Google Veo 3: Is it time for full-fledged, end-to-end AI video?

We've all seen countless AI-generated videos go viral on social media. Most are just outdated internet theory at its worst. Of course, there are some truly excellent videos, but until recently, developing them was very expensive in terms of time, tokens, or even the necessary hardware.

 

Google decided to change that. They launched AI products for every major sector, and Veo is their answer to AI-generated video. It's truly impressive.

Interestingly – in a way – some of Veo's training data comes from YouTube. Google doesn't explicitly explain what that means, but you can draw your own conclusions.

In any case, Veo 3 seems like a major breakthrough: Now you can combine audio creation with video creation, in a way that your video doesn't violate the laws of physics.

What is Google Veo?

Google Veo is a group of AI-powered video creation models that can generate videos from text prompts or still images. The latest model, Veo 3, includes the ability to create original audio along with video (the previous model, Veo 2, only created clips without sound).

Veo 3's ability to create original audio, realistic physics simulation, and advanced prompt comprehension are key features that set it apart from other AI video creation tools . Other AI video creation tools, such as Sora and Runway, currently lack original audio creation functionality.

Overview of Google Veo 3

Veo 3 is truly impressive – people have been using it to revamp their marketing strategies. Before diving into how it works and what it can do, here's a quick overview of Veo's strengths and areas for improvement.

 

Advantage

1. Create original audio and video with natural-sounding voiceovers and background noise or music.
2. Realistic physics simulation for elements such as water, fabric, and light.
3. Excellent camera control and shot composition.
4. Understanding advanced prompts, especially interactive signals.
5. Multiple input options (text, images, frames)
6. Integrated into Flow and Gemini, with an intuitive interface (especially in Flow).
7. It has been continuously improved and has surpassed competitors like Runway or Sora.

Disadvantages

1. Length limit: 8 seconds
2. The character's continuity is inconsistent across scenes, even with detailed prompts.
3. Different prompt interpretations make it difficult to generate repeatable output.
4. Text accuracy is limited in visual elements (e.g., spelling errors).
5. Some errors and issues occur when combining footage or switching between modes.
6. Watermarks are displayed unless you pay for the Ultra plan ($249.99/month).

Where can I access Veo 3?

Veo 3 is available on both the Gemini chatbot and Flow, Google's AI video editing app (Flow is easier to use).

It's also more accessible on Flow because Gemini only offers Google AI Pro subscribers 10 trial Veo 3 videos, while Flow offers 100 videos per month with the same subscription plan.

Flow is also specifically designed for video creation with professional tools such as:

1. Control the camera to accurately frame the shot.
2. Scene building capabilities
3. Project management and organization

Here's an example of a video clip with camera controls:

 

Additionally, it's important to note some geographical limitations, specifically that Flow isn't available everywhere. You can only use Veo 2 through Gemini. Furthermore, it's available in over 70 countries, including the US, Canada, Australia, the UK, and India.

Google Veo price

You need to access Google Veo through a broader Google subscription plan, specifically:

1. Google AI Pro ($19.99/month) gives you 1,000 AI credits monthly. For Flow, that's 100 credits for Veo 3 Quality, 20 credits for Veo 3 Fast, and 10 credits for Veo 2 Fast.
2. Google AI Ultra ($249.99/month) gives you 12,500 AI credits monthly, early access to new features, and no watermark display. Notably, this is the only plan that doesn't include a watermark (Pro subscribers aren't entirely happy about this).

Ultra subscribers also get access to Ingredients to Video. This feature allows you to add individual elements (characters, objects, backgrounds) separately and combine them into scenes for better consistency between shots.

How to create your first video with Veo 3

After signing up for the Google AI Pro or Ultra plan, access Flow (recommended) or use the Gemini app.

Here's how to get started using Google Veo in Flow:

1. Click on the prompt field and describe your footage in detail. Include specific details: setting, characters, action, and camera angles.
2. For dialogue, use quotation marks, for example: "The character says 'specific lines'".
3. Add signs of interaction, for example, "looking directly at each other" or "nodding in agreement".
4. Specify audio details, such as background music type, ambient sounds, etc.

After setting everything up to your liking, create the video and wait for the final result, which usually comes within a few minutes.

Here are a few tips:

1. Avoid using capital letters for emphasis. This can be confusing during the sound creation process.
2. Be extremely specific about the interactions between the characters to avoid the situation where the characters just stare at the camera, as mentioned earlier.
3. For sequences with multiple shots, embrace scene changes instead of trying to maintain consistency.
4. Experiment with different variations of the same prompt to find the most effective method.

And a few notes regarding the limitations:

1. Each video is limited to 8 seconds, which severely restricts storytelling potential. You can't develop complex narratives or show detailed processes.
2. Google AI Pro users will see a watermark displayed on all content they create. Only Ultra subscribers ($250/month) can avoid this.
3. When the author provided a frame containing a profit and loss statement, Veo 3 added text errors such as "Expensestes" instead of "Expenses". The artificial intelligence struggled to generate accurate text in scenes. However, OpenAI's early image processing models also had similar issues – ensuring this wouldn't be a long-term problem.

Veo isn't ready for full-fledged projects from start to finish, but the initial ideas and experiments are certainly promising. And testing its imperfections now will help you visualize how to refine and speed up the video creation process in the future as technology advances.