Via FryAI
“MICROSOFT DEEPFAKE TOOL TOO GOOD TO RELEASE?
|
|
|
|
| “Patience is a key element of success.” (Bill Gates) | |
| What’s up? Microsoft Research has made a breakthrough in animation technology by developing an AI application that converts a still image of a person and an audio track into a lifelike animation with appropriate facial expressions. | |
| How does it work? Named VASA-1, the system is capable of transforming static images—whether photographs, drawings, or paintings—into “exquisitely synchronized” animations that mimic human speech and singing. VASA-1 is trained on thousands of images with a wide variety of facial expressions and can produce 512×512 pixel imagery at 45 frames per second. Videos are generated in about two minutes using a desktop-grade Nvidia RTX 4090 GPU. Potential applications of VASA-1 include creating lifelike avatars for games or simulations, but ultimately human creativity is the only limiting factor. | |
| Is it too good? Due to the potential for misuse, the research team is currently not making the system publicly available. The team stated, “We are dedicated to developing AI responsibly, with the goal of advancing human well-being. Given such context, we have no plans to release an online demo, API, product, additional implementation details, or any related offerings until we are certain that the technology will be used responsibly and in accordance with proper regulations.”” |
0 Responses
Stay in touch with the conversation, subscribe to the RSS feed for comments on this post.