render-song-mv

Name: render-song-mv
Author: gooseworks-ai

1.0kstars193forks

How to install render-song-mv?

gooseworks-ai/goose-skills/render-song-mv

$npx -y skills add gooseworks-ai/goose-skills --skill render-song-mv

Installs into the current project.

Summary

Assemble a song-driven music-video ad from a config — a generated sung track carries the whole narration across N tableaux (one keyframe -> one i2v clip per lyric beat) with NO separate voiceover, captions synced to the song's OWN word timings (script-window, never Whisper) and t

Files · 1

View on GitHub

SKILL.md

1	# render-song-mv
2
3	Assemble a song-driven music-video ad from a config: a purpose-written, sung song is
4	the entire narration (no separate voiceover), and every visual beat is timed to the lyrics.
5	The delivered song sets the timeline; N tableaux (one keyframe → one image-to-video clip per
6	lyric beat, all in a single look pack) are cut to their lyric windows and hard-concatenated
7	on the beat, captions are built from the song's OWN word timings with the hook line landing
8	on the chorus drop, and the spot closes on a PIL brand end card. It reads like a tiny animated
9	music video, not a demo. `scripts/config.example.json` is the worked example (Loóna "Fall In
10	Love With Sleep Again", 28s paper-craft 9:16); `scripts/PIPELINE.md` maps every config block
11	to its step and `scripts/README.md` documents the free assembly.
12
13	## Run
14
15	This is the FREE, deterministic assembly stage — it spends nothing. The three paid
16	inputs are separate capabilities: the sung song (`create-music-elevenlabs`, music_v1,
17	force_instrumental FALSE — the lyrics ARE the script, returns mp3 + `words.json`), one
18	keyframe per tableau (`create-image-fal`), and one Kling 3.0 i2v clip per tableau
19	(`create-video-fal`). Given the delivered song + `words.json` + one clip per beat,
20	`render-song-mv` cuts each clip to its lyric window, hard-concats on the beat, builds the
21	lyric-synced captions, composites the PIL end card, and muxes → the master. Re-cuts reuse
22	the existing song / keyframes / clips and cost $0.
23
24	## Contract (the free assembly)
25
26	- The sung song carries the narration — no separate VO. The generated ElevenLabs track
27	IS the bed and the script (`force_instrumental` false); do not add a spoken voiceover or a
28	second music bed.
29	- Plan the timeline AROUND the delivered song. The song is generated first and reshapes/
30	overshoots length; snap every tableau boundary to the lyric-phrase edges in the returned
31	word timings (`timeline.json`) — never trim the song to a pre-planned grid.
32	- Captions from the song's OWN word timings, not Whisper (script-window). Chunk
33	`audio/words.json` (~3 words at lyric boundaries); accent words get the warm-glow color.
34	Whisper on sung audio returns "🎵 Music Playing 🎵", so it can't caption lyrics.
35	- Land the hook on the chorus drop. Exactly ONE hero tableau (`is_hook`) is timed so the
36	payoff word (`song.hook_word`) sits on the chorus drop; accent that word in the captions.
37	- One look pack for consistency. A single `style_opener` + `negative_tail` + palette drives
38	every keyframe so N beats read as one film; no morph within a clip.
39	- Hard cuts on the beat. Cut each clip to its lyric window and hard-concat — no dissolves
40	(one optional match-cut into the hero reveal).
41	- PIL end card from the real app icon — never AI-render brand text. The lockup is
42	composited deterministically (brand gradient + circular app icon + wordmark + tagline + CTA)
43	from the brand's real asset; a diffusion model garbles a wordmark.
44	- FFmpeg composite, deterministic, FREE. Burn the caption ASS, overlay the end-card PNG on
45	the final window, mux the song, boost the climax beat, loudnorm to −14 LUFS → 1080×1920
46	h264+aac. No paid calls, no keys.

People who liked this also viewed

6 picks

Type

render-song-mv

1.0kstars193forks

View on GitHub

How to install render-song-mv?

gooseworks-ai/goose-skills/render-song-mv

$npx -y skills add gooseworks-ai/goose-skills --skill render-song-mv

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# render-song-mv
2
3	Assemble a song-driven music-video ad from a config: a purpose-written, sung song is
4	the entire narration (no separate voiceover), and every visual beat is timed to the lyrics.
5	The delivered song sets the timeline; N tableaux (one keyframe → one image-to-video clip per
6	lyric beat, all in a single look pack) are cut to their lyric windows and hard-concatenated
7	on the beat, captions are built from the song's OWN word timings with the hook line landing
8	on the chorus drop, and the spot closes on a PIL brand end card. It reads like a tiny animated
9	music video, not a demo. `scripts/config.example.json` is the worked example (Loóna "Fall In
10	Love With Sleep Again", 28s paper-craft 9:16); `scripts/PIPELINE.md` maps every config block
11	to its step and `scripts/README.md` documents the free assembly.
12
13	## Run
14
15	This is the FREE, deterministic assembly stage — it spends nothing. The three paid
16	inputs are separate capabilities: the sung song (`create-music-elevenlabs`, music_v1,
17	force_instrumental FALSE — the lyrics ARE the script, returns mp3 + `words.json`), one
18	keyframe per tableau (`create-image-fal`), and one Kling 3.0 i2v clip per tableau
19	(`create-video-fal`). Given the delivered song + `words.json` + one clip per beat,
20	`render-song-mv` cuts each clip to its lyric window, hard-concats on the beat, builds the
21	lyric-synced captions, composites the PIL end card, and muxes → the master. Re-cuts reuse
22	the existing song / keyframes / clips and cost $0.
23
24	## Contract (the free assembly)
25
26	- The sung song carries the narration — no separate VO. The generated ElevenLabs track
27	IS the bed and the script (`force_instrumental` false); do not add a spoken voiceover or a
28	second music bed.
29	- Plan the timeline AROUND the delivered song. The song is generated first and reshapes/
30	overshoots length; snap every tableau boundary to the lyric-phrase edges in the returned
31	word timings (`timeline.json`) — never trim the song to a pre-planned grid.
32	- Captions from the song's OWN word timings, not Whisper (script-window). Chunk
33	`audio/words.json` (~3 words at lyric boundaries); accent words get the warm-glow color.
34	Whisper on sung audio returns "🎵 Music Playing 🎵", so it can't caption lyrics.
35	- Land the hook on the chorus drop. Exactly ONE hero tableau (`is_hook`) is timed so the
36	payoff word (`song.hook_word`) sits on the chorus drop; accent that word in the captions.
37	- One look pack for consistency. A single `style_opener` + `negative_tail` + palette drives
38	every keyframe so N beats read as one film; no morph within a clip.
39	- Hard cuts on the beat. Cut each clip to its lyric window and hard-concat — no dissolves
40	(one optional match-cut into the hero reveal).
41	- PIL end card from the real app icon — never AI-render brand text. The lockup is
42	composited deterministically (brand gradient + circular app icon + wordmark + tagline + CTA)
43	from the brand's real asset; a diffusion model garbles a wordmark.
44	- FFmpeg composite, deterministic, FREE. Burn the caption ASS, overlay the end-card PNG on
45	the final window, mux the song, boost the climax beat, loudnorm to −14 LUFS → 1080×1920
46	h264+aac. No paid calls, no keys.

People who liked this also viewed

6 picks

Type