creating consistent characters on VEO3 sucks

@julian.stancioff.ai

10.2K followers · usually 1.9K views a post

Views

24.4K

Likes

962

Comments

360

Shares

241

SoundTikTok sound · original sound - julian.stancioff.ai

Promoted product

Veodeepmind.google

Google Veo

10 videos4 creators

Why it worked

The video provides a high-value, actionable 'hack' for a common pain point in a trending technology (AI video generation), which encourages high engagement through comments.

Summary

The creator demonstrates a method for maintaining character consistency in AI-generated videos using Google's Veo tool. He explains a workflow involving screenshotting a character and using a specific prompt to ensure the AI recreates the same subject in subsequent clips.

Structure

1Identify the problem of character consistency in AI video
2Show proof of successful character recreation
3Explain the step-by-step process of screenshotting and prompting
4

Original caption

So I also tried this in 2 other ways: I tried this by prompting the ai with a machine-optimized text prompt and then I tried with the prompt in Python. They both worked well but neither worked better than the regular text prompt. #ai #veo3

More from @julian.stancioff.ai

Product placement

Google Veo (VEO3) is the primary subject of the tutorial.

Call to action

follow me and comment the letters VEO and I'll DM it to you

On-screen text

creating consistent characters on VEO3 sucks VEO3 CONSISTENCY HACK FULL GUIDE+ PROMPT AT THE END

Transcript

creating consistent characters on VEO3 sucks. It's really really hard, but if you're looking for your clips to be longer than eight seconds, you have to do it. And this is the one way I found that actually works. And here's proof. I made the same character multiple times in a row. Now the way to do this is actually pretty simple. So you start with whatever your starting prompt is, doesn't matter what it is. From there you go in like this, you screenshot the character. Next thing you do is add the screenshot, then you give it this prompt. I'm going to give you a screenshot of the Bigfoot character and you're going to describe him in super vivid detail so that we can recreate the exact same scene again. And from there you include your last prompt and then you tell it not to overdo it, keep it simple. Alright, and here is the output again. This was the original input and then this is the output. Super similar, close enough. If anyone wants my full step by step guide, follow me and comment the letters VEO and I'll DM it to you.