Interpolate Clip Sequential (mtb)

Full disclosure up front: this is one of the rare mtb nodes that ships with no description at all - not in the README, not in the node's own metadata. So this page is built from what the input names tell us and how CLIP conditioning works generally, not from the author's own words. If you've used it and know exactly what it does under the hood, take everything below as a reasonable read of the field names rather than gospel.

What the inputs suggest it's doing

You give it base_text - your normal prompt - and a text_to_replace string. The clip input is a standard CLIP model for text encoding. And there's an interpolation_strength dial from 0 to 1. Put together, the shape of this node is a prompt-blending tool for animation: rather than hard-cutting your conditioning from one prompt to a completely different one between frames, it produces a CONDITIONING output that blends toward a version of your prompt with text_to_replace worked in, controlled by how far you turn interpolation_strength up. At 0 you'd expect something close to base_text encoded plain; at 1, something closer to the replaced version. "Sequential" in the name points at this being meant to run across a sequence of frames - you'd typically drive interpolation_strength from a frame counter through something like Fit Number (mtb), easing it from 0 to 1 across your animation so the prompt's meaning drifts smoothly instead of jumping.

This is the ComfyUI-graph equivalent of "prompt travel" - the technique of morphing between two prompt concepts over the course of a generation rather than switching abruptly.

Inputs that matter

• base_text - your starting prompt (multiline, so full-length prompts are fine).
• text_to_replace (default empty) - the word or phrase you're blending toward.
• clip - the CLIP model doing the text encoding, same as any standard prompt-encoding node.
• interpolation_strength (0–1, default 1) - how far along the blend you are.

Output: a single CONDITIONING, ready to feed into your sampler like any other conditioning.

Installing it

• ComfyUI Manager - search "MTB Nodes", install, restart.
• Manual - cd ComfyUI/custom_nodes && git clone https://github.com/melMass/comfy_mtb, restart. A manual clone needs pip install -r requirements.txt run inside the comfy_mtb folder first.

No models beyond the CLIP encoder you'd need for any prompt-conditioning node anyway.

Common issues & troubleshooting

Output doesn't look like it's blending at all. Given the lack of documentation on this node, the safest first move if results look off is a controlled test: hold everything else fixed, sweep interpolation_strength from 0 to 1 in a few steps, and actually look at what each value produces before building a full animation around it. Don't assume the direction (0 = base, 1 = replaced) without checking - it's a reasonable read of the field order, not a confirmed spec.

text_to_replace seems to have no effect. Since the default is an empty string, make sure you've actually filled it in - an empty replacement target likely just gives you base_text regardless of interpolation_strength.

Some mtb nodes fail to load at startup. Normal for this pack - check the console for [comfy_mtb] STATUS and http://127.0.0.1:8188/mtb for what didn't load and why. This node needs nothing beyond a working CLIP model, so it's rarely the one missing.