**BETA**Flex Video Frame Blend ⚡🅡🅞🅣🅘

This one ships flagged BETA in its own display name, and the schema backs that up in a specific, checkable way: the feature_param field's only listed choice is the literal string error_not_implemented. Read that at face value - as of this build, feature modulation isn't actually wired up on this node yet, even though the opt_feature socket exists and accepts a connection. Worth knowing before you spend time debugging a reactive setup that has nowhere to actually land.

How it works

Setting the beta caveat aside, the node itself is a frame-blend effect: it mixes each frame with a neighboring one to fake motion blur or a soft dissolve between frames, rather than true optical-flow interpolation. frame_offset_ratio sets how far away (as a fraction of the sequence) the blended neighbor frame is. blend_strength controls how much of that neighbor gets mixed in, direction_bias weights whether the blend leans toward the frame ahead or behind, and blend_mode picks the compositing operator - normal (plain crossfade), additive, multiply, or screen. motion_blur_strength layers an extra blur pass on top for a softer, more smeared look rather than a crisp double-exposure.

Inputs and outputs that matter

• images (required, IMAGE) - your video frames.
• blend_strength (default 0.5, 0–1) - how much of the neighbor frame blends in.
• frame_offset_ratio (default 0.1, 0–1) - how far away that neighbor frame is.
• direction_bias (default 0.5, 0–1) - weights the blend toward the frame ahead or behind.
• blend_mode - normal, additive, multiply, or screen.
• motion_blur_strength (default 0, 0–1) - an additional blur pass.
• opt_feature (optional, FEATURE) - accepted, but not yet functional per the schema's own error_not_implemented marker.
• Output - a single IMAGE batch.

Installing it

Via ComfyUI Manager: search "RyanOnTheInside," install, restart. Manually:

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

then restart. No model downloads - it's frame compositing, not a learned model.

Common issues & troubleshooting

Feature input does nothing. This is expected given the beta status, not a wiring mistake - don't spend time debugging feature_param/feature_threshold settings on this node until a future release actually implements it.

Result looks like a double-exposure rather than smooth motion blur. That's the nature of a fixed frame-offset blend rather than true optical-flow interpolation - it approximates motion blur by mixing two discrete frames, so fast motion or a large frame_offset_ratio will read as a visible ghost rather than a seamless smear. If you need genuinely smooth interpolation instead, look at the pack's FlexVideoSpeed node or a dedicated frame-interpolation pack.

Being a beta node, expect rough edges. Given it's explicitly marked beta and has at least one unfinished feature (the modulation path), treat this as an experimental effect to try rather than something to build a production pipeline around just yet.