AI-Generated Music Video Using pcode t9rijyq | Music Video Animation

Making an AI music video that does not look like a slideshow of unrelated images is harder than it sounds. Every text-to-image generation is its own roll of the dice, so getting dozens of frames to share one visual identity usually means fighting the tool the whole way. This project took a different approach: lock the style down with a single code and let the lyrics drive the content.

The concept: a modern ballad retelling Mulan

The video sets a retelling of the Mulan legend to a modern ballad, following the arc of a warrior who gives up everything for family and honor. Rather than illustrating the story with live footage or hand-drawn art, every frame was generated from a text-to-image prompt written directly from the song's lyrics. A woman weaving at a loom, a scroll bearing her father's name, a warrior riding beneath the breaking dawn, a homecoming after twelve years away, each lyric became its own still image before the sequence was animated.

Holding one style with pcode t9rijyq

The visual anchor of the project is a special style code, pcode t9rijyq, applied to every prompt to keep the artistic style consistent across dozens of separately generated images. Without something like this, a music video built frame-by-frame from AI stills tends to drift: character faces change, color palettes shift, line weight varies from shot to shot. Using the same pcode across the whole set of prompts kept the frames visually unified, producing the ink drawing aesthetic with a limited color palette that carries through the entire video.

From still images to motion

Once the stills existed, the next step was animation. The project moved the generated images through tools including Midjourney, Runway, Kling, and VEO 3 to bring motion to what started as static frames. This two-stage process, generate a consistent still, then animate it, is a common workaround for the current limits of AI video generation, where getting a long, coherent, stylistically consistent sequence directly from a video model alone is still unreliable. Producing the stills first gives more control over composition and style before handing things off to animation.

The workflow, start to finish

The overall process followed three steps: write text-to-image prompts inspired by the song's lyrics, generate still images from those prompts using the pcode to hold the style, then animate the results. It's a simple pipeline, but it is also a practical template for anyone trying to build a narrative music video from AI tools without a large production budget, write to the lyrics, lock the style, then animate.

Key takeaways

• A dedicated style code applied consistently across every prompt is one of the more effective ways to keep an AI-generated video visually coherent from frame to frame.
• Writing prompts directly from song lyrics is a straightforward way to build a narrative sequence without a traditional storyboard.
• Generating stills first and animating them afterward, using tools like Midjourney, Runway, Kling, or VEO 3, is a practical two-stage workflow for AI music videos.
• The ink drawing aesthetic with a limited palette shows how style constraints can give an AI-generated project a distinct visual identity rather than a generic AI look.

Who this is for

This is useful viewing for anyone experimenting with AI music video production, especially creators trying to solve the consistency problem across a long sequence of generated images. It comes from Humanitarians AI's ongoing experiments in AI-assisted creative production.