It would be very useful to allow pasting an image in the research text box, and not having to save and upload it. Snipping, pasting a screenshot or copying from another page is a very common usage pattern.
In ChatGPT you can paste an image to the chat and ask questions about it. The way it works is that you paste the image and it is not immediately sent.
In research assistant this could be very powerful, for example:
Pasting a snippet from a text with many mathematical notations, asking to explain it in words.
Pasting a screenshot from a paused video, asking what kind of bag the person in the scene is wearing.
Copying an image with a sign in a foreign language and asking for translation.
Not having to go through saving the picture to the computer is important here, because there could be followups (e.g. another part of the same document, another angle of the same scene).
