Projection, seed and GPU model all act on the same free space

README.md

@@ -67,9 +67,11 @@ So: on one machine, with one card, everything here is reproducible to the bit
 
 The full prompt is in [`demo/chess-prompt.txt`](demo/chess-prompt.txt), the settings above in machine-readable form in [`demo/generation-settings.json`](demo/generation-settings.json), and the five renders are in [`demo/`](demo) one by one if you want to step through them.
 
-**What to look at.** Her knee. The prompt asks for it three times and ends on a sentence of its own — *"Her knee never stops bouncing."* It is a continuous involuntary motion with no narrative purpose, which makes it the clearest single sign that a projection carried what was written. Then look at the cat, the laundry and the furniture: named once, anchored nowhere, and they move from one render to the next. That part is expected, and it is the honest limit of the method — what the prompt states is carried, what it leaves open is refurnished by the model.
+**What to look at.** Everything the prompt states is there, on all five: the seated pose, the red dress, the white pieces on her side, the two captured black pawns, the cat, the straw hat, the laundry — and her knee, asked for three times and ending on a sentence of its own, *"Her knee never stops bouncing."* A continuous involuntary motion with no narrative purpose is the clearest single sign that a projection carried what was written, and it carries on the plain matrices too.
 
-One thing none of the five gets right, the 32B included: she lifts a knight and does not put it back on the same square. Object permanence behind an occluding hand, on a grid of sixty-four identical squares, is a limit of the video model rather than of the conditioning.
+**One thing none of the five gets right, the 32B included:** she lifts a knight and does not set it back on the same square, and on the 8B residual there is no knight on the board at all. Object permanence behind an occluding hand, on a grid of sixty-four identical squares, is a limit of the video model rather than of the conditioning.
+
+**And the terrace is furnished differently from one render to the next — that is not infidelity.** The prompt asks for a densely lived-in terrace without anchoring most of it: the cat is *"stretched out asleep in the sun"*, and nothing says where. What is left open, the model invents, and it invents differently depending on **the projection, the seed, and the model of GPU** — all three act on that same free space and none of them touches what was written. Two renders side by side give the impression of a different seed; that impression is what an unconstrained description looks like.
 
 > ⚠️ **Proof of concept — working, but a proof of concept.** It runs and produces good video, and every number below was measured on real hardware. Built and tested on a single setup (Windows 11, NVIDIA, ComfyUI 0.31.0) with deliberately limited exploration.