Mask Math [DEPRECATED] ⚡🅡🅞🅣🅘

Same story as this pack's Audio Latent Blend: the node's own display name says "Mask Math [DEPRECATED]", right there in the ComfyUI search list, so treat it as legacy rather than the tool you reach for on a new build. It still runs - the pack hasn't stripped the code - but everything it does now has a modern equivalent living in the newer "Flex" mask nodes in this same pack, which add the feature-reactive modulation this one never had.

What it actually does

Mask Math combines two masks and applies a time window to when that combination is active - which makes sense given its category is "TemporalMasks." It takes masks (your primary input, single mask or a batch/sequence), optionally mask_b (a second mask to combine against), a combination_method (add, subtract, multiply, minimum, maximum), and a set of post-processing knobs: invert flips black/white on the result, subtract_original peels a portion of the original mask back out after combination, and grow_with_blur softens/expands edges with a Gaussian blur.

The temporal side is the part that separates this from a plain mask-combine node: start_frame, end_frame, and effect_duration define when in a frame sequence the effect is active, temporal_easing (ease_in_out, linear, bounce, elastic, none) shapes how the effect ramps in and out rather than switching on like a hard cut, and palindrome makes the effect play forward then reverse within its duration - useful for a mask that should pulse rather than just appear once.

Inputs and outputs

• masks (required, MASK) - the mask or mask sequence being processed.
• strength (0–1) - overall intensity of the effect.
• combination_method and mask_b - how (and with what) to combine, if you're combining at all.
• start_frame / end_frame / effect_duration / temporal_easing / palindrome - the timing controls that give this node its "temporal" name.
• Output - a single MASK.

If you don't need the timing behavior at all, note that the strength, invert, subtract_original, and grow_with_blur options work as a general mask-touch-up toolkit even with the temporal window left at its defaults (start/end at 0, meaning full range).

Installing it

Install via ComfyUI Manager (search RyanOnTheInside) or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/ryanontheinside/ComfyUI_RyanOnTheInside

Run pip install -r requirements.txt, then restart ComfyUI. This node has no unusual dependency of its own beyond the standard image/mask processing libraries the rest of the pack already needs.

Common issues

The practical trap is discoverability: because it's deprecated but not hidden, it still shows up in node search right alongside the newer Flex mask tools, and nothing stops a new workflow from using it by accident. If you're starting fresh, look at the pack's Flex-prefixed mask nodes first - they cover the same combine-and-shape territory with an opt_feature input this one lacks, meaning the blend strength or timing can react to audio or motion instead of being fixed per-run.

If you do use it: end_frame and effect_duration both default to 0, which the tooltips define as "until end" / "full range" rather than zero-length - so don't be surprised the effect spans your whole sequence by default even though the numeric field reads 0. And as with the rest of this pack, a totally unresponsive node on load points at a missing pip install -r requirements.txt, not a problem with this specific node's logic.