A Co-Pilot extension using the OpenAI Vision support.
The extension requires an OpenAI API key to be set in the environment variable OPENAI_API_KEY
.
You can also do this in a .env
file your home directory.
In the chat view you can activate the participant using @vision
and enter a question about an image. If there is no image reference in the prompt (#image
), the participant will prompt for an image using a file dialog.
You can also refer to an image using an #file
reference in the prompt which allows you to select a file. For example:
@vision #file:sketch2.png what is on this image
From an open .png
file you can use the CoPilot icon in the editor toolbar to initiate a chat about the image shown in the editor.
When the response from the model includes a single html code block in the response, then the chat provides a 'Preview' button to preview the page in the simple browser.
See the folder samples for some example images and questions.