Add option to use boundingBox (instead of boundingSphere) when calculating arcRotateCam lowerRadius. Add orthographic framing #17329
Conversation
|
Please make sure to label your PR with "bug", "new feature" or "breaking change" label(s). |
|
Snapshot stored with reference name: Test environment: To test a playground add it to the URL, for example: https://snapshots-cvgtc2eugrd3cgfd.z01.azurefd.net/refs/pull/17329/merge/index.html#WGZLGJ#4600 Links to test babylon tools with this snapshot: https://playground.babylonjs.com/?snapshot=refs/pull/17329/merge To test the snapshot in the playground with a playground ID add it after the snapshot query string: https://playground.babylonjs.com/?snapshot=refs/pull/17329/merge#BCU1XR#0 |
|
Devhost visualization test reporter: |
|
WebGL2 visualization test reporter: |
|
Visualization tests for WebGPU |
|
I work a lot with BabylonJS and I love getting my hands dirty with the code. For example, I’ve made a “light” version of the OBJExporter that can shrink mesh size by up to 60%, by skipping details like normals, materials, etc. Let me know if you’re interested! |
| const depth = (maximumWorld.z - minimumWorld.z) * 0.5; | ||
| const aspectRatio = this.getScene().getEngine().getAspectRatio(this); | ||
|
|
||
| if (this.mode === Camera.ORTHOGRAPHIC_CAMERA) { |
There was a problem hiding this comment.
@PietroFurlan in your PR the logic here is slightly different - can you confirm which math is correct? do we want the same ortho formula regardless of sphere/box, or should they diffeR?
There was a problem hiding this comment.
@PietroFurlan closign PR for now as it has been open a while, lets discuss the correct approach via these comments and once aligned i can re-open the PR
There was a problem hiding this comment.
If I understood correctly: with an orthographic camera there’s no concept of radius or FOV or else, so the sphere/box mode shouldn’t affect the frustum math. Behavior should be identical in ortho mode. In your PR the ortho option existed only for the box mode; by changing the math only there, I introduced the inconsistency—my bad.
In ortho mode, the math is the same for box and sphere, and should be:
this.orthoTop = height * radiusScale;
this.orthoBottom = -this.orthoTop;
this.orthoRight = this.orthoTop * aspectRatio;
this.orthoLeft = -this.orthoRight;
| this.orthoRight = height * aspectRatio * radiusScale; | ||
| this.orthoLeft = -this.orthoRight * radiusScale; | ||
| this.orthoTop = height; | ||
| this.orthoBottom = -this.orthoTop; |
There was a problem hiding this comment.
there is an error on this formula when passing a radiusScale != 1.
In my last commit you find the correct one:
| this.orthoRight = height * aspectRatio * radiusScale; | |
| this.orthoLeft = -this.orthoRight * radiusScale; | |
| this.orthoTop = height; | |
| this.orthoBottom = -this.orthoTop; | |
| this.orthoTop = height * radiusScale; | |
| this.orthoBottom = -this.orthoTop; | |
| this.orthoRight = this.orthoTop * aspectRatio; | |
| this.orthoLeft = -this.orthoRight; | |
| const depth = (maximumWorld.z - minimumWorld.z) * 0.5; | ||
| const aspectRatio = this.getScene().getEngine().getAspectRatio(this); | ||
|
|
||
| if (this.mode === Camera.ORTHOGRAPHIC_CAMERA) { |
There was a problem hiding this comment.
If I understood correctly: with an orthographic camera there’s no concept of radius or FOV or else, so the sphere/box mode shouldn’t affect the frustum math. Behavior should be identical in ortho mode. In your PR the ortho option existed only for the box mode; by changing the math only there, I introduced the inconsistency—my bad.
In ortho mode, the math is the same for box and sphere, and should be:
this.orthoTop = height * radiusScale;
this.orthoBottom = -this.orthoTop;
this.orthoRight = this.orthoTop * aspectRatio;
this.orthoLeft = -this.orthoRight;5 participants
This is a refactor of the PR proposed here: #17292
Adds new optional param 'mode' to the framingBehavior zoomOnBoundingInfo. If specified, calls into the camera's new _calculateLowerRadiusFromModelBoundingInfo function which respects the passed in mode and adds orthographic framing
If no mode is sent, default to old behavior (using cameras _calculateLowerRadiusFromModelBoundingSphere fn, and not adding orthographic framing)
This maintains backwards compatibility of old behavior while still offering option to use new behavior