Khari Clemens đź‘‹
Founder of Helexeon, LLC.
Helping artists, entrepreneurs, and innovators harness the power of generative AI for storytelling, branding, and automation.
Project:
Prisoner
Music Video
Overview
Prisoner explores captivity, temptation, and the cost of freedom through a grounded, gritty aesthetic. The video follows a character caught between inner conflict and external restraint—moving from stark confinement visuals to surreal breaks of motion and light. The sound drives the narrative: pulsing rhythms, clipped percussion, and restrained melodies that build pressure until the release.
Concept: Portray a psychological prison—where bars, shadows, and routines become metaphors for obsession and self-imposed limits—then fracture that world with moments of rebellion, memory, and desire.
Planning: We storyboarded tight, repeatable motifs (cell, corridor, interrogation, mirror) and contrasted them with abstract sequences (slow-shutter movement, staccato edits, silhouette choreography). Prompts and look-dev were iterated using AI to keep character styling and lighting consistent across shots.
Tools Used: A multi-tool workflow to keep character, lighting, and motion cohesive:
- ChatGPT (Story, shot list, prompt iteration)
- Midjourney / Kling AI / VEO (Image & video generation)
- HeyGen (Lip sync & facial performance)
- Adobe Photoshop (Look-dev, frame paint, plates)
- Adobe Premiere Pro (Edit, timing, rhythm)
- HiggsField / After Effects (Transitions, blends)
- Apple Logic Pro (Production, sound design)
- Suno (Vocal stems / alt takes)
- ElevenLabs (VO textures / whispers)
Challenges
The creative focus was realism with symbolic abstraction: consistent characters and lighting across multiple AI-assisted scenes, lip sync to performance timing, and transitions that feel intentional rather than stitched.
Character Consistency
- Challenge: Keep the lead’s face, wardrobe, and hair consistent across cells, corridors, and dream beats.
- Solution: OmniReference + prompt locking, palette constraints, and reference-guided frames; manual paint-backs for tough cuts.
Lip Sync & Performance
- Challenge: Align mouth shapes and micro-expressions to the vocal while maintaining the video’s tense pacing.
- Solution: HeyGen-based sync passes, then editorial nudging on syllable peaks; added breath/beat holds to sell realism.
Transitions & Visual Rhythm
- Challenge: Avoid “AI montage” feel with disjointed clips; create narrative flow.
- Solution: Designed in/out anchor frames per shot, match-action cuts, light-wipe motifs, and corridor match moves to thread sequences.
Lighting & Mood Continuity
- Challenge: Maintain the same key/fill direction and color temperature across generated scenes.
- Solution: Locked LUT targets and prompt lighting descriptors, then graded plates to a unified contrast curve.
Results/Conclusion:
Prisoner landed with strong emotional clarity—tight lip sync, coherent character continuity, and purpose-built transitions that support the story rather than distract from it. View it now on YouTube (link above) or via the embedded player.
