AI-Cut Studio

💬 Studio — analyze, ask, refine

⚡ One-shot — just make it

🎛 Editor — place it yourself

1 · Source video(s)

Multiple videos = multi-source cut (interviews + b-roll, etc). Merge 2–3 by adding them all.

2 · Your idea

In Studio mode this seeds the questions — you'll refine your vision before it builds.

Custom directives (you choose any part — AI fills only the rest)

Free text → typed slots (style, music, scene kinds, forced moments). Whatever you set here wins; the result shows who decided what.

3 · Style preset

AI brain

Editing brain If the AI author is busy (Gemini overloaded)

Add b-roll cutaways · the AI matches them to what you say

B-roll is placed automatically onto real transcript moments (the AI picks what, the engine picks where + full-screen vs PIP to keep your face when you're the point).

Captions

Put captions on the video

Caption look

★ the default. One choice drives both caption engines, so what you pick is what renders whether the kinetic engine is on or off.

Preview (the real renderer)

Add a video above, then hit “Render preview” to see the look burned over your own footage.

Position on screen

Words per line (0 = auto: 3 on vertical)

Emphasised keywords per line (0 = every keyword)

Kinetic engine (Remotion) — smoother, much slower to render

Off = burned in with ffmpeg (fast, what Auto uses). On = a motion-graphics overlay. Your caption look above applies either way; this only changes which engine draws it.

Kinetic motion override (advanced — leave on “match”)

4 · Format & output

Aspect

Resolution

Output

Draft mode — fast 720p preview (great for refine loops)

5 · Coverage & color

Visual style — ONE look threads the whole video (b-roll, scenes, grade, music vibe) Keep my footage

Your whole video is the base timeline — the AI only adds b-roll, captions and titles on top. It never drops your speech.

Color look

Cinematic LUT grade

6 · Editing extras

Whoosh SFX on cuts

Veo AI video b-roll on hero beats (PAID ~$1/clip)

B-roll is always AI-generated (Gemini images). This adds a real Veo video clip on hero beats instead of a still — metered, one clip per hero beat up to the cap; keep the cap small.

YouTube outro (subscribe + follow card)

Appends a branded end-screen from brand.json (channel name, handle, avatar, colors).

Max b-roll inserts (0 = no cap)

7 · Audio

Music bed (CC track, ducked under voice)

Auto-finds a Creative-Commons track for the video's vibe, levels it under the voice, and ducks it while you speak. Attribution lands in the result notes.

Denoise

Voice enhance (speech)

Target loudness (LUFS)

8 · Studio look

Animated keyword stickers over the talking head

Context-triggered emoji graphics spring in on keywords (the "visual elements without b-roll" look), each with a synced pop.

Max stickers per video (0 = off) Hook guard (seconds) (no b-roll/scene before this; 0 = off) Max element hold (seconds) (0 = default) Default element hold (seconds) (0 = default) Music bed level under voice (LUFS offset) (more negative = quieter bed; 0 = preset) SFX intensity (1 = normal; lower = subtler under VO) SFX min gap (seconds) (restraint — thin dense whooshes; 0 = every cut)

Zoom-punch segments (punch-ins)

SFX on element exits (swoosh-out)

Riser into hero reveals

Subject cutout on a designed background (needs the RVM install)

Mograph clips via Revideo (needs the revideo install)

RENDER

Ready. Add a video, describe your idea, then analyze.

1 · Analyze

2 · Vision

3 · Build

4 · Refine