Google Releases Enhanced AI Veo 2 And Imagen 3 Models

Google releases the latest AI video and image maker tool (photo: dock. Google)

JAKARTA Earlier this year, Google introduced a video creation model called Veo and released the Imagen 3 model to create images. Now, Google is updating the video and image model again.

On December 16, Google officially launched Veo 2 and Enhanced Imagen 3. By releasing Veo 2, this video creation model can now produce high-quality videos with various subjects and styles that users want.

According to Google's explanation, Veo 2 is more detailed and realistic overall. The model is also able to better understand cinematographic language and is able to create tracking shots in low angles, close-up shots, to wide angle shots.

"Veo 2 understands a unique cinematographic language. Ask for a genre, define a lens, suggest a cinematic effect, and Veo 2 will give it a resolution of up to 4K, and be extended to minutes," Google explained on its official website.

Google says Veo 2 is much better than other video-maker models because of the less hallucinations. This Google-made model rarely produces unwanted details, such as additional fingers or strange objects, making the results more realistic.

"Our commitment to the safety and development that is responsible for guiding Veo 2. We have deliberately measured Veo's availability in order to help identify, understand, and improve quality," Google said.

Meanwhile, Imagen 3 image-making models are now able to create brighter, sharper, and better images. The resulting style of art is also more diverse with higher accuracy, such as photorealism, impressionism, abstract, or even anime.

"This increase also follows orders more precisely, and presents more rich details and textures. In a side-by-side comparison between outputs by human appraisers and leading image-making models, Imagen 3 achieves sophisticated results," Google explained.

Imagen 3 is already available on Gemini and now Google will expand its image-making tools to ImageFX. Google started rolling out its features globally on December 16 and will be available in more than 100 countries supporting Google Labs.