The Cure

Workflow. My workflow centered on Grok @imagine Agent Mode’s node-based canvas workspace for image, video, and text generation, iteration, visualization, organization, and scene development. For video generation, I used Grok Imagine exclusively on this project to test its capabilities. I was impressed by @Grok’s ability to create videos that lip-sync on beat to my wife’s performance (tiktok.com/@cbearclarke). There were some tricks I learned to make this work every time I needed it.

I was inspired in part by the recent music video from Heavy Pulp @heavypulp. After watching the breakdown of Heavy Pulp’s process posted by @xAI, I was surprised by how similar many parts of our workflows ended up being. One difference is that I believe Heavy Pulp’s production utilized fully AI-generated assets. I also incorporated a lot of live-action footage and the likeness of a real person captured on a smartphone. I used that footage directly in the final video, and used the captured likeness as a reference for AI video. I created AI-generated content to merge the real footage into the world of the story, and used Adobe Premiere Pro’s AI-powered Object Mask tool to isolate live-action elements and blend them with AI-generated assets. Adobe Audition helped me edit the audio, also recorded on a smartphone.

For image generation and editing for the stills, I used Grok Imagine, ChatGPT, Photoshop, and Adobe Firefly. I worked with LLMs for storyboarding, shot planning, prompt refinement, stylistic consulting, continuity management, scene iteration, and creative decision-making. I provided the models with story outlines, worldbuilding, character references, visual references, and song timing and transcript exported from Adobe Premiere. Adobe Premiere Pro and Adobe After Effects were used for editing, compositing, visual effects, pacing, and final delivery.

Why this is not just “Made with AI.” I think it is wrong to frame this process as “this video was made with AI.” It is more interesting than that. I used all of my creative and technical skills to make this, as I would any project. The difference is that, thanks primarily to Grok, I was able to make this alone in less than a day. In the past, a project like this would have required a team of people and much more time.

I adapted my story into a visual language, designed shots, maintained character and setting consistency across scenes, preserved facial likenesses, unique appearances, tattoos, clothing, locations, and performances. I directed performance through engineered text and image prompts, managed continuity, iterated on hundreds of creative decisions, and shaped the final emotional arc through editing and post-production.