ConditioningSetAreaPercentage_motorway_edition

ConditioningSetAreaPercentage is ConditioningSetArea's friendlier sibling: same idea - confine a conditioning stream to a rectangular region - but the rectangle is specified as fractions of the image (0.0–1.0) instead of raw pixels. That means your region definitions survive a resolution change without redoing the math. This _motorway_edition clone runs that off agilly1989's keyed motorway bag.

If you've seen ConditioningSetArea and bounced off the pixel coordinates, this is the one to reach for: "left half" is just width 0.5, height 1.0, x 0, y 0, and it stays true whether you render 512×512 or 1024×1024. The motorway version changes nothing about that logic - the conditioning is pulled from a bag key, the region is applied, and the result is written back under a key.

The fields

Same shape as the pixel version, with float rectangles:

• MOTORWAY 🚌💨 - in and out.
• INPUT_conditioning_key (default conditioning) - the conditioning to confine.
• width, height - region size as a fraction, 0.0 to 1.0, default 1.0 (full image).
• x, y - top-left corner as a fraction, default 0.
• strength - 0 to 10, default 1.0, the region's influence.
• OUTPUT_CONDITIONING_key (default CONDITIONING) - where the regioned conditioning lands.

Single output, the motorway itself, like every node in the family.

Install and the standing caveat

The easy part first:

cd ComfyUI/custom_nodes
git clone https://github.com/agilly1989/ComfyUI_agilly1989_motorway

Restart ComfyUI, or install ComfyUI_agilly1989_motorway via ComfyUI Manager. Pure Python, no pip requirements, no model downloads.

The caveat, once more for this pack: single-author beta, README banner "BIG BROKEN WITH ASYNC WILL FIX WHEN I GET THE TIME/MOTIVATION," and the machine-generated clones are disabled in current releases - the registration in __init__.py is commented out. Whether this clone shows up in your menu depends on the version you're running; if it doesn't, the stock ConditioningSetAreaPercentage plus a Motorway 0x1 ramp does the same job. And the pack's one failure mode remains the same: 'MotorwayClass' object has no attribute 'hash_' means a key you referenced was never written into the bag. For a node that only reads conditioning and writes CONDITIONING, the entire mystery is usually "did anything upstream actually publish under the default key."