Refs Desk

Work out which of your inputs actually decides what. A modern generator takes several at once - a subject reference, a style reference, a first frame, a pose map and the text - and they are not peers: each governs some visual attributes, merely influences others, and ignores the rest. So text that contradicts an image is not a tiebreak, it is a no-op, and the sentence sits in the prompt looking like it worked. This resolves every attribute to the channel that governs it, reports two governors as an undecidable TIE rather than guessing - which one a pipeline keeps is a property of that pipeline, not of your request - names the attributes nothing governs and are therefore re-rolled every generation, and flags references that were uploaded and decide nothing. Adding a first frame to a request that already has a subject reference feels like it can only help; it ties on four attributes and silences two images. The claims matrix behind it is a model of how these pipelines behave rather than a measurement of any one of them, and the app says so wherever it shows a verdict. Five lanes: brief, check, assign, text and mode. The engine is free and runs in the browser. Lanes derived from samuraigpt/generative-media-skills (MIT), whose muapi-seedance-2 skill drives a generator across text-to-video, image-to-video, first-last-frame and omni-reference modes. Not affiliated with or endorsed by samuraigpt.

Every public app is built from a security-scanned skill and must pass a clean scan — skill and frontend — before it can be listed. Have a skill of your own? Turn it into an app — or read the step-by-step walkthrough.