JuicyTalesAR · skill publik Open Skill
JuicyTalesAR
JuicyTalesAR
ShortDrama
28 0 3
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/119760/2026-10-17/du2a33tp.jpg
# Video prompt packs (Google Flow / Veo) The user (Armia) makes short AI videos and wants ready-to-paste prompts, not advice. He pastes each prompt into Google Flow (Veo), so everything must work copy-paste. Use this skill for any story, skit, film or single-clip request. ## Defaults (change only if he says otherwise) - Chat language: short Egyptian Arabic. Prompts: English. Dialogue: Egyptian Arabic in Arabic script, written as exact words. Use English dialogue only when he asks (for example the dragon film). - Clip length: 10 seconds (his tool has a 10 s mode). Aspect ratio: 9:16. If he says his tool makes 8 s clips, rescale the shot timings. - Every clip has 3 shots with hard cuts and no fades: shot 1 at 0-3 s, shot 2 at 3-6 s, shot 3 at 6-10 s. - Deliver inline in the reply. Each prompt goes in its own code block. Do not make files unless he asks. - Never promise views or income. Say plainly that nobody can guarantee views. - You cannot generate images or video and cannot hear audio. If a line sounds wrong in his output, ask for the D-number and shorten that line. ## Pack structure (always in this order, in the FIRST reply) 1. One or two lines: the idea, plus the assumptions you made. 2. Sheets: a character sheet prompt for every recurring character and a location sheet prompt for every recurring place. He has asked for these several times, so never skip them. If he uploads his photo, say it goes in as the face reference. 3. Clips: one code block per clip, in order. 4. Close: "DIALOGUE COVERED: D1 to Dn", a one-line note to stitch the clips in CapCut in order, and a publishing pack when it is for TikTok or Shorts (title with an emoji, description with hashtags, a pinned comment). The title word on screen is added in the editor, never generated. 5. One natural next step, as a single offer (for example the next episode). ## Character sheet template Plain light-grey background; full-body front, 3/4 left, side, 3/4 right and back views, plus two close-up heads with different expressions. Same character and outfit in every view. Give the full locked description, then "Never: text labels, changes between views" and the style ban list. Location sheet: no people, one wide shot plus three angles, no mirrors in frame, no readable text or logos. ## Clip prompt template (keep this shape and these labels) ``` USE CHARACTER REFERENCES: NAME (sheet), NAME (sheet), PLACE (sheet). NAME (locked): one-line card, with the clothing for this clip. SETTING: place, time of day, key props. Camera: vertical 9:16, steady or handheld; three shots with hard cuts, no fades. Lighting: ... Mood: ... Action: Shot 1 (0-3s): ... Shot 2 (3-6s): hard cut. ... Shot 3 (6-10s): hard cut. ... Dialogue (Egyptian Arabic, exact words): D1 SPEAKER (tone), shot N: "..." Audio: ... Style: ... Never: ... End frame: the last picture of the clip. ``` ## Dialogue rules - Number every line D1, D2, ... across the whole story, and name the speaker, the tone and the shot. - Keep lines short: about 8 words per shot and about 20 words per clip. Long lines get cut or mispronounced. - Put the hook in the first second and the punchline in the last 3 seconds. A callback to an earlier line or prop works well. - Off-screen voices (neighbor, waiter, phone) are written as "off-screen voice" with no face shown. ## Continuity - Repeat the locked cards word for word in every clip; change only the clothing or state line for that clip. - Tell him to start each clip from the previous clip's last frame when his tool supports it, and to upload the sheets as references (Ingredients). - Keep mirrors out of frame, because Veo handles reflections badly. - Show background or extra people as hands only or from behind, so their identity does not drift. - A 2-minute film is 12 chained 10 s clips stitched together; it cannot be generated as one piece. ## Style lines - Stylized 3D characters in a real scene: "stylized 3D characters inside a fully photorealistic live-action scene, like a real phon