Comfy.ICU
ExploreDocsPricing
Support

ComfyUI Extension: ComfyUI_depthMapOperation

Authored by chri002

Created

Updated

7 stars

A simple set of nodes to generate a point cloud from an image and its depth map, perform transformations and some basic operations.

Custom Nodes (0)

    README

    ComfyUI_depthMapOperation

    A simple set of nodes to generate a point cloud from an image and its depth map, perform transformations and some basic operations.

    Here an example of what can be done (not the gif, but the various rotations)

    <div class="grid" align="center" markdown> <img src=".//assets/start.jpg" width=40%> <img src=".//assets/gif.gif" width=40%> </div>

    Installation

    On the console in the custom_nodes filder execute:

    git clone https://github.com/chri002/ComfyUI_depthMapOperation

    Requirements

    these nodes require in order to function:

    • torch
    • numpy
    • opencv-python
    • scipy
    • pandas

    Workflow

    workflow

    Nodes

    Image To Points (Torch)

    Image To Points

    Description

    GPU-accelerated version using PyTorch tensors. Maintains gradient flow and supports automatic device placement.

    Input Parameters:

    • image: Input RGB/RGBA image
    • depth_image: Depth map image
    • depth: Z-axis scaling factor (1-1024)
    • quality: Downsampling quality (1=1 point:1 pixel , 16=16 interpolate points every 2 pixel)

    Output:

    • Points3D: XYZ coordinates + RGB colors

    Transform Points

    Transform Points

    Description

    Applies 3D transformations to point clouds (rotation, translation, scaling).

    Input Parameters:

    • points: Input point cloud (Points3D)
    • rot_x/y/z: Euler angles in degrees
    • trl_x/y/z: Translation offsets
    • scale_x/y/z: Axis-specific scaling factors

    Output:

    • Points3D: Transformed point cloud (XYZ coordinates + RGB colors)

    Points To Image (Orthographic)

    Points To Image (Orthographic)

    Description

    Renders 3D points to 2D image using orthographic projection.

    Input Parameters:

    • images: Template for output dimensions
    • points: Point cloud to render (Points3D)
    • color: Enable RGB coloring

    Output:

    • IMAGE: Rendered grayscale/RGB image

    Points To Image (Projection)

    Points To Image (Projection)

    Description

    Perspective projection renderer with customizable FOV.

    Input Parameters:

    • images: Template for output dimensions
    • points: Point cloud to render (Points3D)
    • color: Enable RGB coloring
    • fov: Field of View in degrees (1-2000)

    Output:

    • IMAGE: Rendered grayscale/RGB image

    Cube Limit

    Cube Limit

    Description

    Filters points within relative cube dimensions (0-100% of original bounds).

    Input Parameters:

    • points: Point cloud to render (Points3D)
    • 6 axis range parameters (x_min-x_max, etc.)

    Output:

    • Points3D: Subset of points within cube (XYZ coordinates + RGB colors)

    Clean Points (KDTree)

    EClean Points (KDTree)

    Description

    Removes outliers using KDTree neighborhood analysis.

    Parameters:

    • points: Point cloud to render (Points3D)
    • k: Minimum neighbors required
    • m: Max neighbor distance threshold

    Output:

    • Points3D: Cleaned point cloud (XYZ coordinates + RGB colors)

    Interpolate Points (KDTree)

    Interpolate Points (KDTree)

    Description

    Generates new points through neighborhood-based interpolation using KDTree. Enhances point cloud density in sparse regions by creating intermediate points between existing neighbors.

    Input Parameters:

    • points: Input 3D point cloud
    • value: (0-1) Blend ratio for new points (0=keep original, 1=full interpolation)
    • n: Number of nearest neighbors to consider (0-32)

    Output:

    • Points3D: Point cloud with added interpolated points (XYZ coordinates + RGB colors)

    Export To PLY

    Export To PLY

    Description

    Exports point cloud to PLY format (ASCII/binary).

    Input Parameters:

    • points: Point cloud to render (Points3D)
    • multiple_files: Split XYZ/RGB data
    • format_out: File encoding format

    Import PLY

    Import PLY

    Description

    Import PLY point cloud files into compatible Point3D format. (Sperimental)

    Input Parameter:

    • .ply file selection

    Output:

    • Points3D: Loaded point cloud data (XYZ coordinates + RGB colors)

    Cloud Points Info

    Cloud Points Info

    Description

    Displays point cloud statistics and coordinate ranges.

    Output:

    • STRING: Formatted summary text

    DEMO FUNCTIONS

    <a target="_blank" href="https://colab.research.google.com/github/chri002/ComfyUI_depthMapOperation/blob/main/test.ipynb"> <img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Test In Colab" height="30px"/> </a>

    TODO

    • Fix artifacts with some extreme values