InstantID Pose Prepare Pipe (JPS)

Pairs with InstantID Pose Prepare Settings (JPS) - the node that frames a pose reference image (crop, pad, resize, flip, sharpen) before InstantID's IdentityNet module reads facial landmarks from it. This pipe splits that bundle back into individual outputs wherever your graph actually consumes them.

What comes out of it

One input, imageprepare_settings - a BASIC_PIPE (JPS's own bundle, a plain tuple, not Impact Pack's object of the same type name). Fourteen outputs:

• resize_to (INT) - resize to target, resize to source, or keep size.
• offset_width / offset_height - positioning nudge.
• crop_left / crop_right / crop_top / crop_bottom and padding_left / padding_right / padding_top / padding_bottom - trim from and add space to each edge independently.
• interpolation - resize algorithm (lanczos, nearest, bilinear, bicubic, area, nearest-exact).
• sharpening (FLOAT) - post-resize sharpen amount.
• flip - No, X-Axis, or Y-Axis, for mirroring a pose reference that's facing the wrong way for your composition.

No mask_type here, unlike its Mask Prepare sibling - this pipe is purely about framing a pose image, not building a mask. This node performs none of the actual crop/pad/resize/flip work itself; that happens in a downstream image-processing node this pipe's outputs feed, before the result reaches InstantID's pose-conditioning input.

Installing it

Ships with the pack. ComfyUI Manager: search JPS Custom Nodes for ComfyUI, install, restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/JPS-GER/ComfyUI_JPS-Nodes.git

Restart afterward, deleting any earlier copy of the pack first. No models needed for this node specifically, though the InstantID pipeline it feeds needs its own checkpoint and InsightFace's ArcFace models - both research/non-commercial licensed, worth knowing before this ends up in a paid product.

Where people get tripped up

Without an InstantID Pose Prepare Settings (JPS) node feeding it, this pipe has nothing to unpack.

The same naming landmine that affects this node's siblings applies here too: imageprepare_settings is a label reused by Image Prepare Settings (JPS), InstantID Mask Prepare Settings (JPS), and InstantID Source Prepare Settings (JPS) for their own outputs, all typed BASIC_PIPE. Since that type carries no schema ComfyUI checks at connection time, it's possible to wire the wrong Settings node into this pipe - the UI won't stop you, since name and type both match - and get back fourteen misaligned values instead of a clean error. If your pose reference isn't ending up cropped, padded, or flipped the way you configured, trace imageprepare_settings back to its source before assuming the Settings node's values themselves are wrong; there's a real chance it's plugged into the wrong one. Past that, if the generated pose doesn't match what you expected even with correct wiring, check resize_to first - "Resize to Target" versus "Resize to Source" versus "Keep Size" produce meaningfully different framing, and it's an easy setting to leave on the wrong option out of habit.