Hello everyone, houdini’s sphere bound. There seems to be an axial misalignment problem.

On a perfect sphere, any rotation would be correct. In this case you're getting the rotation that results in the smallest possible bounding box, taking the subdivisions into account. You can uncheck Oriented Bounding Box if you'd rather have it straight.

Hi,

This is by design.

Oriented Bounding Box (OBB) aligns with the eigenvectors derived from the input geometry's covariance matrix.

From this matrix, it determines the principal directions along which the points spread out the most - these directions are identified by calculating the eigenvectors of the covariance matrix. These eigenvectors do not necessarily align with the world axes because they are intimately related to the geometry's own distribution of points, not the global coordinates.

This approach doesn't necessarily produce the smallest box, but it ensures that the box is aligned with the geometry's natural orientation, thereby efficiently encapsulating the actual shape.

On the other hand, the Minimum Volume Box (MVB) takes a different approach. It seeks to encapsulate the geometry within a box of the absolute minimum volume, regardless of the geometry's principal directions. This process involves a more complex calculation, often exploring various orientations of the box to determine the one that encompasses the geometry with the smallest volume.

Hope this helps.

非常
On a perfect sphere, any rotation would be correct. In this case you're getting the rotation that results in the smallest possible bounding box, taking the subdivisions into account. You can uncheck Oriented Bounding Box if you'd rather have it straight.

animatrix_
Hi,

This is by design.

Oriented Bounding Box (OBB) aligns with the eigenvectors derived from the input geometry's covariance matrix.

From this matrix, it determines the principal directions along which the points spread out the most - these directions are identified by calculating the eigenvectors of the covariance matrix. These eigenvectors do not necessarily align with the world axes because they are intimately related to the geometry's own distribution of points, not the global coordinates.

This approach doesn't necessarily produce the smallest box, but it ensures that the box is aligned with the geometry's natural orientation, thereby efficiently encapsulating the actual shape.

On the other hand, the Minimum Volume Box (MVB) takes a different approach. It seeks to encapsulate the geometry within a box of the absolute minimum volume, regardless of the geometry's principal directions. This process involves a more complex calculation, often exploring various orientations of the box to determine the one that encompasses the geometry with the smallest volume.

Hope this helps.