Add max and min operations to MaskComposite

[NubeBuster]

Problem

MaskComposite's and, or and xor operations round both masks before combining:

https://github.com/comfyanonymous/ComfyUI/blob/master/comfy_extras/nodes_mask.py#L290-L294

torch.bitwise_or(destination_portion.round().bool(), source_portion.round().bool()).float()

MASK is a float tensor in [0, 1], and much of the ecosystem exists to produce intermediate values (blur, feather, grow, segmentation confidence). But there is currently no operation on this node that takes the union or intersection of two soft masks without destroying that feathering. multiply gives a product-style intersection, and add an approximate union, but neither is the standard union/intersection, and the three ops that are named like set operations all binarize.

The failure is silent — the output is still a valid MASK, so nothing complains, and the flattened feather only becomes visible much further downstream (e.g. as a hard seam in an inpaint composite).

Change

Adds max and min to the operation list — the standard fuzzy-set union and intersection, which preserve intermediate values:

elif operation == "max":
    output[:, top:bottom, left:right] = torch.max(destination_portion, source_portion)
elif operation == "min":
    output[:, top:bottom, left:right] = torch.min(destination_portion, source_portion)

Also adds a tooltip to the operation widget noting which operations are arithmetic and which binarize, since that distinction is not visible from the names alone.

Compatibility

Non-breaking. Existing operations are untouched, and the new options are appended to the end of the list. Combo values are serialized by name, so saved workflows are unaffected.

Example

destination [0.0, 0.25, 0.75, 1.0], source [0.5, 0.5, 0.5, 0.5]:

operation	result
max	[0.5, 0.5, 0.75, 1.0]
min	[0.0, 0.25, 0.5, 0.5]
or	[0.0, 0.0, 1.0, 1.0]

Tests

Adds tests-unit/comfy_extras_test/nodes_mask_test.py covering union/intersection semantics, preservation of intermediate values, commutativity and identity cases, plus one test documenting the existing binarizing behaviour of or. 9 tests, all passing.

🤖 Generated with Claude Code

---

PR Reviewed for being correct - Mark

[github-actions]

✅ All contributors have signed the CLA. Thank you! This PR is ready to be merged.
_{Posted by the CLA Assistant Lite bot.}

[coderabbitai]

📝 Walkthrough

Walkthrough

MaskComposite now supports documented max and min operations. These operations combine overlapping masks with elementwise tensor maxima or minima and preserve soft-mask values. Unit tests cover intermediate values, commutativity, identity cases, and existing binary or behavior.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Title check	✅ Passed	The title clearly and concisely describes the main change: adding max and min operations to MaskComposite.
Description check	✅ Passed	The description explains the new operations, compatibility, tooltip, and unit tests, all of which match the changeset.

✨ Finishing Touches 💡 1

🛠️ Fix failing CI checks 💡

• Create stacked PR
• Commit on current branch

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[NubeBuster]

I have read and agree to the Contributor License Agreement