Vector & Watercolor Story Studio · publiczny skill Open Skill
Vector & Watercolor Story Studio
This will create a 45 second - 2 minute episode, (the length is set by how many shots the story actually needs rather than a fixed number.) Just select continue the series when you want more episodes. Highlights: 1. It locks in skin tones so your characters won't drift to other colors half way through your project. 2. It lets you select one voice to be used for the narration if you have more then one character each character who talks gets their own locked voice 3. It mutes the audio baked into each clip so it can't clash with the narration layered in afterward 4. Captions are generated for the finished episode Why your time and credits aren't wasted: Nothing past checkpoint 1 gets built on reference art you haven't approved, and nothing past checkpoint 2 gets built on keyframes you haven't approved - so you're never paying to redo downstream work because something upstream was wrong. And if you only want to change one thing afterward - a single keyframe, a clip, the voiceover, the music, a caption - it redoes just that piece and reassembles around it, instead of regenerating the whole episode from scratch. Here's the complete breakdown... Give it one photo and a one-line idea, and you can "continue the series, next episode." to create a series. That one photo does double duty: it becomes your character/scene/prop reference art (Checkpoint 1), and it anchors the same look across every keyframe built after it, one keyframe image per shot of the episode, usually 3 to 6 shots depending on how the story's paced. Each approved keyframe is turned into its own short video clip, and all of those clips are stitched together into one finished, captioned episode, typically landing around 1–2 minutes total by default, since the length is set by how many shots the story actually needs rather than a fixed number. Here's exactly what happens, step by step, and where you get a say before anything moves forward. 1. Idea & style lock You share your reference photo and your one-line premise (or point it at an existing series to continue). Pick a visual style - high-contrast vector illustration, Afrocentric watercolor-painterly, or torn-paper collage. For an ongoing series the style and characters are already locked in from before, so this step is skipped and it just picks up where you left off. 2. The script gets written This isn't a bare shot list - it writes real narration and dialogue for the episode (unless you specifically want it silent), sized to a short vertical episode told in a handful of shots. 3. Reference art - Checkpoint 1 Your uploaded photo becomes the visual anchor for that character: it generates character, location, and prop reference art that matches your photo's actual style and features - it won't push a stylized or faceless reference toward realism, and it won't invent a face that isn't in your photo. You see every piece of reference art together and approve it - or flag just the ones that need a fix - before a single frame of the actual episode gets generated. 4. Scene keyframes - Checkpoint 2 Once references are approved, it generates one key still image per shot, then runs its own consistency pass across the whole set - same character face, same outfit, same visual style, same recurring props, shot to shot. You see the full batch together and approve it (or send back just the shots that need fixing) before any audio or video generation starts. 5. Voiceover & music The first time a character - or the narrator - speaks, you're given a few different voice options to choose from (different takes on age, tone, and pacing). Whichever one you pick becomes that character's locked voice for the rest of the episode, and every future episode in the series, so it never drifts or suddenly sounds like someone else mid-scene. If the episode calls for music, a bed is generated or chosen to match the mood and visual style. 6. Video clips - Checkpoint 3 Using each approved keyframe as its anchor image, a video clip is generated for every shot, then the whole set is shown to you together before anything else happens. Flag what's off and only those specific clips get redone. 7. Final mix This is where it comes together: any audio baked into the raw video clips is muted first, so it can't clash with anything. The clips are stitched into one continuous episode in order, your voiceover is layered in at full, clear volume, and the music bed is kept low underneath it - quiet enough that it never competes with or buries the words, never just dumped in at a random level for you to catch after the fact. 8. Captions Once the mix is locked in, captions are generated for the finished episode. Why your time and credits aren't wasted Nothing past checkpoint 1 gets built on reference art you haven't approved, and nothing past checkpoint 2 gets built on keyframes you haven't approved - so you're never paying to redo downstream work because something upstream was wrong. And if you only want to change one thing afterward - a single keyframe, a clip, the voiceover, the music, a caption - it redoes just that piece and reassembles around it, instead of regenerating the whole episode from scratch.
ShortDrama
18 2 9
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/925535/2026-09-22/fjx6kxpk.png
--- name: vector_watercolor_story_studio description: Use when the user wants to make a short vertical animated episode - from a one-line idea to a finished piece: series bible and script, character/scene/prop reference art in her signature visual styles (high-contrast vector, Afrocentric watercolor, torn-paper collage), scene keyframes, voiceover, video clips, a stitched final cut, and captions. Locks skin tone and character identity across every scene and episode - no drift. Built for characters of any shade of Brown, fantasy, sci-fi, Afrofuturist, Afrocentric, chibi/brat serialized shorts, or any new series idea. Also use to continue an existing series (e.g. Ice Princess on Wheels, Sonny Cub), start a brand-new series, revise a script, swap a character's look, redo a keyframe or clip, or regenerate audio/captions. --- # Vector & Watercolor Story Studio Turn a one-line idea (or an uploaded reference image plus a storyline) into a finished short vertical episode, moving through a fixed pipeline with three standing approval checkpoints. Built around Luna's / Pixel & Pose's own production habits, not a generic drama-short template. ## Non-negotiables (any miss is a violation) - Visual style must be locked before any generation call. See `references/03-world-style.md`. Never blend two styles in one project unless the user explicitly asks for a crossover. - Every image/video generation prompt ends with the locked style's style-lock phrase, unless reference-image fidelity (below) overrides it for that asset. - Every generation call includes the standard negative prompt unless the locked style is explicitly 3D/CGI: "NOT 3D rendered, NOT CGI, NOT Pixar style". - Prompts are structured as **Subject + Scene + Motion**, camera movement named explicitly, motion sourced from the subject rather than the camera wherever possible. - Default orientation is 9:16 vertical portrait. Only switch to landscape if the user explicitly asks. - No on-image text/lettering baked into generated art unless a shot specifically calls for it. - The text shot list (`references/05-storyboard.md`) is internal planning only - never shown to the user as a multi-panel contact-sheet/grid image. Each shot's keyframe (`references/06-keyframes.md`) is its own standalone image, generated and reviewed individually as a set - never merged into one grid image. - For an existing series, pull established facts from `references/03-world-style.md` instead of re-inventing them. For a brand-new series, run full intake (`references/01-intake.md`) instead. - **Reference image fidelity**: if the user uploads or points to a reference image for a character/scene/prop, that image's exact style, medium, and facial treatment overrides the general style-lock description - never push a stylized/illustrated/silhouette reference toward photorealism, and never invent facial features (eyes/nose/mouth) on a character whose reference has none. Applies to reference art, keyframes, and video alike. See `references/03-world-style.md`. - **Actually write the story.** Even when the user dictates a full rundown of shots/beats themselves, that's the raw pitch, not the finished episode - write real narration/dialogue for it (unless the user explicitly wants it silent). See `references/02-story.md`. - **Never auto-insert the creator's own visual signature** (silver-white hair, rectangular glasses, gold hoops) as a stand-in/avatar character in an episode. It's opt-in only - use it only when the user explicitly asks for a version of herself in that piece. - The user's uploaded art is used **as-is** as the reference for any character it already shows - never regenerated into a character card. See `references/04-assets.md`. - Each project's look is captured in **one rich sentence** (medium, palette, motifs) and repeated verbatim at the end of every generation prompt. Keep prompts short: padding a prompt with paragraphs gives the image model less to hold onto, not more. - No video channel has a negativ