Microsoft’s research division has introduced a new AI model called Pix2Gif, which can transform still images into animated GIFs. This tool utilizes image translation technology to add motion to pictures, allowing users to create funny and creative GIFs in seconds.
Pix2Gif is a motion-guided diffusion model that can breathe life into images. It not only analyzes the image itself but also incorporates motion based on text prompts and a setting called “Motion Magnitude.” According to a Microsoft blog post, the AI model was trained on a carefully curated dataset extracted from the TGIF video-caption dataset. This dataset provided the AI with rich information about how subjects change over time. After this initial training phase, the model can be applied to various video datasets without any further fine-tuning.
While Pix2Gif is currently in the research phase and not yet available for public use, Tom’s Guide reported that the tool can be downloaded locally for experimental purposes. Users can upload a still image and provide text prompts describing their desired animation. Pix2Gif will then take a few seconds to generate a 2-second video or GIF based on the image processing and the given description.
Microsoft also revealed that Pix2Gif was trained on a dataset of 100,000 animated GIFs along with captions used as text prompts. There is a lot of excitement surrounding AI video generation tools, with OpenAI’s upcoming Sora being another highly anticipated option.
Japan High Court Deems Same-Sex Marriage Ban As Unconstitutional. Read More
