How I Make Cinematic AI Ads with Fable 5.2 & Seedance 2.5
Offers practical workflows for solving character consistency and asset realism in AI commercials, detailing specific prompt formulas and model selection strategies.
What this lesson covers
Dan Kieft breaks down his full production pipeline for creating realistic AI commercial ads, emphasizing the importance of input asset quality over simple prompt rewriting. He demonstrates how to construct effective character sheets, select optimal image models, and generate consistent video assets using Higgsfield and other tools.
Key takeaways from the creator
AI-extracted notes, not independently verified product claims. Timestamp links let you check each point in the original video.
- 01:09 ↗
High-quality input assets are essential for realistic AI video output, as generating video from poor or plastic-looking assets yields low-quality results regardless of prompt adjustments.
- 02:16 ↗
Creating a video character sheet on a neutral gray background prevents bright light bounce from carrying over into subsequent video generations.
- 03:30 ↗
Structuring a character sheet with a three-quarter face close-up alongside headless front and back full-body views gives the AI a clear face reference and body proportions without introducing multiple conflicting face references.
- 05:36 ↗
Generating a video character element by linking an asset with a 4 to 30-second audio recording allows AI systems to simulate realistic voice synthesis and appearance.
- 06:18 ↗
Models like Soul Cinema can generate varied, realistic human faces and locations without plastic skin textures at lower credit costs compared to broader diffusion models.
- 08:28 ↗
Defining subtle realistic details—like a slightly crooked cap or tight uniform—rather than broad cartoonish traits prevents AI character models from producing exaggerated or fake expressions.
Workflow outlined in the video
- Create a dedicated project space in your image generation tool to keep character, location, and object assets organized.
- Generate character sheets on a neutral gray background to avoid studio light bleed in video outputs.
- Include a 3/4 view face close-up and headless front/back body views on character sheets to lock in facial geometry and body proportions.
- Upload 4 to 30 seconds of clean vocal audio to construct a video element that models custom voice characteristics.
- Detail specific clothing, age, physical traits, and subtle flaws in character prompts rather than requesting generic AI descriptions.
- Run image generation tasks one batch at a time rather than generating all variants in a single run to ensure maximum output diversity.
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
165,921 views captured 21 November 2026. ReelStack recommendation score: 56/100. Formula: radar-v5. This is ReelStack’s calculation, not a YouTube rating or a measure of factual accuracy.
Show the calculation
- 65% performance: 1.56× views vs 10 other channel videos observed at a similar age. View-evidence factor 99% (views / (views + 1,000))
- 25% freshness: 65/100 with a 14-day half-life
- Momentum pending: collecting daily snapshots; its 20% weight goes to observed performance, not free points
- 10% engagement: 5/100; likes + 4× comments, smoothed with a 500-view neutral prior