I’m pleased to announce that version 4.8 of my Image Description Toolkit (IDT) is now available. If you have what Microsoft has been branding as a Copilot+ PC, IDT can now use the AI hardware built into these devices to describe your images.
What this means in the real world is quality image descriptions generated on your own device in about four to six seconds per image. Nothing is uploaded, and there’s no API key, no account, no internet connection and no cost, however many images you describe.
It is really great to have image descriptions from the operating system itself on both Mac and Windows. IDT has supported Apple Intelligence on the Mac since version 4.6, and Windows AI joins it in 4.8.
Using Windows AI
To use Windows AI specifically, you need a Copilot+ PC running Windows 11 24H2 or later. When you run the installer, leave the “Set up Windows AI” option checked. Then choose Windows AI as your provider in ImageDescriber, or use --provider Windows-AI with the idt command line.
So far I’ve only tested the Windows AI support on ARM hardware (Snapdragon). I don’t have one of the Intel or AMD Copilot+ PCs, so if you do, I’d love to hear whether it works for you.
You can get the latest release of IDT from the IDT 4.8.0 release page on GitHub.
What IDT Is
I wrote a blog post earlier this year giving a detailed explanation of what IDT is and what it is not. The short version: if you have dozens, hundreds or thousands of pictures you want described, IDT uses a range of AI technologies to streamline the description process. Point IDT at a collection of images, pick a description provider and a description type, and let the tool work. You’ll end up with a saved description for each image, including frames IDT extracts from your videos. You can optionally embed the descriptions in copies of the image files, and more.
Comments