The Impact of Google Veo 3: How Will Super-Evolved Video Generation AI Transform SNS Video Strategies?

Generative AI is evolving at an extremely rapid pace and has been actively utilized in the business arena in recent years.

Video generation AI, in particular, is making a significant impact in advertising and marketing due to its high production efficiency, limitless creativity, and remarkable realism that can be mistaken for the real world.

This time, we focused on Google's video generation AI model, "Google Veo 3" (hereafter Veo 3), and the potential for short video initiatives.

Toshihiro Iyoda and Sohei Mochihara from Dentsu Inc., who oversee Toyota Motor Corporation's (hereinafter Toyota) global SNS initiatives, discussed the potential for utilizing video generation AI in the marketing domain with Randy Han, who supports Toyota as part of Google's global sales team.

Journey on a whole different scale.
https://www.youtube.com/shorts/ln3UAkOs9c8

 

What rumbles but never crumbles?
https://www.youtube.com/shorts/rhPSf1GJ7Ew

Can You Even Tell It's AI-Generated? 3 Evolutionary Points of Google Veo 3

Iyoda: Mochihara and I currently manage Toyota's global SNS operations and content production. We faced challenges in improving the production efficiency of our 150 short videos created and posted annually, and in keeping pace with rapidly changing SNS trends.

Amidst this, Toyota and Dentsu Inc. participated in a Google-hosted workshop in May 2025. There, we had the opportunity to trial Veo 3, which had just launched in Japan that same month. We were strongly impressed by its high performance and potential within the social media marketing domain.

Mochihara: Regarding this workshop, I was so impressed by footage created with the previous model, Veo 2, that I reached out to Randy to make it happen. Incidentally, that footage was created by a creator named Laszlo Gaal. It was truly an impactful piece that generated significant buzz.

Randy: Originally, the plan was for a "Veo 2 workshop," right? But then Veo 3 was released just three days before the workshop was scheduled to happen (laughs). As a result, we were able to give participants early access to Veo 3.

Google Veo
https://gemini.google/overview/video-generation/?hl=ja

Google Veo is a video generation AI model announced in May 2024. Veo 2 followed about six months later in December 2024, and Veo 3 was announced in May 2025. They're upgrading versions at a very rapid pace, roughly every six months.

Iyoda: That really highlights the speed of AI evolution. So, what aspects of Veo 3 have evolved compared to Veo 2?

Randy: There are three major improvements: "video generation that reproduces object motion," "handling complex effects," and "enhanced accessibility."

First, regarding video generation that replicates object motion: Veo 3 can now generate more natural videos by reproducing physical properties without needing detailed prompt instructions. For example, it can simulate "real-world physical laws" like gravity. It can easily generate scenes like water flowing down or splashing on the ground. Also, if you prompt it with "a car driving," it generates movement that accurately portrays a car driving correctly, not like a human running. The output to prompts has become more natural overall.

Second, it enables complex cinematography that expands creators' creative scope. By reproducing object motion, advanced camera work like drone shots and panoramic shots is now achievable, adding features that creators find easier to use.

Finally, accessibility has improved. While Veo 2 was a standalone tool, Veo 3 can now be accessed from various Google services. Integration with platforms like Gemini and Google Photos (currently US only) has begun. The goal is to make it a tool for expressing free ideas, accessible to more users.

Mochihara: At the workshop, I was shocked by the high quality of the demo videos created by the creators. Even at the Veo 2 stage, I thought the realism was impressive, but Veo 3 is now almost indistinguishable from actual footage. I was amazed that "just typing a prompt can create video of this level."

Iyoda: What caught my attention was the simultaneous generation of video and audio. I was particularly amazed by the high accuracy of the lip-syncing—how people's mouths move perfectly in sync with the audio. The consistency between the spoken content and the person's movements is flawless, and even Japanese dialects can be reproduced with a single prompt.

