The Music Video Problem
A typical music video has 60–120 cuts. The artist appears in multiple outfits, locations, and lighting conditions across different shooting days. Then you take all of that footage and try to impose a single, cohesive AI-stylized aesthetic on it.
Without consistency tools, the result looks like a collage. The artist looks like three different people. The palette shifts with every location change. Viewers perceive the inconsistency before they consciously register why the video feels "off."
Building the Style Pack First
The most important step in a music video stylization workflow is building the Style Pack before touching individual clips. Identify 3–4 clips that represent the range of lighting conditions in your footage — a bright outdoor shot, a dark indoor shot, and a mid-range shot. Run the analysis pipeline against these clips.
The analysis extracts the dominant palette, and the system suggests a prompt fragment. Review the palette, adjust if the automatic extraction picked up outlier colors, and finalize your prompt fragment. For music videos, this might be: "neon-lit, high contrast, saturated, music video aesthetic" or "bleach bypass, analog, gritty, 16mm."
This one Style Pack now travels with every clip in your project.
Character Cards for the Artist
Create a Character Card for your primary artist using 3–5 reference photos from the shoot. Choose photos that cover:
- Different angles (front-facing, 3/4 profile)
- Different expressions (neutral, performing, emotional)
- The primary wardrobe look for the video
This reference set gives the IP-Adapter enough variation to represent the artist's identity robustly rather than over-indexing on a single angle or expression.
The Render Queue Workflow
Once your Style Pack and Character Card are set up, the workflow is straightforward:
- Export clips from your Resolve timeline as ProRes or DNxHR (lossless source quality matters for stylization)
- Add clips to the ShotLock Queue with your Style Pack and Character Card selected
- Enable batch consistency mode for clips from the same scene — this locks the seed across the batch so adjacent clips share the same stochastic starting point
- Submit the queue and let it run
- ShotLock auto-imports the processed clips back into Resolve's media pool
Handling Multiple Outfits and Looks
Music videos often feature multiple looks. Create a Character Card per distinct wardrobe — the artist in the club outfit, the artist in the desert scene. Apply the relevant Character Card when queueing clips from that look. The Style Pack stays consistent; the Character Card switches with the look.
What to Expect
Expect a significant reduction in visible drift across the timeline — not perfect frame-by-frame freeze-frame consistency, but a coherent visual identity that holds across cuts. Higher character lock strength (0.85+) gives tighter identity but less stylistic freedom. For most music video work, 0.75–0.80 is the sweet spot.