MultiPipe 1.5 Out

This is the receiver end. If you've loaded an AegisFlow-style workflow, MultiPipe 1.5 Out is the socket sitting at the end of every cell - the node that takes one fat wire and gives you your individual components back. It's the unpack half of the SD 1.5 pipe pair, and its whole job is making the boundary between two parts of a graph a single connection instead of a dozen.

You hand it one PIPE_LINE and it splits out the twelve things the 1.5 side of this pack collects: image, mask, latent, model, vae, clip, positive, negative, and four ints (image_width, image_height, latent_width, latent_height). Where each one goes is mostly obvious from the socket type - image into a VAE decode or preview, model + clip + positive + negative into a KSampler, latent into that same sampler, mask into inpainting or compositing, and the ints into anything resolution-driven.

Two outputs deserve a closer look. First, unlike its XL sibling, this node passes the original pipe through as its first output - handy if you want to keep the bundle for a second consumer or chain another node off the same pack. Second, it emits a "discord link" string. Wire it to a Show Text if you want a Discord invite parked on your canvas; more realistically, ignore it.

Mechanically there's nothing to tune: the pipe is a fixed-order Python tuple, and this node reads it positionally, exactly as af_pipe_in_15 wrote it. That's where the one real gotcha lives - pairing matters. Wire a pipe that the XL In node built into this 1.5 Out and every output shifts, because the XL tuple inserts an sdxl_tuple and a refiner quartet in the middle. Keep the 1.5 In and Out as a pair (they default to 512 dimensions, which matches SD 1.5's native resolution), and don't cross the streams.

If you're only here because a workflow wouldn't load without it, that's fine too. This is a passive unpacking node - nothing generates, nothing samples. Its whole contribution is keeping a large workflow readable, which is the same quiet job quality-of-life nodes like rgthree's do across the ecosystem.

Install is the standard two-liner:

cd ComfyUI/custom_nodes
git clone https://github.com/aegis72/aegisflow_utility_nodes

Restart ComfyUI and you're done. Or use ComfyUI Manager → Install Custom Nodes → search "AegisFlow Utility Nodes" → install. There are no model downloads and no requirements.txt in the repo; the pack is one self-contained file. Some of its fx siblings auto-install a pip library on first import, but this node has no extra dependencies at all - just restart and it'll be in your node list under AegisFlow/passers.