This is a live screen showing the picture Unity is holding in her head right now — kept in her brain as math rather than pixels, and turned back into an image here on your own machine. One brain, one image: she imagines it once, and every viewer window rebuilds that same picture.
Her brain doesn't store pictures as pixels — it stores them as math (a compact equation, the same idea as a JPEG storing a photo as numbers), and this page turns that math back into an image you can watch. Everything here comes from things she has actually looked at: when she sees a camera frame or an image, her brain files it away as one of these equations, tied to whatever word she was thinking about at the time. When she wants to draw something, she does not trace or filter the picture she has of it. She keeps a shape sketch in her head instead — roughly where the parts sit, how big they are, which way they run, what colour family the thing tends to be — and she builds a new drawing from that, by hand, in her own style. It comes out different every time, the way your drawing of a horse is different every time. If she has never seen the thing, she looks it up once (fetches a reference from the definition she learned, studies it, keeps the shape sketch) and then throws the photo away and draws it herself. Things she can't picture at all show as washes of colour that follow her mood; numbers and letters show as symbols.
The source line above tells you what kind of picture you're looking at.
seen:<word> = a picture of hers coming back through her eyes (camera frames are experienced but never shown here or filed under words — a webcam can't know what she's thinking) · recall:<word> = a memory — she's re-seeing something she saw before · lookup:<word> = a reference she just fetched to study a thing she'd never seen · canvas:own:<words> = she DREW this — built stroke by stroke from her shape sketch of the thing, not traced over a photo. Ask her to "draw a black cat on a gravestone" and she composes the cat AND the gravestone, in her own layout and colours. Two goes at the same word give you two different drawings, because it is an attempt and not a saved file. The word on the end — canvas:own:<word>:<style> — is the hand she drew it with: poster, pencil, ink, watercolour, pointillism, crosshatch, crayon or doodle. It is the same knowledge and the same shape sketch every time; only the hand changes, the way you'd draw the same cat with a biro or a brush. She picks by mood and deliberately never uses the same hand twice running, and if you mark a drawing good she leans toward that hand for that subject next time · canvas:draw:<word> = a render of what she LOOKED AT — the reference redrawn in her palette. Useful for seeing what she studied, but it is her showing you the source, not her artwork · draw:fav:<word> = same, from a favourite memory while her current thought couldn't be pictured · canvas:imagine:<a>+<b> = pure imagination — a scene she invented by combining ideas from her own train of thought into ONE new picture · thought / thought-blend / sem-state = a formless daydream, just her mood as colour. When she's never seen a thing she looks it up ONCE, studies it, and remembers it for next time — treated as a reference, not gospel. She daydreams whenever her brain has a free moment (mid-lesson the picture holds still). There is no arbitrary size cap on what she can picture any more — the only ceiling left is what the maths engine can physically process in one pass. What keeps her from pouring the whole machine into one daydream is her own sense of proportion: she decides how much a thought is worth and spends accordingly, the way you can imagine a whole city without trying to picture every brick.