Smart Point

Smart Point is a two-widget node that outputs exactly one thing: a normalized POINT - a single [x, y] coordinate in 0–1 space. That sounds like almost nothing, and honestly it is. But it's the seed for the whole point system in this pack, and the moment it earns its keep is when you pair it with SmartBackgroundRemove, the pack's flood-fill background remover. Click (well, type) where the background is, and the flood fill starts growing from there like a magic wand in Photoshop.

How it works

You get two FLOAT widgets, x and y, both defaulting to 0.5 and stepping by 0.01. The node doesn't look at any image - it just packages the numbers into a POINT. Two things worth knowing:

• Coordinates are normalized, not pixel values. 0.5, 0.5 is the center of whatever image you connect downstream, regardless of resolution. 0, 0 is the bottom-left corner, because ComfyUI uses a bottom-left (math-style) Y origin; the interactive node in this pack flips Y for you so a point you drag at the top comes out as a small Y.
• It's a manual counterpart to SmartImagePoint, which lets you drag dots directly on an image canvas. Use SmartPoint when you want a coordinate you can drive from a number, a seed, or a prompt - or when you don't want to click a canvas at all.

Where it plugs in

The POINT output feeds the point consumers in this pack:

• SmartBackgroundRemove (start_point) - the point you click is where it samples the background color and flood-fills outward to delete it. One point means "one color region removed."
• SmartPointSet → wraps it into a POINT_SET so it can join others.
• SmartDrawPoints takes a POINT_SET (not a raw point), so you'll go through the set nodes first.

For the single-point case, the practical wiring is: SmartPoint → SmartBackgroundRemove, then out to the rest of your workflow.

Installing it

Smart Point ships in ComfyUI-SmartImageTools - there's no standalone node. Install the pack once and you get all of these nodes:

cd ComfyUI/custom_nodes/
git clone https://github.com/slvslvslv/ComfyUI-SmartImageTools
pip install -r ComfyUI-SmartImageTools/requirements.txt

Then restart ComfyUI. ComfyUI Manager also finds it by searching "ComfyUI-SmartImageTools". The pack's actual requirements.txt pulls in scikit-learn, scikit-image, numpy, Pillow, opencv-python, and numba - heavier than the README advertises, and the first start after install can be slow while scikit-learn and friends compile or download. That's a one-time cost.

Gotchas

• A POINT only means something to nodes that understand the pack's POINT type. It won't drop into arbitrary node inputs - the ports only match other SmartImageTools point sockets.
• Defaults are center. Forgetting to change x/y means flood removal starts at the middle of your image, which is rarely where the background is.
• This is a genuinely small pack, so don't expect a huge tutorial trail - the best documentation is the pack source and the small set of point-aware nodes you can wire together in ten minutes.

Smart Point is the boring plumbing node of this pack, and that's fine. It's the piece you build multi-point sets out of, and multi-point sets are what make flood-fill background removal work on anything messier than a studio backdrop.