds);
-      const double corners[8][3] = {
-        { bnds[0], bnds[2], bnds[4] }, { bnds[1], bnds[2], bnds[4] },
-        { bnds[0], bnds[3], bnds[4] }, { bnds[1], bnds[3], bnds[4] },
-        { bnds[0], bnds[2], bnds[5] }, { bnds[1], bnds[2], bnds[5] },
-        { bnds[0], bnds[3], bnds[5] }, { bnds[1], bnds[3], bnds[5] }
-      };
+      const double corners[8][3] = { { bnds[0], bnds[2], bnds[4] }, { bnds[1], bnds[2], bnds[4] },
+        { bnds[0], bnds[3], bnds[4] }, { bnds[1], bnds[3], bnds[4] }, { bnds[0], bnds[2], bnds[5] },
+        { bnds[1], bnds[2], bnds[5] }, { bnds[0], bnds[3], bnds[5] },
+        { bnds[1], bnds[3], bnds[5] } };
       for (int i = 0; i < 8; ++i)
       {
         double p[4] = { corners[i][0], corners[i][1], corners[i][2], 1.0 };

mwestphal · 2026-05-04T10:43:33Z

HI @soumyasavarn

It looks lke you are using Generative AI for coding, please note in the MR description how it was used and which model in compliance with our AI policy: https://f3d.app/dev/AI_POLICY

mwestphal added the ci:fast label Mar 30, 2026

soumyasavarn added 3 commits March 30, 2026 08:55

style: fix formatting

3159616

test: add recording for TestInteractionAxesGridSwitchFile

1256a37

test: add baseline for TestInteractionAxesGridSwitchFile

8c57bee

mwestphal self-requested a review March 30, 2026 09:49

soumyasavarn marked this pull request as ready for review March 30, 2026 09:49

mwestphal reviewed Mar 30, 2026

View reviewed changes

Comment thread application/testing/tests.interaction.cmake Outdated

mwestphal requested changes Mar 30, 2026

View reviewed changes

fix: correct recording format and remove wrong baseline

9440172

soumyasavarn marked this pull request as draft March 30, 2026 11:42

mwestphal reviewed Mar 30, 2026

View reviewed changes

Comment thread vtkext/private/module/vtkF3DRenderer.cxx Outdated

mwestphal reviewed Mar 30, 2026

View reviewed changes

Comment thread application/testing/tests.interaction.cmake Outdated

soumyasavarn and others added 3 commits March 30, 2026 17:32

Update vtkext/private/module/vtkF3DRenderer.cxx

27960e7

removed unnecessary comment Co-authored-by: Mathieu Westphal <mwestphal.f3d@protonmail.com>

Update application/testing/tests.interaction.cmake

95741a4

Co-authored-by: Mathieu Westphal <mwestphal.f3d@protonmail.com>

test: add correct baseline for TestInteractionAxesGridSwitchFile

41a1ae8

soumyasavarn marked this pull request as ready for review March 30, 2026 12:43

soumyasavarn requested a review from mwestphal March 31, 2026 04:07

mwestphal reviewed Mar 31, 2026

View reviewed changes

Comment thread testing/baselines/TestInteractionAxesGridSwitchFile.png Outdated

mwestphal reviewed Mar 31, 2026

View reviewed changes

Comment thread application/testing/tests.interaction.cmake

mwestphal requested changes Mar 31, 2026

View reviewed changes

soumyasavarn marked this pull request as draft March 31, 2026 07:45

soumyasavarn force-pushed the fix/axes-grid-not-recomputed-on-file-switch branch from 4398688 to d3e2e36 Compare March 31, 2026 17:57

soumyasavarn force-pushed the fix/axes-grid-not-recomputed-on-file-switch branch from c96b6e4 to c24ded7 Compare May 4, 2026 04:50

Delete testing/baselines/TestInteractionAxesGridSwitchFile.png

0d77fe6

deleting earlier baseline

soumyasavarn and others added 2 commits May 4, 2026 08:42

mwestphal added the triage:stalled label May 28, 2026

Uh oh!

Conversation

soumyasavarn commented Mar 30, 2026 • edited by mwestphal Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Describe your changes

Issue ticket number and link if any

Checklist for finalizing the PR

AI Disclosure

Continuous integration

Uh oh!

soumyasavarn commented Mar 30, 2026

Uh oh!

Uh oh!

mwestphal left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

mwestphal left a comment

Choose a reason for hiding this comment

Uh oh!

soumyasavarn commented Mar 31, 2026

Uh oh!

mwestphal commented Mar 31, 2026

Uh oh!

mwestphal commented Mar 31, 2026

Uh oh!

mwestphal commented May 3, 2026

Uh oh!

soumyasavarn commented May 3, 2026

Uh oh!

mwestphal commented May 4, 2026

Uh oh!

soumyasavarn commented May 4, 2026

Uh oh!

github-actions Bot commented May 4, 2026

Uh oh!

mwestphal commented May 4, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

soumyasavarn commented Mar 30, 2026 •

edited by mwestphal

Loading