Nadji95

F*CKING GENIUS... this 2D cartoon series hit 2.7M views 🤯 the trick: the comments decide what happens next → 46.4K followers with just 12 posts → 308K, 265K and 250K on the next parts every episode ends with a choice, the comments vote, and the next part opens with "69% SAID: CITY" that's why people keep coming back. I wrote a Claude Code + Higgsfield prompt that rebuilds this whole format with your own character steal it below ↓ <role> Act as my animation director, scriptwriter and production engineer. Build a faceless 2D animated "you decide" story series inspired by my attached reference. Use Claude Code, Higgsfield (images, video, voice), Python, HTML/Canvas and FFmpeg. Original characters and stories only, never copy the reference. </role> <reference> Analyze my attached MP4 with FFmpeg first: pacing, palette, characters, camera moves, captions, on-screen graphics and the ending. The style: > dark teal and navy noir with warm orange lamps, red neon and rain > simple characters with big round white faces, tiny eyes, beanies and jackets > detailed backgrounds: highways, motels, cars, back rooms > HUD at the top: cash counter, crew count, heat % > animated maps, evidence cards, scales, counters > one word of each caption highlighted in red </reference> <series> Pitch 5 original series concepts (heist, survival, mystery, money decisions) and wait for my pick. Each episode is 60–90 seconds and ends with a two-option decision. Episode 2+ opens with "XX% SAID: [choice]" based on the comments I paste in, then continues the story from that choice. </series> <script> 0–5s: second-person hook ("You just robbed a bank...") 5–20s: the situation and the stakes 20–45s: first problem, explained with a graphic 45–65s: a twist that changes everything 65–80s: two choices with clear risks Last 3s: "Comment A or B. Part X tomorrow." Short spoken sentences. Every sentence has its own visual. Label anything fictional as fiction. Show me the script and storyboard before generating anything. </script> <visuals> 1080x1920, 30 fps. Generate one character sheet first and reuse it as the reference for every keyframe. Generate each scene as a keyframe in Higgsfield, then animate it with image-to-video: rain, flickering neon, moving headlights, slow push-ins. Never use one still image for a whole scene. </visuals> <graphics> Build every piece of text and every graphic in HTML/Canvas, never inside AI images: HUD counters, maps with moving pins, evidence cards, warning stamps, the decision panel and the "XX% SAID" title card. Animate numbers counting up and cards sliding in. </graphics> <audio> Generate a calm, suspenseful narrator voice. Get word-level timestamps with Whisper and use them to drive the captions and every graphic cue. Cut pauses longer than 0.3s. Add rain ambience and low tension music under the voice. </audio> <delivery> Render with FFmpeg, then check sync, caption spelling and the safe zones for TikTok, Reels and Shorts before giving me the final MP4, the cover frame, and the caption for the post. </delivery>