Summary
This beginner guide from Dan Kieft explains how to start making AI videos in 2026, from choosing a platform to generating a first clip while keeping credit costs down. He recommends using a platform that offers multiple video and image models, since the best model changes constantly, and demonstrates the workflow on Higgsfield, which sponsored the video.
The core method is reference-to-video rather than plain text-to-video. The workflow runs from ideation to asset creation, then video generation, then editing and assembly. Dan shows how to build character sheets, location sheets, props, and outfits, including a cheap route through Cinema Studio and a template-plus-LLM prompt for more control. His key cost advice is to generate many images, which are very cheap, so that video generations, which are expensive, start with the right ingredients.
He compares image outputs across several models, including Nano Banana 2.1, and names the video models he currently favors, such as Seedance 2.5, Kling 4, MiniMax Hailuo 3 Max, and Flux 3. He also shares a structured video prompt format built from a goal, character references, scene description, and shot-by-shot stages, so viewers can reproduce consistent results.
📺 Source: Dan Kieft · Published October 22, 2026
🏷️ Format: Tutorial Demo
