ComfyUI Node
Reconstruct Poisson (backend)
A ComfyUI node in geompack/reconstruction with 5 inputs and 2 outputs.
Reconstruct Poisson (backend)
- points
- reconstructed_mesh
- info
◄poisson_depth8►
◄poisson_scale1.1►
◄estimate_normalstrue►
◄normal_radius0.10►
Categorygeompack/reconstruction
Inputs (5)
| Name | Type | Default | Description |
|---|---|---|---|
| points | TRIMESH | — | |
| poisson_depth | INT | 81–12 | Octree depth for the Poisson solver. Higher values capture finer detail but use exponentially more memory and time. 6=coarse, 8=balanced, 10+=high detail. |
| poisson_scale | FLOAT | 1.11–2 | Scale factor for the reconstruction grid relative to the bounding box. Values >1.0 add padding to avoid boundary artifacts. 1.1 is usually sufficient. |
| estimate_normals | COMBO | true | Re-estimate point normals using k-nearest neighbors. Poisson reconstruction requires oriented normals — enable this if the input has no normals or unreliable normals. |
| normal_radius | FLOAT | 0.100.001–10 | Search radius for normal estimation via k-nearest neighbors. Should be 2-3x the average point spacing. Only used when estimate_normals is true. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| reconstructed_mesh | TRIMESH | — |
| info | STRING | — |