Flex Mask Interpolate ⚡🅡🅞🅣🅘

A linear crossfade between two masks is fine until it isn't - a hard edge that eases in feels more natural than one that snaps or ramps at a constant rate, and a radial reveal reads completely differently from a straight blend. Flex Mask Interpolate gives you both: a real set of interpolation curves to blend masks toward mask_b, plus (like the rest of this pack's Flex family) an optional FEATURE input so the blend amount itself can move with audio or motion instead of sitting at a fixed value.

How it works

interpolation_method is the core choice - nine options ranging from the expected (linear, ease_in, ease_out, ease_in_out, cubic, sigmoid) to the more visually distinct (radial, which interpolates outward from a center point rather than uniformly; distance_transform, which blends based on distance from the mask's edges rather than a flat time curve; and random_noise, which breaks the transition up with noise instead of a clean gradient). max_blend caps how far the interpolation can go - 1.0 lets it fully reach mask_b, lower values hold back some of the original. blend_mode (normal, add, multiply, overlay, soft_light) then determines how the two masks actually combine once the interpolation weight is computed, and invert_mask_b flips the second mask before any of that happens.

Layered underneath is the same general mask toolkit you'll see across this pack's Flex mask nodes: invert flips the final output, subtract_original peels back a portion of the source mask after blending, and grow_with_blur softens or expands edges. mask_strength scales the overall effect independently of max_blend.

Inputs and outputs

• masks (required, MASK) - the base mask or sequence.
• mask_b (required, MASK) - the second mask to interpolate toward.
• interpolation_method - the curve shape, the setting that actually defines the character of the transition.
• opt_feature (optional, FEATURE) - drives the blend reactively if connected; otherwise the transition is fixed.
• Output - a single MASK.

One thing worth knowing before you rely on the feature input

The feature_param dropdown on this node - the setting that's supposed to pick which parameter the feature modulates - currently only exposes a placeholder value in its schema rather than a real option like max_blend. In practice, that means feature-driven modulation on this specific node isn't fully wired up the way it is on its sibling Flex nodes. The deterministic side (everything not touching opt_feature) works exactly as described above; if you were counting on this one to react to audio the same way Flex Image Posterize or Flex Mask Morph do, test it before building a workflow around that assumption.

Installing it

Install via ComfyUI Manager (search RyanOnTheInside) or manually:

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

Then pip install -r requirements.txt and restart ComfyUI. No unusual dependency for this specific node beyond the pack's standard image/mask stack.

Common issues

Beyond the feature_param gap above, the most common surprise is interpolation_method mattering far more than max_blend - people dial in max_blend expecting to control the look, when it's actually interpolation_method that decides whether the transition looks like a fade, a radial wipe, or a noisy dissolve. If the blend looks flat or unremarkable, try radial or distance_transform before assuming the node isn't doing anything interesting.