Of course, we can generate images from text via diffusion models.
My idea is now pretty simple. Assume, I have a set of icons (in the form of SVGs).
Is there any model I could use to enter some text and the model generates another image (SVG) solely based on those icons?