GPv3: Merge by distance #113917

New Issue

Pratik Borhade · 2023-10-19T07:40:29+02:00

Pratik Borhade commented

2023-10-19 07:40:29 +02:00

Merge from selected/all points based on threshold value

Define operator GREASE_PENCIL_OT_stroke_merge_by_distance in grease_pencil_edit.cc
Create operator property threshold (int)
Iterate through all the drawings at current scene frame (retrieve_editable_drawings)
- Get geometry of this drawing: bke::CurvesGeometry &curves = drawing.strokes_for_write();
- Define an array merge_indices with the size of the points as well as a counter for the number of points to be merged for later.
- Loop though each individual curve using threading::parallel_for(curves.curves_range(), ...
  - Get a selection mask of the selected points in the curve (see for example ed::greasepencil::smooth_curve_attribute)
  - Get the selection ranges using IndexMask::to_ranges()
  - For every selection range:
    - Create an array distances that computes the accumulated distance for every point to the first point (e.g. the fist point has a distance of 0, every next point has a distances of prev_distance + distance to previous point)
    - Build a KDTree_1d for the distances (BLI_kdtree_1d_new, BLI_kdtree_1d_insert, BLI_kdtree_1d_balance)
    - Use BLI_kdtree_1d_calc_duplicates_fast to find the points that need to be merged. The return value is the number of points that need to be merged, e.g. remove. Make sure to accumulate this count into the counter declared above so that we can calculate the size of the final CurvesGeometry.
    - Write the resulting indices from BLI_kdtree_1d_calc_duplicates_fast into merge_indices. Make sure to use the correct offsets.
- Create a new CurvesGeometry with the same number of curves and calculate the number of points (remaining after merges).
- Calculate the new offsets using the merge_indicies array (see logic in point_merge_by_distance.cc) as well as a OffsetIndices<int> map_offsets.
- Transfer all the point attributes:
  - Iterate over all source attributes and use bke::attribute_math::DefaultMixer<T> mixer to mix the values for all the destination points. Use the map_offsets to find the source points to mix the attributes from (There might only be one point if it just need to be copied, that's fine).
- Copy all the curve attributes

Merge from selected/all points based on threshold value - Define operator `GREASE_PENCIL_OT_stroke_merge_by_distance` in `grease_pencil_edit.cc` - Create operator property `threshold` (int) - Iterate through all the drawings at current scene frame (`retrieve_editable_drawings`) - Get geometry of this drawing: `bke::CurvesGeometry &curves = drawing.strokes_for_write();` - Define an array `merge_indices` with the size of the points as well as a counter for the number of points to be merged for later. - Loop though each individual curve using `threading::parallel_for(curves.curves_range(), ...` - Get a selection mask of the selected points in the curve (see for example `ed::greasepencil::smooth_curve_attribute`) - Get the selection ranges using `IndexMask::to_ranges()` - For every selection range: - Create an array `distances` that computes the accumulated distance for every point to the first point (e.g. the fist point has a distance of 0, every next point has a distances of prev_distance + distance to previous point) - Build a `KDTree_1d` for the distances (`BLI_kdtree_1d_new`, `BLI_kdtree_1d_insert`, `BLI_kdtree_1d_balance`) - Use `BLI_kdtree_1d_calc_duplicates_fast` to find the points that need to be merged. The return value is the number of points that need to be merged, e.g. remove. Make sure to accumulate this count into the counter declared above so that we can calculate the size of the final `CurvesGeometry`. - Write the resulting indices from `BLI_kdtree_1d_calc_duplicates_fast` into `merge_indices`. Make sure to use the correct offsets. - Create a new `CurvesGeometry` with the same number of curves and calculate the number of points (remaining after merges). - Calculate the new offsets using the `merge_indicies` array (see logic in `point_merge_by_distance.cc`) as well as a `OffsetIndices<int> map_offsets`. - Transfer all the point attributes: - Iterate over all source attributes and use `bke::attribute_math::DefaultMixer<T> mixer` to mix the values for all the destination points. Use the `map_offsets` to find the source points to mix the attributes from (There might only be one point if it just need to be copied, that's fine). - Copy all the curve attributes

Pratik Borhade added the

labels 2023-10-19 07:40:44 +02:00

Pratik Borhade added this to the Grease Pencil project 2023-10-19 07:40:59 +02:00

Hans Goudey commented

2023-10-19 09:52:28 +02:00

I think the actual merging is fairly tricky here, at least doing it in a performant way.

The same behavior should be used as a geometry node (possibly the existing "Merge by Distance" node), that's something to keep in mind.

Rather than having an "unselected" toggle, IMO it would be simpler and clearer to just require selecting every point for that case. All it would require is first pressing "A".

I think the actual merging is fairly tricky here, at least doing it in a performant way. The same behavior should be used as a geometry node (possibly the existing "Merge by Distance" node), that's something to keep in mind. Rather than having an "unselected" toggle, IMO it would be simpler and clearer to just require selecting every point for that case. All it would require is first pressing "A".

👍 1

Falk David commented

2023-10-19 14:27:54 +02:00

After a discussion with @HooglyBoogly I edited the task to reflect the changes.

Falk David referenced this issue

2023-10-25 18:43:46 +02:00

GPv3: Community Tasks: Overview #114158

Prikshit singh commented

2023-12-18 18:13:10 +01:00

While solving this issue I got stuck at Create an array distances that computes the accumulated distance for every point to the first point (e.g. the fist point has a distance of 0, every next point has a distances of prev_distance + distance to previous point) How will I get the points and calculate distance for every selection range , could you give me an overview of how should I approach this and steps after this. Thank you.

While solving this issue I got stuck at ```Create an array distances that computes the accumulated distance for every point to the first point (e.g. the fist point has a distance of 0, every next point has a distances of prev_distance + distance to previous point)``` How will I get the points and calculate distance for every selection range , could you give me an overview of how should I approach this and steps after this. Thank you.

Prikshit singh referenced this issue

2024-01-09 14:27:10 +01:00

GPv3: Sculpt Mode #116899

Chao Li commented

2024-04-06 17:48:07 +02:00

Hello. Is this task currently being worked on?