Considering its use in Toyota's video initiatives, the high fidelity in reproducing car engine sounds and driving noises is also welcome. While there are still occasional minor inconsistencies in detail upon close inspection, it can produce perfectly adequate footage for static car shots or wide-angle cuts where fine details aren't visible. I believe Veo 3 has already reached a level where companies can effectively utilize it for social media.

Toyota has already started producing videos using Veo 3 for its global social media content, and we are currently showcasing prototypes. The two videos introduced at the beginning feature the Toyota Land Cruiser 300 as the star: one presents a unique world where it races through a wilderness made of cookies, while the other offers a realistic, powerful visual experience evoking the energy and grandeur of nature. Both were primarily created by professional video creators using Veo 3. and while they are already reaching a practical level from a marketing perspective, they also hint at further evolution.

When integrated with other products, it enables easier creation of higher-quality visuals.

Iyoda: Video generation AI is evolving rapidly, but there's also the challenge of how we utilize it. Mochihara-san, from a production perspective, are there any tips for creating good visuals?

Mochihara: I think it really comes down to "how accurately you can write the prompt." That said, even I sometimes struggle to verbalize the images I have in my mind. For example, the term "drone shot" isn't something someone outside professional video production would easily think of (laughs). No matter how much the tools evolve, the output will still vary depending on the user.

In those cases, I sometimes use other generative AI tools like Gemini, a chat-based AI assistant, to generate prompts for Veo 3. Furthermore, Veo 3 isn't just about "Text to Video" where you give instructions via text prompts. Its "Image to Video" and "Video to Video" capabilities, which accurately reproduce subjects by feeding in other images or videos, are also highly effective. We leverage other Google products for video generation too. This approach makes video creation relatively easy even for those without production experience.

Google's creative ecosystem is particularly appealing. With Veo and Gemini, and the ability to generate videos from Google Photos images (currently only in the US), it offers a well-equipped production environment leveraging a rich suite of products.

Randy: Production processes like "Text to Video" and "Image to Video" are entirely new methods born in the era of generative AI. Consequently, even within Google, the "Veo Usage Guidelines" are frequently updated. We're currently accumulating production know-how, like "writing prompts this way achieves that result."

Generative AI evolves so rapidly that if Veo 4 or Veo 5 were announced later, it's unclear whether the guidelines and know-how we've built up now would still be applicable (laughs).

Mochihara: It's unusual to create guidelines for a product you've built yourselves while actually using it, isn't it? I think that's both the challenge and the appeal of generative AI. And in terms of its evolutionary direction, you're aiming for a future where guidelines become unnecessary, right?

Randy: Yes, we're evolving with the goal of "extending human creativity," making it easier to maximize human creativity. Regarding Veo's prompt guidelines, we regularly publish them on the official Google Cloud blog. If you're interested, please take a look.

https://deepmind.google/models/veo/prompt-guide/ 
https://blog.youtube/news-and-events/made-on-youtube-2025/

SNS Strategy: Keywords for the Future are "Volume" and "Video Diversity"

Iyoda: We've discussed the current state of video generation AI. Now, let's consider how companies can leverage video generation AI and integrate it into their SNS strategies.

With video generation AI entering the business realm, I imagine it will also change posting strategies for corporate social media marketing. What are your thoughts?

Randy: First, regarding production, video generation AI will solve various constraints like budget and time, significantly expanding the number and range of proposals.

Regarding the difference from traditional video production, a creator who prototyped Toyota videos using Veo 3 said:

"It allows us to focus purely on 'how to present it' and 'how to convey the message,' without worrying about constraints like location, time, or weather."

With budgetary and time constraints removed, we can expect even more diverse visual expressions than before.

Iyoda: I also believe "diversity in video" will be a key concept. AI makes it easy to create things that were previously impossible, clearing away all constraints in the production process. Working in such an environment makes it easier for ideas to emerge that make you think, "I never would have thought of that." In other words, precisely because "anyone can create," the value lies in the unique sensibilities that make something "only that person could create."

