abyz22_Topipe

If you've ever dragged MODEL, CLIP, VAE and two conditioning wires across a busy workflow, you've already wanted abyz22_Topipe. It bundles the standard ingredients - model, CLIP, VAE, positive, negative, and an image - into a single PIPE output. Its siblings in the same pack (abyz22_Frompipe, abyz22_Editpipe, abyz22_Convertpipe) unpack or swap individual fields later in the graph. One fat wire instead of six skinny ones, and suddenly you can actually see what's connected to what.

How it works: there's no magic and no processing. Topipe just wraps the inputs in a dict - pipe["MODEL"], pipe["CLIP"], pipe["VAE"], pipe["POSITIVE"], pipe["NEGATIVE"], pipe["IMAGE"], plus the optional latent_Image and latent if you pass them. Frompipe reads the same keys back out. It's the author's own pipe convention, and if you've used Impact Pack's BasicPipe the idea will feel familiar - the field names are close. Don't assume cross-compatibility with Impact's pipe consumers, though. The slots aren't guaranteed to line up, so use it with this pack's Frompipe/Editpipe family and you're on safe ground.

Inputs: all six required - MODEL, CLIP, VAE, Positive, Negative, IMAGE - the same ingredients you'd normally feed into a KSampler. The optional latent_Image and latent get stashed in the pipe so a downstream Frompipe can hand them back out without you re-routing them.

Output: a single pipe. In the pack's own sample workflow, one Topipe feeds roughly ten Frompipe nodes spread across a detect-crop-refine graph built on Impact Pack's SEGS machinery. Without the pipe that graph would be a spaghetti of parallel wires that nobody could read. That's the real reason to reach for this node: not because it adds capability, but because it turns an unreadable graph into a legible one.

Gotchas worth knowing before you rely on it:

• No validation, no defaults. Leave latent empty and Frompipe hands back None for that slot. If a node downstream can't take None you get a confusing error in the middle of a long graph, and the error won't point at the pipe.
• Because a pipe is just a dict, mismatched keys surface at consume-time, not build-time. When something breaks, check the Frompipe side first - it's where missing keys announce themselves.

Install is the standard custom-node routine for this pack: ComfyUI Manager → search image_control, or cd ComfyUI/custom_nodes && git clone https://github.com/abyz22/image_control, then restart. The pipe nodes need no models and no special dependencies beyond the pack's requirements.

Bottom line: if your graphs are small, you don't need this. The moment you're running a multi-stage detail loop with a dozen stages, a pipe bundler stops being a nicety and starts being the difference between a graph you can edit and a graph you're afraid to touch.