Self-hosted, Self-Improving, Federated Everything App — dev management, agent orchestration, persistent mind, digital twin, second brain, and creative pipeline from a single dashboard over Tailscale to port the user into the OS
JavaScript ai-agentsdashboarddeveloper-toolsdigital-twinquantified-selfsecond-brainself-hostedtailscale
Tell the treatment draft not to write song times, clip lengths, trim notes or "no lip-sync" into motion prompts
Claudef78621af1872
diff --git a/server/services/musicVideo/treatmentDraft.js b/server/services/musicVideo/treatmentDraft.jsindex 12b5f775e6..3ad2ba9da1 100644--- a/server/services/musicVideo/treatmentDraft.js+++ b/server/services/musicVideo/treatmentDraft.js@@ -417,7 +417,7 @@ ${project.styleReferences?.length ? '- "styleLook": one short sentence summarizi- Two or three recurring "motifs" (an image, object, color or gesture) with how each changes across the arc.- The balance of "performance", "cutaway" and "graphic" shots as percentages, with a rationale.- "graphicLanguage": a concise visual note for the graphic cards and typography (HUD, pictograms, counters), honoring any graphic direction in the brief.-${hasLyrics ? '- "lyricInterpretation": what the lyrics mean and how the picture interprets them (not word-for-word illustration).\n' : '- The song is instrumental: set "lyricInterpretation" to null and invent no lyrics.\n'}- For EACH shot: "mode" (performance|cutaway|graphic), "route" (generated|supplied-asset|code-2d — code-2d becomes a code-rendered title card for a sung line or a moved still; prefer it where a video model cannot hold the requirement), "focalSubject", "framing", "negativeSpace" (none|upper|center|lower — the region kept clean for the separately composited text), "typographyRole" (none|subtitle|hero; must be none on instrumental shots), "emphasis", "transitionIn", "transitionOut", "rationale", and a "framePrompt" (the opening still) and "prompt" (the motion) for the image/video model.+${hasLyrics ? '- "lyricInterpretation": what the lyrics mean and how the picture interprets them (not word-for-word illustration).\n' : '- The song is instrumental: set "lyricInterpretation" to null and invent no lyrics.\n'}- For EACH shot: "mode" (performance|cutaway|graphic), "route" (generated|supplied-asset|code-2d — code-2d becomes a code-rendered title card for a sung line or a moved still; prefer it where a video model cannot hold the requirement), "focalSubject", "framing", "negativeSpace" (none|upper|center|lower — the region kept clean for the separately composited text), "typographyRole" (none|subtitle|hero; must be none on instrumental shots), "emphasis", "transitionIn", "transitionOut", "rationale", and a "framePrompt" (the opening still) and "prompt" (the motion) for the image/video model. The motion "prompt" is plain physical action: no song times, clip lengths, trim notes ("use the first 2 s") or "no lip-sync" line, which the video model cannot act on.- For EACH shot also supply "medium" (procedural|still|existing-footage|generated-footage) and "mediumRationale". Medium is independent of performance/cutaway intent. Retain director pins. Tie the medium to the shared motifs, their evolution across repeated sections, and entry/exit transitions.- MEDIA MODE: ${musicVideoMediaMode(project)}. Code authors scene composition, characters, environment, camera and timing in every mode. Code-only forbids images even for visual guides; code-images forbids generated or imported footage.- For code-first, prefer procedural or still. Existing footage must be selected already; generated footage is an explicit exception within the allowance. At zero allowance never assign generated footage: report unmet performance intent rather than quietly generating it. This is planning, not a claim that procedural rendering is ready.