Currently we traverse a directory and assume all the files are images. While is good practice to separate your assets (imho) we should not assume each file is an image or we'll end up creating image tags for assets that are not images and possibly crash Janus.
Instead, we should detect the file extension and create tags based on this.
Alternatively, we could add additional prompts to ask for various asset types (images, objects, scripts) and just output those with the assumption that the user is truthful.
Currently we traverse a directory and assume all the files are images. While is good practice to separate your assets (imho) we should not assume each file is an image or we'll end up creating image tags for assets that are not images and possibly crash Janus.
Instead, we should detect the file extension and create tags based on this.
Alternatively, we could add additional prompts to ask for various asset types (images, objects, scripts) and just output those with the assumption that the user is truthful.