Display points #27

New Issue

Vectornaut · 2024-12-04T20:10:07Z

Vectornaut commented

2024-12-04 20:10:07 +00:00

Show points in the Display view.

Styling

Emphasize that a point has no part. For example:

No partial occlusion, even when the point representation (presumably disk or sphere) is only partially shaded by a surface

A point coincident with a codimension 1+ entity is never occluded by that entity.
Maybe a point that's constrained to be on a surface should be flattened onto the surface? (Likely this would be a later experimental refinement of the visual representation.)

Labeling

Label is never occluded by the surface the point is on (which is a corollary of (5) and (2)).
Label always has the same apparent occlusion as the point.
Label stays in the same screen orientation and position, relative to the point, under most or all rotations.

Labels are always "upright" for reading in the display, unless explicitly customized on a per-point basis or generically. (Such angled label text can be a later enhancement.)
Maybe it would be okay for the label's screen orientation to roll when the camera rolls (respecting previous bullet).
Ultimately the apparent relationship of label to point should be customizable (e.g. by dragging the label, but perhaps in other ways, such as a selection of placement algorithms) on a per-point basis, falling back to a (potentially customizable) general default.

Show points in the `Display` view. ### Styling 1. Emphasize that a point has no part. For example: - No partial occlusion, even when the point representation (presumably disk or sphere) is only partially shaded by a surface 2. A point coincident with a codimension 1+ entity is never occluded by that entity. 3. Maybe a point that's constrained to be on a surface should be flattened onto the surface? (Likely this would be a later experimental refinement of the visual representation.) ### Labeling 4. Label is never occluded by the surface the point is on (which is a corollary of (5) and (2)). 5. Label always has the same apparent occlusion as the point. 6. Label stays in the same screen orientation and position, relative to the point, under most or all rotations. - Labels are always "upright" for reading in the display, unless explicitly customized on a per-point basis or generically. (Such angled label text can be a later enhancement.) - Maybe it would be okay for the label's screen orientation to roll when the camera rolls (respecting previous bullet). - Ultimately the apparent relationship of label to point should be customizable (e.g. by dragging the label, but perhaps in other ways, such as a selection of placement algorithms) on a per-point basis, falling back to a (potentially customizable) general default.

Vectornaut added reference main

2024-12-04 20:10:21 +00:00

Vectornaut removed reference main

2024-12-04 20:10:24 +00:00

Vectornaut added the

enhancement

label 2024-12-04 20:10:42 +00:00

glen commented

2024-12-04 21:25:16 +00:00

Note that a first PR would presumably only implement the core of the features listed above; its merge should be accompanied by filing issues with the remaining desired features and resolving this overall one.

glen commented

2024-12-04 21:56:52 +00:00

Are points already fully supported in the Assembly and Engine? If not, it would seem that issues for such support should be filed now, and at least the Assembly one should be a prerequisite of this one.

glen commented

2024-12-04 22:04:15 +00:00

Oops 2 is not quite right. A point on the occluded hemisphere of a sphere is occluded (by the "front" hemisphere of the sphere) just the way that occluded hemisphere is. So the language of 2 needs amendment to capture this. Maybe:

A point coincident with a codim 1+ entity always has the same occlusion as the corresponding Euclidean point on that entity, never more from the immediate neighborhood of the entity itself (but it could be occluded by some disparate region of the entity in which it lies).

Oops 2 is not quite right. A point on the occluded hemisphere of a sphere is occluded (by the "front" hemisphere of the sphere) just the way that occluded hemisphere is. So the language of 2 needs amendment to capture this. Maybe: 2. A point coincident with a codim 1+ entity always has the same occlusion as the corresponding Euclidean point on that entity, never more from the immediate neighborhood of the entity itself (but it could be occluded by some disparate region of the entity in which it lies).

glen commented

2024-12-04 22:05:45 +00:00

Same caveats with 4 (feel free to just remove that proviso if you like, as it is a corollary of 5 and 2 anyway).

Sign in to join this conversation.