The long-awaited Google I/O2024 developer conference kicked off Tuesday with an expected slew of announcements covering AI, applications and future tools. 

The phrase AI was mentioned more than a hundred times on the day of the keynote alone. But alongside all the Gemini hype, there are some really interesting video and image generation news coming out of google labs that portends the future. 

Note that the emphasis is on the future. Sadly, the best of these "experiments" are currently only available on a wait and see basis. Here's how to access already active tools and get to the waiting list for those models that have not yet been launched.

At the top of the list is Google text to video's new VideoFX, which replaces OpenAI's Sora. The new tool is based on Google DeepMind's Veo model, which allows users to generate 1080p quality video clips "over 1 minute" in length. 

The video clips on the Veo demo page are impressive and promise some cool upcoming features such as clip extensions, video from still images, and mask editing. 

Unfortunately, this new wonder is not yet available, you need to sign up for the waiting list of projects to gain access Google text to image model Imagen2 was launched 3 months ago and the technology is already available on the ImageFX website

. 

However Adventurous Soul 3 models who want to test the brand new update Imagen, must join again...Yes, the waiting list of trusted testers. 

The results of the current picture are good, but certainly nothing to shout from the AI pioneers. Here we expect the new version offers significant improvements.

MusicFX was launched on May 12 last year and was a solid attempt at the AI text to music generator at the time. 

However, 5 months equals 2 AI lifetimes, and the tool is now hopelessly overtaken by newcomers like udio and Suno. And things will probably only get worse with the upcoming release of ElevenLabs Music.

Nevertheless, Google is bravely fighting to stay relevant, with an upgrade to MusicFX at this week's I/O showcase. The new DJ mode allows you to mix various genres along with text prompts, using sliders to adjust the intensity of each. 

There's no vocal action yet, but when compared to the rich complexity of services like Udio, the results are good, but still sub-par. The good news is that both DJ and creator modes are now available at the AI Test kitchen location.

Overall, it was a hard 12 months for Google that increasingly looks like the flat-legged giant caught a nap at the AI kitchen table. 

The company is gradually offering new applications, but many of them seem to be a desperate attempt to catch up with more agile and creative rivals. 

In a keynote on the power of Gemini1.5 to provide multimodal speed to NotebookLM, a small statement at the bottom right of the presentation screen declared "Audio is pre-generated." Not a great look for the creators of DeepMind.