> ## Content Index
> Fetch the complete content index at: https://nordisk-ai.ghost.io/llms.txt
> Use this file to discover other available public pages before exploring further.

# Microsoft's AI Turns Still Images Into GIFs
- URL: https://nordisk-ai.ghost.io/microsoft-s-ai-turns-still-images-into-gifs/
- Published: 2024-03-27T07:00:00.000Z
- Updated: 2026-08-27T17:22:07.000Z
- Description: Microsoft's latest innovation in artificial intelligence: Pix2Gif, a groundbreaking model that turns still images into GIFs in just a few seconds.
- Author: Anders Hagen
- Tags: AI Models, AI, AI Sweden, Nordisk Ai, Sweden, Sweden, #Migrated-1787841634512, #wp, #wp-post, #Import 2026-08-27 14:41, #Import 2026-09-07 13:41

Microsoft's research division has recently launched a new artificial intelligence model, [Pix2Gif](https://hiteshk03.github.io/Pix2Gif/?ref=aiartweekly), which can turn still images into GIFs in just a few seconds. This model uses a diffusion approach similar to that found in many AI text-to-video models, such as Runway and Pika Labs, but with a unique approach to "image translation." It's built on Stable Diffusion as its base model.

The researchers behind the new, publicly available model have focused on controlling motion through a text prompt together with the image input. The model "spatially transforms" the features of the original image based on the text prompt.

### How Pix2Gif Works

By giving the model an image of a train and the text prompt "moving on the tracks," it creates a GIF in which the train appears to move. Although it's just a 2-second GIF with limited motion, this represents a new step in AI generation that combines images, video, and 3D models.

![](https://storage.ghost.io/c/ae/3a/ae3ae746-0116-4fb1-8b39-8b60b897bca8/content/images/2026/08/pix2gif-hur-fungerar-den.jpg)

### Example

### Availability and Use

Unlike many other new AI models, this is more of a research project than a publicly available product from Microsoft that could be integrated into Copilot. However, the code is available for download and can be run locally on a computer. For now, the best way to use it is via the test environment, where you can give the model an image and a text prompt to generate a GIF.

In my experiments, it took about a minute to generate a 2-second GIF from a still image, though this could be faster with a powerful GPU.

### Use Cases for Pix2Gif

The motion-controlled transformation aspect is arguably more interesting than the ability to create a GIF from a still image, since this could be applied to other video-like models in the future and potentially allow for editing of GIFs.

Such technology could be integrated into Microsoft Designer or Paint to facilitate animation, or even AI-assisted image adjustments.

To train the model, 100,000 animated GIFs with suitable captions were used, from which frames were extracted and the captions used as text prompts.

It's unclear where the GIFs came from, which may be why the project is still considered a research project. A publicly available, commercial model would require licensed data, especially if it were to be integrated into a Microsoft product.

For now, Pix2Gif is a fun tool for experimenting with what a still image might look like with a few more frames. Personally, I'll stick with AI video tools like Runway, Pika Labs, and the upcoming OpenAI Sora.