What are the key takeaways from this Marketing Against the Grain episode?

Key insights include: **Reference Images Over Text Prompts:** Create detailed reference images in Nana Banana Pro for each character, setting, and prop before generating video clips. This ingredients-to-video approach ensures visual consistency across all eight-second segments, unlike text-to-video which produces inconsistent results.; **Four-Step Production Framework:** Start with a strong original concept, build a storyboard with scene descriptions and dialogue using ChatGPT, generate reference images for each scene in Nana Banana Pro, then create eight-second clips in Veo 3.1 Flow and edit together in iMovie or CapCut.; **AI Amplifies Expertise, Not Ideas:** AI tools accelerate execution for domain experts but cannot generate creative concepts independently. The taste and original idea must come from the creator. AI excels at refining solutions, not identifying what solution to build in the first place.

What did Kieran Flanagan discuss on Marketing Against the Grain?

Kieran Flanagan demonstrates his four-step framework for creating professional AI video ads using Veo 3.1 and Nana Banana Pro, reducing production time from thirty hours to just a few hours through strategic reference image creation. Key topics include: **Reference Images Over Text Prompts:** Create detailed reference images in Nana Banana Pro for each character, setting, and prop before generating video clips. This ingredients-to-video approach ensures visual consistency across all eight-second segments, unlike text-to-video which produces inconsistent results.; **Four-Step Production Framework:** Start with a strong original concept, build a storyboard with scene descriptions and dialogue using ChatGPT, generate reference images for each scene in Nana Banana Pro, then create eight-second clips in Veo 3.1 Flow and edit together in iMovie or CapCut..

How long is this episode of Marketing Against the Grain?

This episode is 20 minutes long. SignalCast provides an AI-generated summary so you can get the key insights in about 3 minutes.

Marketing Against the Grain

How to Make the Most Realistic AI Videos (Step-by-Step Tutorial)

January 8, 2026

20 min episode · 2 min read

Kieran Flanagan

Episode

20 min

Read time

2 min

Topics

Fundraising & VC, Artificial Intelligence, Software Development

AI-Generated Summary

Published Jan 8, 2026

Key Takeaways

✓Reference Images Over Text Prompts: Create detailed reference images in Nana Banana Pro for each character, setting, and prop before generating video clips. This ingredients-to-video approach ensures visual consistency across all eight-second segments, unlike text-to-video which produces inconsistent results.
✓Four-Step Production Framework: Start with a strong original concept, build a storyboard with scene descriptions and dialogue using ChatGPT, generate reference images for each scene in Nana Banana Pro, then create eight-second clips in Veo 3.1 Flow and edit together in iMovie or CapCut.
✓AI Amplifies Expertise, Not Ideas: AI tools accelerate execution for domain experts but cannot generate creative concepts independently. The taste and original idea must come from the creator. AI excels at refining solutions, not identifying what solution to build in the first place.
✓Scene-by-Scene Image Building: Work through each scene systematically, asking ChatGPT what reference images are needed, generating Nana Banana prompts, and feeding previous character images to maintain facial consistency while updating clothing or settings. Total production cost remains under two dollars for complete videos.

What It Covers

Kieran Flanagan demonstrates his four-step framework for creating professional AI video ads using Veo 3.1 and Nana Banana Pro, reducing production time from thirty hours to just a few hours through strategic reference image creation.

Key Questions Answered

•Reference Images Over Text Prompts: Create detailed reference images in Nana Banana Pro for each character, setting, and prop before generating video clips. This ingredients-to-video approach ensures visual consistency across all eight-second segments, unlike text-to-video which produces inconsistent results.
•Four-Step Production Framework: Start with a strong original concept, build a storyboard with scene descriptions and dialogue using ChatGPT, generate reference images for each scene in Nana Banana Pro, then create eight-second clips in Veo 3.1 Flow and edit together in iMovie or CapCut.
•AI Amplifies Expertise, Not Ideas: AI tools accelerate execution for domain experts but cannot generate creative concepts independently. The taste and original idea must come from the creator. AI excels at refining solutions, not identifying what solution to build in the first place.
•Scene-by-Scene Image Building: Work through each scene systematically, asking ChatGPT what reference images are needed, generating Nana Banana prompts, and feeding previous character images to maintain facial consistency while updating clothing or settings. Total production cost remains under two dollars for complete videos.

Notable Moment

The creator spent ten hours on a single scene trying to make both characters speak dialogue simultaneously, eventually discovering the workaround required killing one character off-screen so the AI would allow the remaining character to deliver lines independently.

Know someone who'd find this useful?