Microsoft's research division has unveiled a new artificial intelligence model that can convert a still image into a GIF in seconds

Pix2Gif uses the same type of diffusion model found in many AI text-to-video models such as Runway and Pika Labs, but takes an "image translation" approach

In a published paper on the new model, the researchers state that it focuses on guiding movement through text prompts in parallel with image input It "spatially transforms" original image features based on text prompts

If you give it an image of a train and say "moving on the tracks," the train should appear to be moving This is a new approach to AI generation, combining images, video, and 3D models

Like many recent AI models, this is more of a research project than a publicly available Microsoft product that could be incorporated into Copilot However, it is possible to try it out and download the code

In other words, you can run it locally on your laptop For now, it is best to use a test environment where you can give an image or text prompt and get a gif

In my experiments, it took about a minute to generate a 2-second gif from a still image

All of the gifs in this story were created using this tool These are AI images generated using Adobe Firefly and animated through Pix2Gif

Motion guided warping may be more interesting than the ability to create GIFs from still images

It could be incorporated into Microsoft Designer or Paint to make animation easier and use AI to adjust images

To train the model, they used 100,000 animated GIFs with appropriate captions, extracted frames from the GIFs, and used the captions as text prompts

It is not clear where the GIFs came from, which may be why they are reserved as a research project A generally available and commercial model would require licensed data, especially if incorporated into a Microsoft product

For now, this is a fun diversionary tool and a way to test what a still image might look like with a few more frames Personally, I'm going to stick with AI video tools like Runaway, Pika Labs, and OpenAI Sora