render-split-screen-creator

Name: render-split-screen-creator
Author: gooseworks-ai

1.0kstars193forks

How to install render-split-screen-creator?

gooseworks-ai/goose-skills/render-split-screen-creator

$npx -y skills add gooseworks-ai/goose-skills --skill render-split-screen-creator

Installs into the current project.

Summary

Assemble a split-screen creator ad from a config — a two-zone vertical composite where a supplied AI-creator lip-sync take fills the BOTTOM ~48% while real 16:9 product/demo clips run uncropped in the TOP ~52%, each top clip contain-fit with a darkened blurred cover-scale fill of

Files · 1

View on GitHub

SKILL.md

1	# render-split-screen-creator
2
3	Assemble a split-screen creator ad from a config: a two-zone vertical
4	(1080×1920, 9:16, ~40s) format where an AI creator talking-head anchors the
5	BOTTOM ~48% of the frame and real 16:9 product/demo clips run uncropped in
6	the TOP ~52%, a 3px brand-color divider between the zones. The creator delivers
7	the whole VO cold-to-camera and each top clip proves the claim its VO line makes.
8	This capability is the FREE, deterministic assembly + captions — the two-zone
9	composite (contain-fit + blurred-cover fill + divider + creator slice), the
10	hard-concat, the end card, and the word-level caption burn from the assembled cut.
11
12	`scripts/config.example.json` is the worked example (Perplexity concept-10
13	"Bloomberg terminal", ~40s 1080×1920 9:16, 6 scenes + an end card);
14	`scripts/PIPELINE.md` maps every config block to its source step and
15	`scripts/README.md` documents the free assembly.
16
17	## Run
18
19	This is the FREE, deterministic assembly + captions stage — it spends
20	nothing. The paid inputs are separate steps — the VO (`create-vo-elevenlabs`,
21	ElevenLabs `eleven_v3` with-timestamps, sliced into per-scene windows), the
22	photoreal MEDIUM chest-up AI-creator anchor (`create-image-gpt-image-fal`,
23	`gpt-image-2`) — shot at a natural webcam distance (headroom + shoulders, real
24	room), not a plain-background close-up headshot (see the anchor note below), and
25	the whole-VO lip-sync (a paid VEED Fabric 1.0 @ 720p take — a no-atom step,
26	`image_url` = the anchor, `audio_url` = the vo mp3, ~$0.15/sec, ~$5.90 for a 39s
27	VO; run its calls sequentially, `veed/fabric-1.0` storage-auths 403 under
28	parallel load). Given the creator lip-sync take + the per-scene VO timing + one
29	16:9 top clip per scene + the scene-1 hook graphic + the end-card clip,
30	`render-split-screen-creator` composites the two zones, hard-concats the scenes,
31	appends the end card, transcribes the assembled cut, and burns the captions → the
32	master. Re-cuts reuse the existing VO / lip-sync / clips and cost $0.
33
34	## Contract (the free assembly)
35
36	- Two-zone split, top ~52% / creator ~48%. The TOP zone runs the real 16:9
37	product/demo clip contain-fit (uncropped), the BOTTOM zone is the creator
38	lip-sync framed head-to-shoulders: scale-to-width × a small ZOOM (~1.15–1.2)
39	then crop the zone with a downward offset so the face sits upper-middle and the
40	shoulders enter the bottom. (A plain cover + crop-toward-top shows only the head and
41	cuts the shoulders — and it can't rescue an anchor that was shot too close; fix the
42	anchor distance first.) Tune zoom/offset visually against the source video's creator
43	framing — it's a FREE re-assemble, no VEED re-run. A 3px brand-color divider separates
44	the zones. Canvas 1080×1920, `top_height` ~998. Keep every stacked height EVEN
45	(998 + 4 divider + 918 = 1920) — libx264 rejects odd dimensions.
46	- Anchor = photoreal MEDIUM shot, not a studio headshot. The lip-sync only looks as
47	good as the anchor. It must be photoreal/candid (real lived-in room, natural skin
48	cues), framed chest-up at a natural webcam distance (headroom + shoulders), not a
49	plain-background close-up and not a phone-selfie pose copied from another format
50	(e.g. `ugc-walk-and-talk`). VEED Fabric handles photoreal fine (unlike Seedance).
51	- Blurred-cover fill, never black bars. The top clip's letterbox margins are
52	filled with a darkened blurred cover-scale of the same clip — a flat
53	charcoal/black bar reads cheap.
54	- One claim per scene, shown as it's said. Each top clip is windowed
55	(`top_start`/`top_end`) to the on-message segment that proves its VO line.
56	Never loop a short clip — set the window and the assembler speed-fits it to
57	the scene (looping replays into a sparse/black tail).
58	- **The creator VO is the entire audio bed — no separate musi

