How to Make the Most Realistic AI Videos (Step-by-Step Tutorial) - Marketing Against the Grain Recap
Podcast: Marketing Against the Grain
Published: 2026-01-08
Duration: 20 minutes
Summary
Kieran Flanagan outlines a comprehensive guide to creating realistic AI videos using tools like VO 3.1 and Nano Banana Pro, emphasizing the importance of creative concepts and reference images for consistency.
What Happened
Kieran Flanagan spent over 30 hours creating a two-minute AI video ad, learning valuable lessons along the way that he shares in this episode. He developed a four-step process that can reduce the time needed to make a professional-looking AI video to just a few hours. The key tools he used were VO 3.1 for video creation and Nano Banana Pro for generating consistent reference images, which are critical for maintaining visual coherence in AI-generated videos.
Flanagan emphasized the importance of starting with a strong creative concept, something that AI tools alone cannot generate. He shared how AI acts as a thought partner in refining ideas but lacks the ability to originate truly compelling concepts. He demonstrated this by creating a video set in a 1976 sitcom style, featuring ChatGPT as a comically unreliable buddy.
The process begins with creating a storyboard using ChatGPT to plan the scenes, visual descriptions, and dialogues. This step is crucial for setting the tone and ensuring that the video flows logically. Flanagan recommends using storyboards to outline the sequence of scenes and to detail each visual element and dialogue.
Reference images play a vital role in the process, ensuring that all elements within scenes remain consistent. Flanagan used Figma to organize these images, allowing him to maintain continuity in character appearance and setting across different video scenes. He praised Nano Banana Pro for its ability to generate high-quality, consistent images.
Flanagan also highlighted how using reference images before video creation leads to better outputs, as opposed to relying solely on text-to-video processes. He noted that spending time on these images significantly enhances the final video's quality.
Finally, Flanagan shared his experience of editing the video using simple tools like iMovie and CapCut, which are accessible to non-experts. He discussed common challenges, such as copyright issues with names, and how he overcame them. His insights make the process more approachable for those new to video creation.
Key Insights
- Creating a professional-looking AI video can be reduced to a few hours by following a four-step process, utilizing tools like VO 3.1 for video creation and Nano Banana Pro for generating consistent reference images.
- AI tools like ChatGPT can assist in refining creative ideas but lack the capability to originate compelling concepts, emphasizing the need for a strong creative foundation when starting a project.
- Using storyboards to plan scenes, visual descriptions, and dialogues is crucial for maintaining logical flow and setting the tone in AI-generated videos.
- Organizing reference images with tools like Figma ensures visual coherence across video scenes, significantly enhancing the final output quality compared to relying solely on text-to-video processes.