ComfyUI Node

Cloud Points Info

A one-glance stats readout for your point cloud

By chri002·Created about a year ago·Updated about a year ago· 15
Cloud Points Info
  • points
  • STRING

Cloud Points Info is the debug print statement of this pack. Wire any Points3D cloud into it and it spits back a short text summary - how many points, and the min/max range along each axis. That's the entire feature, and honestly it's all it needs to be.

This is the node you add to the graph when something looks wrong and you want numbers instead of guesses. Cloud unexpectedly huge after an interpolation step? The count tells you at a glance. Cloud rendering as a flat smear? The Z range will tell you whether depth survived the trip through your transforms. When you're stringing Image To Points → Transform → renderer together and the output isn't what you expected, dropping this node on the wire is the fastest way to see whether the problem is the geometry or the render. Think of it as the print(points.shape) you'd write if you were debugging the code by hand.

How it works

The node squeezes the cloud, counts the points, and measures min/max on X, Y and Z. There's not much more to it - the work is trivial, which is why it's basically instant even on million-point clouds.

Output: STRING - a formatted summary. Since it's plain text, you can feed it into any text-display node (or a Show Text / preview node) and read the result. The string looks roughly like:

Points : 1048576
X : [0.0, 1023.0]
Y : [0.0, 1023.0]
X : [−6.0, −0.2]

Install

Same as the rest of the pack. ComfyUI Manager → "ComfyUI_depthMapOperation", or:

cd ComfyUI/custom_nodes
git clone https://github.com/chri002/ComfyUI_depthMapOperation

Restart. No model downloads, no keys; deps (torch, numpy, opencv-python, scipy, pandas) are already in a stock ComfyUI.

Where people get burned

Two little gotchas, both more annoying than broken. First, the third range line is printed with an "X" label instead of "Z" - a typo in the source. The numbers under it are the Z range, not a duplicate of X, so don't let the label throw you off. Second, the pack is batch-size-1 everywhere, and this node follows suit: a batched cloud will confuse the shape math. Feed it a single cloud and the readout is accurate. And if you're used to this pack's quirks, no surprises here - small one-maintainer utility, does one thing, does it fine.

CategorydepthMapOperation

Inputs (1)

NameTypeDefaultDescription
pointsPoints3D

Outputs (1)

NameTypeDescription
STRINGSTRING