array1 t

A transpose is one of those operations you never think about - until a node hands you data in the wrong orientation and nothing downstream lines up. array1 t is the transpose node for this 3D pack: give it a LIST, get back its transpose. If you've ever had a camera array that's columns-where-you-wanted-rows, this is the fix.

The single input is array_input (a LIST) and the single output is array_output (a LIST). Under the hood it converts to a numpy array and transposes - but the 1D behavior is worth knowing because it's a mild surprise. A 1D list like [1, 2, 3] gets converted to a column: [[1], [2], [3]], i.e. an N×1 2D list. A 2D list gets a proper numpy.transpose, swapping rows and columns: a 2×3 becomes a 3×2. So this node pushes everything toward 2D - which, for this pack's camera and batch arrays, is usually exactly what you want, because the pose converters expect rows of fixed width.

Empty input returns an empty list rather than erroring. That's the graceful path, and it means this node is safe to leave in a graph even before data flows.

Where does a transpose earn its keep here? Building camera-pose arrays is the classic case. If you assemble per-axis columns (all the radii in one list, all the elevations in another) and need to get them into row form for array1 to camposes - which wants each camera as one row of six values - you transpose your way there. It's also the general "reshuffle this 2D list" tool when you're converting between list layouts mid-workflow.

One honest caveat: for arrays deeper than 2D it still transposes, but numpy's default behavior on higher dimensions can give you something you didn't intend. In practice the pack's lists are 1D or 2D, so you're unlikely to hit it - just be aware the node doesn't warn you.

This is part of 807502278/ComfyUI-3D-MeshTool, a one-author 3D utility pack. Pure Python, no models. Install via ComfyUI Manager (search "ComfyUI-3D-MeshTool") and restart, or:

cd ComfyUI/custom_nodes
git clone https://github.com/807502278/ComfyUI-3D-MeshTool
cd ComfyUI-3D-MeshTool
pip install -r requirements.txt

And the pack's usual habit: node renames between releases turn old instances red in saved workflows - recreate the node and reconnect. When your data's the right values in the wrong shape, this is the one-line fix.