> ## Documentation Index
> Fetch the complete documentation index at: https://comfytv.org/llms.txt
> Use this file to discover all available pages before exploring further.

# Mesh Boolean

> Combine two meshes with CSG — union to merge, difference to cut one out of the other, intersect to keep only the overlap — positioned with an on-card gizmo.

<Frame caption="Mesh Boolean · screenshot coming soon">
  <img src="https://mintcdn.com/comfytv/dMkMxJDZdIo4YzAU/images/nodes/_placeholder.svg?fit=max&auto=format&n=dMkMxJDZdIo4YzAU&q=85&s=b54171099e5b02d4313fe581171ebbcc" alt="Mesh Boolean" width="800" height="450" data-path="images/nodes/_placeholder.svg" />
</Frame>

## What this node does

Mesh Boolean takes two `COMFYTV_MODEL` inputs (`model` = A, `model_b` = B) and combines them on a signed-distance voxel grid, returning one new model. Choose the `operation` and position B (and A) with the gizmo in the node body; the transforms are stored internally and applied before the cut.

Because it works on an SDF grid, the result is watertight but its face count depends on `resolution` — a higher grid gives a sharper cut with more faces. It runs the vendored mesh3d backend (▶ Run). Both inputs are required: the stage errors if `model` or `model_b` is empty. The card's 3D preview snapshot becomes the `image` output.

Outputs are ComfyTV snapshots, not native tensors — insert a **Bridge** to feed native ComfyUI nodes ([bridges docs](https://github.com/jtydhr88/ComfyTV/blob/main/docs/bridges.md)).

## When to use it

* Drill holes or cut slots — subtract a cylinder or cube (B) from a body (A) with `difference`.
* Weld two overlapping shapes into one solid with `union`.
* Keep only the shared volume of two shapes with `intersect` (e.g. carve a shape to a mold).

## Parameters

### operation

The CSG operation: `union` (merge both), `difference` (remove B's volume from A), `intersect` (keep only the overlap). Default `union`.

### resolution

SDF voxel grid resolution. Default `256`, range `32`–`1024`. Higher = sharper cut and more faces; follow with a **decimate** in **Mesh Ops** if you need an exact face count.

### smooth\_iters

Taubin smoothing iterations applied to the result. Default `0`, range `0`–`20`. A few passes soften the stair-stepping that a low `resolution` can leave along the cut.

### model (A) / model\_b (B)

The two `COMFYTV_MODEL` inputs. `model` is A, `model_b` is B. For `difference`, B's volume is removed from A. Both are required at run time.

### transform\_a / transform\_b

Internal (hidden). The position / rotation / scale (TRS) of each model, set by the gizmo in the node body. You place the meshes visually; you never edit this JSON by hand.

### captured\_image

Internal (hidden). The 3D preview snapshot URL that becomes the `image` output.

## Outputs

| Output    | Type            | Meaning                    |
| --------- | --------------- | -------------------------- |
| **model** | `COMFYTV_MODEL` | The combined mesh (GLB)    |
| **image** | `COMFYTV_IMAGE` | Snapshot of the 3D preview |

## Tips

* Order matters for `difference`: A is the body, B is the tool removed from it. Swap the wires if the cut is inverted.
* The output can carry more faces than you want; chain a **Mesh Ops** decimate to bring it to a budget.
* If a thin cut looks jagged, raise `resolution` or add a couple of `smooth_iters` rather than both at once.

## Related nodes

* **Mesh Primitive** — make the cutting tool (cylinder for a hole, cube for a slot).
* **Mesh Ops** — decimate / remesh / weld the boolean result.
* **Mesh Bake Maps** — bake detail after cleanup and unwrap.