I Made AI Videos That Got 6M Views & Started a Trend — Here's the Exact Workflow

What this lesson covers

Karima outlines her viral video generation pipeline using character consistency sheets, environment setting, and multi-asset video prompting. She demonstrates feeding personal selfies and prop reference photos to Claude to generate multi-angle character sheets before generating cinematic shots using Seedance 2.5 in Higgsfield AI.

Key takeaways from the creator

AI-extracted notes, not independently verified product claims. Timestamp links let you check each point in the original video.

1. 01:14 ↗

   Combine multi-angle personal selfies with Pinterest outfit references in Claude to generate a detailed multi-angle character sheet prompt.

2. 02:36 ↗

   Create dedicated prop reference sheets for key objects held by characters to maintain item consistency across video sequences.

3. 05:16 ↗

   Write video prompts from a camera operator perspective, specifying starting elevation, camera movement, timing, and framing shifts.

4. 05:59 ↗

   Add explicit physics instructions in your prompt to define weight, momentum, and realistic body mechanics during character actions.

5. 06:46 ↗

   Use Seedance 2.5 in Higgsfield AI for multi-asset video generation requiring consistent facial identity, camera trajectories, and embedded audio effects.

6. 07:33 ↗

   Render initial prompt tests at 480p resolution and run single generations first to check action physics before spending credits on 720p or batch runs.

Workflow outlined in the video

1. Take clear, well-lit personal selfies from multiple angles and gather outfit inspiration photos.
2. Upload selfie and outfit references into Claude to generate a comprehensive character sheet prompt with positive rules.
3. Generate character turnarounds and key prop sheets using an image generator in 16:9 aspect ratio.
4. Feed the approved character, prop, and location assets back into Claude to draft a shot script with explicit camera motion and physics cues.
5. Load assets into Higgsfield AI using Seedance 2.5, run a low-resolution test generation, tweak prompt physics if needed, and produce final renders.

Before you use this workflow

These notes describe the source video at its publication date. Model access, pricing, connectors and interfaces may have changed. Check the original source and the provider’s current documentation before installing an add-on, connecting an account or spending credits.

ReelStack has not independently tested this workflow. Preview one representative shot and check motion, continuity and output quality before applying it to a full production. No result, cost saving or model capability is guaranteed.

Explore the tools

How to interpret the numbers

151,698 views captured 27 September 2026. ReelStack recommendation score: 54/100. Formula: radar-v5. This is ReelStack’s calculation, not a YouTube rating or a measure of factual accuracy.

Read our methodology and limitations →