Let’s explain to you what is Seedance 2.0the model of artificial intelligence designed by ByteDance. Specifically, it is a multimodal AI model capable of generating video from prompt descriptions, but also using other references such as audio, images and video.
Let’s start by reminding you what Seedance is, so you know what exactly we are referring to. Then, we will tell you the main features of its new version, as well as the differences with its predecessor and some other things that you should keep in mind, such as when will it be available for use in Spain.
What is Seedance and how it works
Seedance is an artificial intelligence model created by ByteDance, the company that created the TikTok social network. This model is used for generate video from textbeing the competition of models such as I see from Google either sora from OpenAI.
The truth is that ByteDance is one of the five most relevant Chinese companies in AI. In fact, it has its own family of models called Seedin which we have a Seedream to create images, Seededit to edit images, Seed3D to create three-dimensional models, Seed LiveInterpret to translate voice in real time, Seed-Music to create music and more.
As in other similar models, its operation is simple. You simply describe what you want to see through a prompt, uploading a reference image if you want, and Seedance will create a video of several seconds with the scene you have described.
To create the video, you will first be able to process and understand what you ask with natural languagesince it has been trained to understand the way we speak and express ourselves. It will then proceed to create the video, including synchronized audio based on the instructions you give it.
Seedance tries to stand out from the competition with physical realism, so that objects move naturally and consistent with physics. Also with visual consistency, and duration, since it generates videos up to 10 seconds long.
Seedance 2.0 Features
As we all know, the race in generative artificial intelligence is frenetic, so every few months a new model comes out, revolutionizing everything. In this aspect, Seedance 2.0 is now the one that is surprising the most, being a multimodal model capable of combining images, video, audio and text to generate content.
Think about this. You can add audio of a sound or conversation and generate the video from it, and even include another video or photo to use as a reference. Everything will be able to process simultaneously.
The model supports precise reference capabilities, and is capable of exporting results in 2K resolution with 30% faster generation speeds than the previous version 1.5. It has also improved the level of understanding of the physics of sound,
This new version of the model also introduces the so-called multi-lens storytellinga term that means you can keep the same character, with the same clothes and features through different camera cuts. Something that will allow the video to be more complex but maintain cohesion between the cuts.
When will Seedance 2.0 arrive?
ByteDance has not yet included Seedance 2.0 in the website of their modelsthat It is only available in testing phase for a very limited group of people. Even so, it is to be hoped that little by little it will also begin to reach everyone. You have to be careful, because there are also fake pages that use the name Seedance 2.0 as a claim.
In Xataka Basics | How to create videos with artificial intelligence: 13 essential free tools
GIPHY App Key not set. Please check settings