ByteDance is ready to make a number of investments and studies in the development of artificial intelligence models. Most recently, ByteDance together with a number of researchers introduced its own video generation model, named “Goku”.
“Goku” was introduced in open source form, allowing various researchers and developers to also review its offerings. It is said that this model is based on Transformer whose flow is refined, thus improving the relationship and interaction between image and video tokens. This also enables the production of high-quality visuals.
Goku is seen to support several generation supports, including text to video, image to video, and text to image.
ByteDance through the relevant page Goku also stated that it can be used for advertising video generation – thus dramatically reducing production costs. The videos produced are also seen to be more realistic than various other offerings on the market.
For those of you who are interested, you can check out the research paper and code via the official Goku website today.