"The Honor 600 series will be the first phones with a unified multi-modal video generation model that runs on the device itself, allowing for video generation and editing based on natural language commands."
"You start by giving the AI one to three images to use as reference and use text commands to tell it what you want the video to be, resulting in a 3-8 second video."
"Honor is giving you a lot of control with this. For example, you can give the AI the start and end images and it will 'fill in the blanks' in between, guided by your instructions."
"AI models can get confused easily - since they only have a few images to work with, they can't see objects from all sides, so they can get tripped up as they try to draw those objects from unseen angles."
Honor's 600 series, launching on April 23, introduces AI Image to Video 2.0, allowing users to generate videos using images and natural language commands. The new feature utilizes a unified multi-modal video generation model that operates on-device. Users can input one to three images and provide text commands to dictate video content. The AI can also fill in gaps between start and end images. Pre-made templates are available for cinematic techniques, although AI models may struggle with limited image perspectives.
Read at GSMArena.com
Unable to calculate read time
Collection
[
|
...
]