People who liked this also viewed

6 picks

Type

render-split-screen-creator

1.0kstars193forks

View on GitHub

How to install render-split-screen-creator?

gooseworks-ai/goose-skills/render-split-screen-creator

$npx -y skills add gooseworks-ai/goose-skills --skill render-split-screen-creator

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# render-split-screen-creator
2
3	Assemble a split-screen creator ad from a config: a two-zone vertical
4	(1080×1920, 9:16, ~40s) format where an AI creator talking-head anchors the
5	BOTTOM ~48% of the frame and real 16:9 product/demo clips run uncropped in
6	the TOP ~52%, a 3px brand-color divider between the zones. The creator delivers
7	the whole VO cold-to-camera and each top clip proves the claim its VO line makes.
8	This capability is the FREE, deterministic assembly + captions — the two-zone
9	composite (contain-fit + blurred-cover fill + divider + creator slice), the
10	hard-concat, the end card, and the word-level caption burn from the assembled cut.
11
12	`scripts/config.example.json` is the worked example (Perplexity concept-10
13	"Bloomberg terminal", ~40s 1080×1920 9:16, 6 scenes + an end card);
14	`scripts/PIPELINE.md` maps every config block to its source step and
15	`scripts/README.md` documents the free assembly.
16
17	## Run
18
19	This is the FREE, deterministic assembly + captions stage — it spends
20	nothing. The paid inputs are separate steps — the VO (`create-vo-elevenlabs`,
21	ElevenLabs `eleven_v3` with-timestamps, sliced into per-scene windows), the
22	photoreal MEDIUM chest-up AI-creator anchor (`create-image-gpt-image-fal`,
23	`gpt-image-2`) — shot at a natural webcam distance (headroom + shoulders, real
24	room), not a plain-background close-up headshot (see the anchor note below), and
25	the whole-VO lip-sync (a paid VEED Fabric 1.0 @ 720p take — a no-atom step,
26	`image_url` = the anchor, `audio_url` = the vo mp3, ~$0.15/sec, ~$5.90 for a 39s
27	VO; run its calls sequentially, `veed/fabric-1.0` storage-auths 403 under
28	parallel load). Given the creator lip-sync take + the per-scene VO timing + one
29	16:9 top clip per scene + the scene-1 hook graphic + the end-card clip,
30	`render-split-screen-creator` composites the two zones, hard-concats the scenes,
31	appends the end card, transcribes the assembled cut, and burns the captions → the
32	master. Re-cuts reuse the existing VO / lip-sync / clips and cost $0.
33
34	## Contract (the free assembly)
35
36	- Two-zone split, top ~52% / creator ~48%. The TOP zone runs the real 16:9
37	product/demo clip contain-fit (uncropped), the BOTTOM zone is the creator
38	lip-sync framed head-to-shoulders: scale-to-width × a small ZOOM (~1.15–1.2)
39	then crop the zone with a downward offset so the face sits upper-middle and the
40	shoulders enter the bottom. (A plain cover + crop-toward-top shows only the head and
41	cuts the shoulders — and it can't rescue an anchor that was shot too close; fix the
42	anchor distance first.) Tune zoom/offset visually against the source video's creator
43	framing — it's a FREE re-assemble, no VEED re-run. A 3px brand-color divider separates
44	the zones. Canvas 1080×1920, `top_height` ~998. Keep every stacked height EVEN
45	(998 + 4 divider + 918 = 1920) — libx264 rejects odd dimensions.
46	- Anchor = photoreal MEDIUM shot, not a studio headshot. The lip-sync only looks as
47	good as the anchor. It must be photoreal/candid (real lived-in room, natural skin
48	cues), framed chest-up at a natural webcam distance (headroom + shoulders), not a
49	plain-background close-up and not a phone-selfie pose copied from another format
50	(e.g. `ugc-walk-and-talk`). VEED Fabric handles photoreal fine (unlike Seedance).
51	- Blurred-cover fill, never black bars. The top clip's letterbox margins are
52	filled with a darkened blurred cover-scale of the same clip — a flat
53	charcoal/black bar reads cheap.
54	- One claim per scene, shown as it's said. Each top clip is windowed
55	(`top_start`/`top_end`) to the on-message segment that proves its VO line.
56	Never loop a short clip — set the window and the assembler speed-fits it to
57	the scene (looping replays into a sparse/black tail).
58	- **The creator VO is the entire audio bed — no separate musi

People who liked this also viewed

6 picks

Type