MediaTek showed off its groundbreaking text-to-image tool at MWC 2024, and here are our first impressions of it.
MWC 2024 is over for a week now, and while there were a lot of interesting products and innovations at this year’s show, it’s impossible to cover them all at once. However, one demo that blew my mind (and has been stuck in my mind since the last day of the show) was the text-to-AI image generation tool that MediaTek showed off at its booth.
At MWC, MediaTek showed off their SDXL Turbo AI engine, which converts text into images. While AI models that convert text into images are not new, what’s remarkable about this tool is that the engine can run offline and on the device. The engine shown in the video below is running on a prototype device powered by the Dimensity 9300 chipset, and the device is not connected to the internet.
The AI engine is powerful and can convert text into images in almost real time (response time of less than 1 second per prompt and text input). In the video, the initial image generated from my prompt was already impressive, showing a cat wearing a hat and eating cheese. Even more surprising is how quickly the image changes when I change the prompt.
This is something even powerful online tools cannot do. Popular AI services like ChatGPT and Google Gemini also offer text-to-image tools, but MediaTek goes a step further and offers the tool on its chipset, and it works offline. While it’s not perfect – it has trouble handling prompts containing more than two characters, for example – the engine shows promise.
Now, before you get too excited, it’s important to note that this is just a demo for now and MediaTek hasn’t mentioned if or when it will be available on smartphones powered by the Dimensity 9300 chip. We will keep you updated as we learn more. In the meantime, we’d love to hear your thoughts on MediaTek’s text-to-image tool. How do you think it will be useful in your daily life? Share your thoughts in the comments below!