Just a week after the launch of the Sora from the OpenAI, Google has made official a few important innovation with respect to the pattern of its artificial intelligence. Among the highlights of the model Veo 2, it is able to generate a natural, more realistic.
Straight out of a post published on the official blog of Google, on December 16, 2024 learn it for the announcement of the new version of the model, a text-to-video, followed, among others, with a re-run of the new text in the image-Imagen 3. Google says that the Veo 2 “provides a better understanding of the physics of the real world, and the nuancave of the movement, and human expression, which is that it helps to improve the detail and realism of the overall”.
Veo 2 can produce videos up to the resolution is 4K, for a maximum of a few minutes. Google says that the model is also able to understand the demands of the cinema, trying to explain that it is likely to be included in the request, for example, the index in connection with the use of a “lens of 18 mm” to take effect with the wide-angle on which is the. known in the past. You can also play around with depth of field, looking troubled background. According to BigG, Veo 2 is also less prone to halucinacioneve than the other, the more the better.
Veo 2 is currently in the middle of a presentation about the vehicle VideoFX of Google Labs, though, you need to join on a list waiting to try out the service, and the model currently, it is not accessible from the Reporting. In any case, the results are then integrate the filigranin invisible SynthID to help the user to identify the contents of the set up with this tool. You can also find the videos and an impressive demonstration of the set up, with the Veo 2 by following the link to the source.
However, in terms of model-Imagen 3-text-in-image, it now generates more images of the light, and to be composed much better. Modelja also, you can track me to the correct styles of art. Imagen 3 are distributed in more than 100 countries in the ImageFX, the means for generating the image of the Google Labs. However, in this case, the Country is not among the countries involved.
Google took also the opportunity to declare the experiment and his most recent in the field of the production of images with IT. This is a device called a Whisk, which has just been launched in the United States of America. This service is a combination of the Imagen for 3, with the ability to understand and detailed visual of the Gemini. You can “connect” to the three images in order to perform a kind of remiksi, which is connected with the creation of the single. In short, we're trying to make the most of the easy delivery of the order to HIM, i do Gemini get automatically a new title for detailed images. More details in the video below.
Discussion about this post