Google Gemini launches video generator: How to make AI clips using Veo 2 — Step-by-step guide

Google Gemini's Veo 1 AI video generator: Silicon Valley technology giant Google has rolled out its video generator feature on Gemini AI. This is an update the gives users the option to generate more than just images, but also videos, albiet short ones, at present.

Veo 2 is designed to produce high-resolution, detailed videos with cinematic realism using AI. Google claims that the model delivers “fluid character movement, lifelike scenes and finer visual details”, by “better understanding real-world physics and human motion”, across styles and subjects. We take you through a step-by-step guide on how to create AI videos on Google's Gemini using Veo 2.

Notably, in December 2024, rival OpenAI introduced Sora Turbo — its new and more advanced AI video generation model, designed to create realistic videos from text prompts. The tool was rolled out to OpenAI users with ChatGPT Plus and ChatGPT Pro.

Veo 2: Creating AI Videos with Gemini — What and Who?

Called Veo 2, at present it allows for the creation of eight-second long videos using text prompts. Veo 2 allows a user to generate an eight-second video clip at 720p resolution, delivered as an MP4 file in a 16:9 landscape format.

Google in its official blog post announcing the rollout said that you will have a monthly limit on how many videos can created, but did not state a number. But stated: “We will notify you as you approach” the monthly limit.

The roll out was announced on April 15 local time (April 16 IST), for Gemini Advanced users globally on web and mobile. For Google One AI Premium subscribers, the feature is also available in all languages Gemini supports.

Further, Google Labs is also making Veo 2 available through Whisk — which allows for the generation of AI videos using both text and image prompts, the company blog post added.

Here is a Step-by-Step Guide: How to Generate AI Video Using Veo 2?

• To generate AI videos on Gemini, select Veo 2 from the model dropdown 1.
• Once selected, describe the scene you want to create in the prompt — be it a short story format, specific scenery, or a visual concept i.e. fantasy, realism, unreal combinations and more.
• Google advices the prompt description be as detailed as possible to enable Veo 2 to “bring your ideas to life” and ensure control over the final AI video output.
• You can choose to share the video you create using the share button to social media platforms such as YouTube Shorts and TikTok.

How Does Veo 2 Work? Examples of Prompts Used:

The following are the various prompts that Google used to create AI videos using Veo 2. Check the results below:

• Prompt 1: “Time-lapse of a voxel-style ice cream melting under a clear blue sky. The pink, grey, and white scoops slowly lose their shape, dripping down the yellow, blocky cone held by a large, voxel hand with a blue sleeve. Focus on the melting process, highlighting the changing texture and form of the voxel elements as the ice cream deforms and drips, creating a colorful, pixelated melt against the vibrant blue backdrop.”

• Prompt 2: “An animated shot of a tiny mouse with oversized glasses, reading a book by the light of a glowing mushroom in a cozy forest den.”

• Prompt 3: “A wide, slow-panning shot of an enormous glacial cavern, bathed in eerie twilight. Pale cyan light filters from above, illuminating frozen candy figures within the ice walls. Two figures in white exosuits, their helmet lights casting beams, trudge through the center. Capture the cavern's scale and stillness.”

• Prompt 4: “Prompt: Aerial shot of a grassy cliff onto a sandy beach where waves crash against the shore, a prominent sea stack rises from the ocean near the beach, bathed in the warm, golden light of either sunrise or sunset, capturing the serene beauty of the Pacific coastline.”

Watch Video: How to Use Veo 2, Whisk?

Concern about Misuse, DeepFakes. What Google Says?

While many industry experts have raised concerns over misuse of AI generated videos for fake news, disinformation, and unconsented imagery i.e. Deepfakes, Google said it has taken “important steps” to ensure a “safe experience” through Veo 2 on Gemini.

What do these steps include? One is watermarking each frame of the video created so that it can be clearly identified AI generated content.

Another way is what Google calls “red teaming”. What is red teaming? It is a military term borrowed to describe a team of hackers who simulate various adversaries in order to prepare the organisation for against any future attacks on its AI systems.

Besides this, the company also claimed that it would continue to ensure that its AI generated videos do not violate its policies.