AFTERDARK — Cinematic Sensual Drama Studio · публичный навык Open Skill
AFTERDARK — Cinematic Sensual Drama Studio
AFTERDARK — Cinematic Sensual Drama Studio — публичный Open Skill на Digen. Откройте, чтобы увидеть, что создаёт этот процесс.
ShortDrama
124 13 73
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/3677592/2026-10-01/uzy27nst.png
# AFTERDARK — Cinematic Sensual Drama Studio Premium vertical mini-series for adult audiences: **beautiful + seductive + emotionally intense + visually cinematic + highly addictive**. Sensuality is strong but the story stays a drama, never pornography. All romantic/sensual characters are clearly consenting adults aged **21+**. ## Reference files - `references/creative-direction.md`: the full AFTERDARK creative spec (story types, visual standard, camera language, lighting, color palette, locations, character design, wardrobe, sensuality engine, kiss direction, physical chemistry, semi-nude cinematic language, intimacy transitions, dialogue style, hook/retention/cliffhanger engines, sound design, music style, output format, consistency rules, content boundaries). **Read it before writing any story or shot.** ## When to use - User wants a new sensual adult mini-series / microdrama episode from a one-liner, idea, or existing characters. - User wants the next episode, a rewrite, a single-shot fix, or a restyle that must preserve the approved characters/look. - User wants a full series concept (title, logline, characters, season arc, episode structure) generated automatically from one sentence. ## Steps 1. **Plan.** Use `write_todos` to list the stages (story → character refs → storyboard → clips → voice → music → mix) and update it as you progress. 2. **Recall continuity.** Call `memory_query` (kind `episode`/`project`) for any existing AFTERDARK series the user has worked on — characters, relationships, previous kisses/betrayals/secrets, wardrobe, locations, unresolved cliffhangers, timeline. If nothing is found, start fresh. 3. **Read the creative spec.** Call `read_skill_file(name="afterdark", path="references/creative-direction.md")` and follow it for story, shot, and visual direction. 4. **Story generation.** Build the complete concept (series title, viral hook, logline, genre, visual style, main characters with full profiles, season premise, episode title, shot-by-shot list with camera/lens/location/lighting/action/dialogue/sound/music/emotional intention, cliffhanger, next-episode tease). Save it with `create_document(content=..., name="<Series> — Episode <N> script", content_type="text/markdown", entity_type="scripts")`. If the user gave only one sentence, make strong creative decisions automatically — don't force them to fill gaps. 5. **Character reference images.** For each main character, generate a consistent reference with `generate_image(model='t2i', orientation='portrait', resolution='720P', prompt=<full character sheet: face, hair, age, body, wardrobe identity, style>, name="<Character> ref", entity_type="characters")`. Use bare `t2i`/`i2i` (not `.hd`) for adult/sensual content. Keep faces, hair, age, skin tone, body type, tattoos, jewelry, and wardrobe identity fixed across the whole series. 6. **Storyboard images.** Generate one image per shot with `generate_image(model='i2i', input_urls=[<character ref(s)>], orientation='portrait', prompt=<shot: camera, framing, location, lighting, action, physical interaction>, name="Shot <NN>", entity_type="storyboard_images")`. Pass the relevant character refs to keep identity consistent. Follow the camera/lighting/color guidance in the spec. 7. **Storyboard videos.** Turn each storyboard image into a short clip with `generate_video(model='i2v', input_urls=[<shot image>], orientation='portrait', duration=~5–10, prompt=<motion: push-in, dolly, kiss escalation, hand touch, etc.>, name="Clip <NN>", entity_type="storyboard_videos")`. **Video generation does not support NSFW/adult content on any channel** — keep clips to kissing, caressing, silhouettes, implied intimacy, and fade-to-black; never explicit acts. For stronger identity consistency you may use `model='ref2v'` with character refs instead. 8. **Voiceover.** Generate dialogue with `generate_tts(model='tts.design', text=<line in the user's request language — Spanish by default>, voice_instruction=<description of