Bytedance Omnihuman 1 Ai Tool Show How Realistic Ai Generated Deepfakes

Tiktok’s parent company bytedance has launched an Artificial Intelligence System that can create a complete video by making a complete video with the help of only one photo. This model also seems to challenge the existing Deepfec video tools. This new AI system of bytedance is named Omnihuman-1. Omnihuman-1 is an AI system that takes a reference to a one photo and creates a video of it, in which a person can be seen talking, gesturing, and even playing a music instrument. Is.

The AI ​​system of bytedance has come in the Omnihuman-1 market which can bring revolutionary changes in deepfhek video. Said about omnihuman (via) It is being said that this weak signal can create a very real -looking human video despite input. Especially on the basis of audio, it can make a realistic video which performs much better than the present methods. Its biggest feature is that it can support the image input of any aspect ratio. Whether it is a portrait photo, a half-body photo, or a full body image. It can give high-quality results by taking any input.

Omnihuman-1 can spontaneously create animation, adjust the body aspect ratio. And even surprisingly, can modify the existing video with great accuracy. This new AI system model of bytedance Omnihuman-1 has been trained on a video of 19 thousand hours. However, the model is also not completely flawless. It seems to struggle while working with low-quality images and some other inputs.

Researchers have also shared some sample videos on the project page of Omnihuman-1. It can be clearly seen in these videos how capable this tool is. The video given as an example also shows the movement of the hand and body. Also, the video can be generated from several angles. It can also make a video with an animated character. Apart from this, video can also be generated with history related celebrities.

It takes it alive with just one photo and feels like it is a real video of the subject with photo. Omnihuman-1 tool of course is seen in the race to make the best and most real-looking footage of AI-birthted humans.