By combining the unique sensibilities of such creators with social media trends, and having video generation AI support the production process, I feel truly great things can be created.

Furthermore, as video generation AI shortens the production process, it will become possible to prioritize planning and generate more ideas. In that context, how to select and manage vast numbers of ideas is an area we need to consider going forward. To name one example, in the SNS domain where speed is crucial, I believe we should also aim to simplify the "process of proposing plans to clients."

Mochihara: For SNS, short videos will undoubtedly become increasingly important. As the consumer behavior of "watching many short videos" becomes widespread, companies are also shifting towards "producing many short videos."

For Toyota's global SNS initiatives we manage, maintaining a 365-day presence requires a certain volume of posts to engage followers and deepen brand awareness. Strategically calculating volume and frequency, we produce around 150 short videos annually.

Iyoda: I've been asked before, "Why make so many?" (laughs). My personal answer is simply that having diverse content makes it more enjoyable for viewers. That entertainment value truly comes from having volume. When it comes to "creating volume," video generation AI plays a significant role, including its efficiency in production.

Mochihara: Beyond that, leveraging generative AI is also crucial for "expanding human creativity." Our goal with social media campaigns is to turn video viewers into fans of our brand or products. To achieve that, we need interesting content that people actually want to watch. We're counting on video generation AI to deliver unique outputs that only it can create.

Iyoda: However, as mentioned earlier, the fact that "anyone can create" inevitably leads to the commoditization and trivialization of video content. The challenge for creators is how to add an edge and express individuality within this context. We also focus on the human creativity Randy mentioned earlier. By streamlining processes with generative AI, humans can dedicate more time to the ideation and planning stages. will become increasingly important going forward.

Exploring the production process in the AI era while confronting both its negative and positive aspects

Iyoda: Finally, I'd like to discuss the future of video generation AI and how we engage with it. First, what are your thoughts on Veo's future outlook?

Randy: As we've discussed several times today, Google will continue providing better tools so more people can freely generate diverse creative work.

Beyond that, for Veo specifically, we're focused on improving accessibility and expanding the user base, driven by our desire for it to be used not only in business but also by general users.

We also plan to evolve Veo through iterations like Veo 4 and Veo 5, incorporating feedback and input on usability and features from companies like Dentsu Inc. and creators.

Mochihara: Regarding video generation AI overall, I feel we've entered an era where experimentation is possible, thanks to the perfect synergy between the current SNS environment that encourages "creating a lot" and the characteristic of video generation AI that allows "creating various things easily without worrying about restrictions." I believe video production using AI will increase across various fields going forward. I'm really looking forward to that future.

However, issues surrounding rights like copyright and portrait rights in AI utilization are extremely important and represent a challenge for the entire creative industry. For Veo 3, we offer rights assurance . As a company, we need to properly address these aspects while continuously thinking about how to strike the right balance to create interesting content.

Iyoda: I was deeply struck by a comment from an overseas creator: "I'm so glad I was born in this era." While video generation AI might seem like a negative force to some in the advertising industry, it actually brings many positive elements—like efficiency and enhanced human creativity.

As Dentsu Inc., we want to thoroughly consider the production process—how to understand these aspects and move forward proactively in partnership with client companies—and explore the appropriate approach.

AI-Driven SNS Solution "VERTICAL"

A consortium project involving multiple companies, including Dentsu Inc. and Dentsu Live Inc. It tackles fundamental challenges in the SNS domain, leveraging powerful players and a deep understanding of clients, users, and communication perspectives. This vertically integrated approach enables rapid, decisive advancement.
 

Related Article:
Users Who Won't Even Tap "Like" Are Increasing!? Understanding the "Three Principles" of Short Videos
Capture Global Fans with "AI×SNS"! The Cutting Edge of SNS Global Strategy