ComfyUI Node
Preview Gaussian Camera
A ComfyUI node in GaussianPack with 8 inputs and 1 output.
Preview Gaussian Camera
- extrinsics
- intrinsics
- image
◄ply_path—►
◄image_width1024►
◄image_height1024►
◄near_plane0.01►
◄backgroundblack►
◄force_backendauto►
CategoryGaussianPack
Inputs (8)
| Name | Type | Default | Description |
|---|---|---|---|
| ply_path | STRING | Path to a 3DGS PLY (HYWM2GaussianTrain.ply_path, SharpPredictGaussiansFromMetricDepth, MergeGaussians, or any external 3DGS tool). | |
| extrinsics | EXTRINSICS | [N, 4, 4] world-to-camera. Renders N frames. | |
| intrinsics | INTRINSICS | [N, 3, 3] or [3, 3] pixel-K (or normalized-K, auto-rescaled). Single [3,3] is broadcast to all N views. | |
| image_width | INT | 102432–8192 | — |
| image_height | INT | 102432–8192 | — |
| near_planeopt | FLOAT | 0.010.001–100 | Near-plane culling distance in world units. Gaussians with depth < near_plane are dropped. |
| backgroundopt | COMBO | black | Background color where transmittance survives all the gaussians. |
| force_backendopt | COMBO | auto | auto = try gsplat (fast CUDA); fall back to pure-torch on ImportError or non-CUDA device. gsplat = error if not available (use to verify fast path). torch = always use the pure-PyTorch fallback (slower; useful for AMD/Mac/CPU debug). |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| image | IMAGE | — |