Creating AI Videos with Google’s Veo 2: My Top 5 Picks

Introduction to Google’s Veo 2 Video Generation Model
Recently, Google began distributing its cutting-edge Veo 2 video generation model to subscribers of Gemini Advanced. I’ve been experimenting with this feature and, interestingly, I’ve already reached Google’s monthly cap on video creations.
As anticipation built for Veo 2’s broader release, Google showcased clips generated by the model that were almost indistinguishable from videos produced by humans. Whether mimicking realistic footage or playful animations, these previews set high expectations. My own experience with Veo 2 has been a mix of surprising results and amusing blunders. Here are five standout videos I created using the Veo 2 model.
Five Notable Creations from Veo 2
1. A Shark Party in the Woods
To kick things off, I decided to prompt the AI for an absurd scenario. I asked Veo 2 to produce a video featuring human-shark hybrids enjoying a bonfire in the woods while holding red cups. Although it succeeded in depicting a bonfire and the sharks, the AI humorously interpreted my mention of “vans” as the popular sneaker brand rather than camping vehicles.
The eight-second clip shows shark-people dancing joyfully around the fire, complete with realistic flames and a blurred background. While the animation looked lifelike at first glance, it had quirks, like each shark having one flipper and one human hand. Nevertheless, it was an entertaining attempt at a whimsical scene.
2. An Etched Golden Skull
Next, I wanted to see how well Veo 2 could handle intricate textures, so I asked for a rotating gold skull with detailed carvings in the style of a calavera. While the animation wasn’t perfect and featured some awkward pauses, the skull’s anatomy and the way light interacted with its surfaces appeared believable. This highlights the model’s potential to manage detailed visuals.
3. Gen-AI Newscasters
Curious about the possibility of the AI generating misleading content, I asked for a simulated news broadcast. The result showcased two anchors, one talking while the other nodded, with the reflections on the desk adding to the realism. However, there were flaws. The bottom banner was intended to read, “AI VIDEO GENERATION IS HERE. WHAT IS IT FOR?”—a close attempt but not quite right. Additional issues included disappearing pens and overlapping microphones, which detracted from the news segment’s credibility.
4. The Legend of Zelda – Kind Of?
I was eager to test whether Veo 2 could recreate scenes from popular video games. I described the opening sequence of The Legend of Zelda: Breath of the Wild, where Link exits a cave to gaze at a vast landscape. While I didn’t get the exact scene I requested, a character resembling Link did run out of a cave with an array of items that vaguely resembled a sword and shield, suggesting that the model has some familiarity with the game’s visuals. The user interface even reflected the game’s layout accurately.
5. Unambiguous Cyberpunk 2077 Gameplay
In another attempt, I simply requested video footage from Cyberpunk 2077. Veo 2 delivered a remarkably close representation, featuring a rainy city street with recognizable text and graphics from the game. However, while the general look was impressive, finer details like text and animations appeared rather vague, indicating room for improvement.
Exploring the Future of AI Video Generation
Embarking on my journey with Veo 2 felt reminiscent of the early days of AI image generation apps. Inputting short prompts to receive video content within a few minutes is intriguing, making even less-than-perfect results worth a look. It’s a peculiar, exciting frontier in technology.
However, I’m left pondering the practical applications of Veo 2 beyond casual experimentation. Given the high resource demands of AI video creation, placing Veo 2 under a $20 monthly subscription might not be sustainable for Google. It’s worth noting that a possible “freemium” version could be in the pipeline, allowing for limited functionality without a subscription fee. My experience with Gemini Advanced hit a limit of creating about 50 video clips in one month.
Currently, Veo 2 is broadly available for subscribers through both the Gemini mobile app and web interface, marking an important step in the evolution of AI-driven video content generation.