It's odd to consider but absolutely correct, that though the cinema concerns itself with images, most filmmakers do not think in images. Narrative and its logic of words and events has for the most part, since the inception of sound, informed most all cinematic production. Not all, by any means, but for the predominance of films made, yes.
What is to think in images?! Luis Bunuel had a daily ritual where late in the afternoon he would sit down to a Martini and think, think in images, let images come to him.
This thinking he considered a muscle and required exercise.
In any event Bunuel, like filmmaker, Harmony Korine, see, envision an image and the image leads, and is, its own narration. These images have nothing to so with a shot list or story boarding - no, the image emerges in the mind, in the imagination as just that, an image, and the image and images form a flow or beget another image. This image is a kind of encounter, a juxtaposition, a pre-rational or non-linguistic event.
Jean-Luc Godard has as immense facility for image and words and turns words into images. Hence all the puns and textual play, the word is an image. William Burroughs also works words as image, and as such, changes the tense of words to become a tense of images.
Where do images come from and how to think in images and then let the image happen in the mind and then to live in a form. Perhaps Marshal McLluhan has spoken most precisely about form and image, the multiple and the unconscious.
Nick Roeg makes films in images, he may not think in images, but he sees in images. Woody Allen films the flurry of words. He has a visual imagination but in the sense that Buster Keaton does. I would venture to say that Passolini thinks more visually being a poet than Fellini. David Fincher, Steven Soderberg, Gus Van Saint make pretty images, but these are images made by cinematographers.
The image is something much more intractable, much more puzzling, more ambivalent, more evocative, more slippery, more illogical, if we mean by logic something additive, it adds up, no the image does not add up, is not this or that, it is an image.