New app that helps users create unbelievable fake videos with AI

ByteDance, the company behind TikTok, has launched a groundbreaking artificial intelligence (AI) model called OmniHuman-1, which has been gaining significant attention for its ability to transform photos and sound bites into highly realistic videos. This new model can produce videos of people speaking, singing, and moving with an unprecedented level of accuracy and realism, far surpassing current audio-conditioned human video generation techniques, as stated by the ByteDance team in their technical paper.

While deepfake technology is often associated with fraud and entertainment, ByteDance’s model showcases the increasing advancements in AI made by Chinese developers, despite global concerns over their rapid progress. The OmniHuman-1 model is not yet available to the public, but sample clips have already gone viral, including one featuring a highly convincing 23-second video of Albert Einstein giving a speech.

ByteDance’s rise in the AI industry is emphasized by the Doubao app, which is currently the most popular consumer-facing AI app in China. Their model’s training strategy mixes diverse data sets, including text, audio, and movement, to generate videos with realistic facial expressions, head gestures, and body proportions. The AI has already shown potential for applications ranging from entertainment to educational content, as demonstrated in a TED-style speech demo with realistic hand movements.

Despite Washington’s efforts to curb Chinese AI developments, ByteDance remains a leader in the field, with its Jimeng AI platform being regularly updated with new capabilities, positioning the company at the forefront of AI-driven video generation.

Pharmacist arrested in Davanagere for sharing improper videos on social media

Bengaluru Woman Sexually Assaulted, Blackmailed with Private Videos

#ByteDanceAI #DeepfakeTech #VideoGeneration #OmniHuman